RES EARCH

◥ tic cells, through the nascent mesoderm, inter-

RESEARCH ARTICLE SUMMARY mediate mesoderm, and coelomic epithelium
states. FOSLCs exhibited a transcriptional pro-
DEVELOPMENTAL BIOLOGY file and cellular composition similar to those

Generation of ovarian follicles from mouse

in embryonic ovarian somatic cells at embryo-
nic day 12.5. When FOSLCs were aggregated

pluripotent stem cells

with PGCLCs derived from mESCs, the PGCLCs
entered meiosis, and subsequent oocyte growth
accompanied the development of FOSLC-
Takashi Yoshino, Takahiro Suzuki, Go Nagamatsu, Haruka Yabukami, Mika Ikegaya, Mami Kishima, derived follicles in culture. PGCLC-derived
Haruka Kita, Takuya Imamura, Kinichi Nakashima, Ryuichi Nishinakamura, Makoto Tachibana, oocytes developing in the FOSLC-derived
Miki Inoue, Yuichi Shima, Ken-ichirou Morohashi, Katsuhiko Hayashi* follicles were capable of fertilization and de-
veloped to live offspring. These results demons-

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
trate the reconstitution of functional follicle
INTRODUCTION: Germ cells develop in a spe- RATIONALE: During mouse development, the structures that are fully capable of supporting
cific environment in the reproductive organs. embryonic ovaries originate from the nascent oocyte production.
Throughout oogenesis, oocytes are encapsulated mesoderm, followed by the intermediate meso-
by somatic cells in follicle structures that provide derm and coelomic epithelium at the genital CONCLUSION: Our results demonstrate that
numerous signals and components essential ridge region. For the formation of embryonic functional gonadal somatic cells can be in-
for key events in oocyte development, such as ovarian somatic cells from mouse pluripo- duced from mESCs through a faithful differ-
meiosis and growth. The interaction between tent stem cells, appropriate signals need to entiation process in culture. The generated
the oocyte and the somatic follicular cells is be provided in culture to mimic those em- material may serve as a useful source to re-
regulated in a stage-dependent manner. Re- bryonic events. Using mouse embryonic stem place embryonic ovarian tissue for in vitro
cently, in vitro gametogenesis, reconstitution cells (mESCs) harboring reporter constructs gametogenesis. Furthermore, this system con-
of germ cell development in culture using plu- that monitor the expression of key genes for tributes to a better understanding of gonadal
ripotent stem cells, has been developed in each step, we set out to explore culture con- somatic cell differentiation and the inter-
mammalian species, including mice and hu- ditions for the recreation of the differentia- actions between oocytes and follicular somatic
mans. In mice, functional oocytes can be pro- tion process. Faithful gene expression and cells. Because it does not require embryonic
duced from pluripotent stem cell–derived functionality should be conferred in induced gonads, the methodology opens the possibility
primordial germ cell–like cells (PGCLCs) by embryonic ovarian somatic cells under the for application in other mammalian species
reaggregation with embryonic ovarian so- appropriate conditions. The functionality of with fewer ethical and technical concerns.
matic cells at embryonic day 12.5. Therefore, the induced cells should be verified by the This system will accelerate our understand-
in vitro gametogenesis is expected to be an ability to support the generation of functional ing of gonadal development and provide an
innovative means of producing a robust num- oocytes capable of fertilization and subsequent alternative source of gametes for research
ber of oocytes in culture. This should be par-
ticularly useful for application to humans and
development. and reproduction.
▪
endangered animals. However, the in vitro re- RESULTS: Based on reporter gene expression,
constitution of germ cell development is high- we determined a series of culture conditions
ly dependent on the somatic cell environment that recreate the differentiation process from The list of author affiliations is available in the full article online.
*Corresponding author. Email: hayashik@[Link]
provided by embryonic ovarian tissue, which pluripotent cells to gonadal somatic cells in
Cite this article as T. Yoshino et al., Science 373, eabe0237
is difficult to obtain from mammalian species. a stepwise manner. Under these conditions, (2021). DOI: 10.1126/science.abe0237
Here, we provide a model system that recon- mESCs differentiated into fetal ovarian soma-
stitutes the ovarian somatic cell environment tic cell–like cells (FOSLCs) expressing Nr5a1, a READ THE FULL ARTICLE AT
using mouse pluripotent stem cells. representative marker gene of gonadal soma- [Link]

FOSLCs

Mouse Reconstitution of Follicular development

pluripotent follicle structures with oocyte growth Offspring
stem cells

PGCLCs

Reconstitution of follicle structures, including oocytes, entirely from mouse pluripotent stem cells. Illustrations on the left show a schematic overview of
reconstitution of both FOSLCs and PGCLCs from mESCs. Oocytes in the reconstituted environment gave rise to offspring after fertilization. The right image represents
fully grown cumulus-oocyte complexes derived from FOSLCs (red) and PGCLCs (blue).

Yoshino et al., Science 373, 298 (2021) 16 July 2021 1 of 1

RES EARCH

◥ (LPM) are formed. As a step toward in vitro

RESEARCH ARTICLE reconstitution of the somatic gonad, we deter-
mined a culture condition that efficiently
DEVELOPMENTAL BIOLOGY induces the IMM from mouse embryonic stem
cells (ESCs) by focusing on T and platelet-derived
Generation of ovarian follicles from mouse growth factor receptor-a (Pdgfra) expression:
T is expressed in the nascent mesoderm at the
pluripotent stem cells primitive streak and then eventually restricted
in the notochord, whereas Pdgfra is expressed
Takashi Yoshino1, Takahiro Suzuki2,3, Go Nagamatsu1, Haruka Yabukami2, Mika Ikegaya2, in a lateral part of the nascent mesoderm that
Mami Kishima2, Haruka Kita1, Takuya Imamura1,4, Kinichi Nakashima1, Ryuichi Nishinakamura5, eventually differentiates into the PM, IMM,
Makoto Tachibana6, Miki Inoue7, Yuichi Shima7,8,9, Ken-ichirou Morohashi7,8, Katsuhiko Hayashi1* or LPM (3, 4) (Fig. 1A). For evaluation of the
culture conditions, female ESCs harboring
T nEGFP‐CreERT2/+ [T–green fluorescent protein

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
Oocytes mature in a specialized fluid-filled sac, the ovarian follicle, which provides signals needed for
meiosis and germ cell growth. Methods have been developed to generate functional oocytes from (T-GFP)] (5) (fig. S1B) were first differentiated
pluripotent stem cell–derived primordial germ cell–like cells (PGCLCs) when placed in culture with into epiblast-like cells (EpiLCs) (6) and then
embryonic ovarian somatic cells. In this study, we developed culture conditions to recreate the cultured in a U-bottomed plate with various
stepwise differentiation process from pluripotent cells to fetal ovarian somatic cell–like cells combinations of BMP4 and a WNT agonist,
(FOSLCs). When FOSLCs were aggregated with PGCLCs derived from mouse embryonic stem cells, the CHIR99021 (CHIR) (Fig. 1B). T-GFP expression
PGCLCs entered meiosis to generate functional oocytes capable of fertilization and development to was observed in cell aggregations cultured in
live offspring. Generating functional mouse oocytes in a reconstituted ovarian environment provides the presence of BMP4 or CHIR at 2 days of
a method for in vitro oocyte production and follicle generation for a better understanding of culture (D2) but disappeared at D4 (fig. S2A).
mammalian reproduction. Endogenous T protein was also detected in
T-GFP–positive cells (fig. S2B). Fluorescence-

I
activated cell sorting (FACS) analysis showed
n mammalian species, oocytes are grown Reconstitution in vitro of the entire pro- that in the presence of BMP4 or CHIR, most of
in the ovarian follicles for a long period of cess of follicular development would enable the cells expressed both T-GFP and PDGFRA
time to acquire competence of fertilization. a better understanding of oocyte development at D2, and then expressed only PDGFRA at D4
In mice, the interaction of oocytes with and robust production of oocytes in culture. (Fig. 1C and fig. S2C), indicating that the
surrounding somatic cells commences at Recently, we developed a culture system that nascent mesoderm–like cells were lateralized
embryonic day (E) 10, when the primordial produces functional oocytes from mouse plu- during the culture period.
germ cells (PGCs) migrate into the two genital ripotent stem cell–derived PGC-like cells Under these conditions, we monitored the
ridges. Somatic cells in the genital ridge pro- (PGCLCs) by reaggregation with female go- expression of Osr1 and Foxf1, which are repre-
vide signal(s) for the proliferation of PGCs nadal somatic cells isolated from E12.5 mouse sentative marker genes for IMM and LPM, re-
while proliferating themselves to form a pair embryos (2). This system is expected to pro- spectively (7, 8) (Fig. 1D and fig. S1, A and B),
of gonads. Upon sex determination at around vide a means of producing a robust number by using Foxf1-tdTomato/Osr1-GFP reporter
E12, female gonadal somatic cells start to dif- of oocytes in culture and should be particu- ESCs (fig. S3A). Foxf1-tdTomato was induced
ferentiate into granulosa cells and interstitial larly useful for application to humans and by BMP4 in a dose-dependent manner (Fig.
cells, which eventually form ovarian follicle endangered animals. To enable in vitro gen- 1E and fig. S3, B and C), consistent with
structures (1). After puberty, primary oocytes eration of mouse follicular development, it is evidence that BMP4 lateralizes the mesoderm
begin to grow to mature oocytes, and this pro- also necessary to develop a culture system that in vivo (9). Osr1-GFP was induced at a high
cess is tightly associated with the development allows the induction of functional female go- concentration of CHIR with BMP4, but the
of ovarian follicles that provide the support nadal somatic cells from mouse pluripotent effect was attenuated by an increased con-
required for oocyte growth and maturation. stem cells. By combining in vitro oocyte and centration of BMP4 (Fig. 1E and fig. S3, B and
somatic gonadal cells, it might then be pos- C), suggesting a mutually exclusive function
sible to generate a functional ovarian follicle of BMP and WNT signaling on the determi-
1
Department of Stem Cell Biology and Medicine, Graduate for fertilization and embryonic growth. nation of LPM and IMM. Supporting this
School of Medical Sciences, Kyushu University, Higashi-ku, observation, quantitative polymerase chain
Fukuoka 812-8582, Japan. 2Laboratory for Cellular Function ESCs differentiate into gonadal somatic cells reaction (Q-PCR) analysis of the marker gene
Conversion Technology, RIKEN Center for Integrative under defined conditions
Medical Sciences, Yokohama, Kanagawa 230-0045, Japan. expression showed that a high concentration
3
Functional Genomics, Graduate School of Medical Life During mouse development, the pluripotent of CHIR promoted the expression of the IMM
Science, Yokohama City University, Yokohama, Kanagawa, epiblast undergoes multiple steps to form the genes but prevented that of the LPM genes
230-0045, Japan. 4RNA Biology and Epigenomics Team/
LMCP, Program of Biomedical Science, Graduate School of
embryonic gonads (fig. S1A). During gastru- (fig. S3D). Under these conditions, the expres-
Integrated Sciences for Life, Hiroshima University, Higashi- lation, the pluripotent epiblast undergoes sion of the PM markers Uncx4 and Tbx2 re-
Hiroshima City, Hiroshima 739-8511, Japan. 5Department epithelial-to-mesenchyme transition along the mained at a very low level. Based on the
of Kidney Development, Institute of Molecular Embryology and
Genetics, Kumamoto University, Chuo-ku, Kumamoto
primitive streak, followed by bilateral ingress enrichment of the Osr1-positive/Foxf1–negative
860-0811, Japan. 6Laboratory of Epigenome Dynamics, underneath the epiblast layer (Fig. 1A). The cell population and transcripts of the IMM
Graduate School of Frontier Biosciences, Osaka University, Suita, distance from the primitive streak is impor- marker genes, we fixed the concentrations of
Osaka 565-0871, Japan. 7Department of Molecular Biology,
tant for cell fate determination during meso- BMP4 and CHIR at 1 ng/ml and 14 mM, respec-
Graduate School of Medical Sciences, Kyushu University,
Higashi-ku, Fukuoka City 812-8582, Japan. 8Department of derm development; that is, along with the tively, for the subsequent culture experiments.
Systems Life Sciences, Graduate School of Systems Life mediolateral axis, the notochord, the paraxial It is known that the mesoderm after gastru-
Sciences, Kyushu University, Higashi-ku, Fukuoka City 812-8582, mesoderm (PM), the intermediate mesoderm lation is anteriorized by retinoic acid (RA) and
Japan. 9Department of Anatomy, Kawasaki Medical School,
Kurashiki City, 701-0192 Okayama Prefecture, Japan. (IMM; which includes somatic precursors of in contrast posteriorized by fibroblast growth
*Corresponding author. Email: hayashik@[Link] the gonads), and the lateral pate mesoderm factor (FGF) and Wnt signaling (10–12). In

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 1 of 8

RES EARCH | R E S E A R C H A R T I C L E

Fig. 1. Sequential differen-

A in vivo C T(-) PDGFRA(-) T(+) PDGFRA(-) T(+) PDGFRA(+) T(-) PDGFRA(+)
tiation to the gonadal (Cell population I) (Cell population II) (Cell population III) (Cell population IV)
in vitro

Pdgfra(antibody)
somatic cell precursor. epiblast
A (%) (%) (%) (%)
IV III 100 100 100 100
(A) Marker genes for P
monitoring nascent meso- D2
50 50 50 50
derm differentiation. The left II
C20 C20 C20 C20
I 20 20 C14 20 C14 20 C14 C14
14
diagram shows embryonic Pdgfra 0 3
8
0 3
8
14
C1
C3
0
C8

3
8
14
0 C1
C3
C8
8
14
C1
C3
C8
C3
C8

3 C1

IR

IR

IR
B10 B10 B10

IR
B10
regions expressing T (blue) T T-GFP 10 3 1 B3 10 3 B1 1 C0 10 3
B3 1
B1 C010 3 1 B3
B1 C0 B3 C0

H
B1

H
1 0 1 B0 1 0 B0 1 B0 B0

C
0

C
0

C
BMP4 BMP4 BMP4 BMP4
and Pdgfra (yellow) in the (%) (%) (%) (%)
mesoderm. White arrows B (Days of culture) 100 100 100 100

D0 D1 D2 D3 D4
indicate the direction of the D4
50
spreading mesoderm. The BMP4 50 50
C20
50
C20 C20 C20
(0, 1, 3, 10ng/ml) 14
20 20 C14
C8 20 C8
C14
20 C14 C14

right diagram shows an CHIR 0 3

8
0 3
8
14
C1
C3
0 3
8
14
0 C1
C3
3
8
14
C1
C3
C8

C1
C3
C8

(0, 1, 3, 8, 14, 20µM)

IR
IR

IR

IR
B10

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
B10 B10
10 1 B3 10 1 10 1 10 1 B3 B3
expected sequence of nas- C0 C0 C0
3 3 3 3 C0

H
B1 B1

H
B1

H
1 1 0 B0 1 1 0 B0 B0

C
0 0

C
EGF BMP4 BMP4 BMP4 BMP4
cent mesodermal differentia- (50ng/ml)

tion in the FACS analysis. E Osr1(-) Foxf1(-) Osr1(+) Foxf1(-) Osr1(+) Foxf1(+) Osr1(-) Foxf1(+)
A, anterior; P, posterior. D in vivo in vitro
Foxf1-tdTomato
A (Cell population V) (Cell population VI) (Cell population VII) (Cell population VIII)
VIII VII
(B) Culture conditions tested P
(%) (%) (%) (%)
LPM 100 100 100 100
for nascent mesoderm dif-
ferentiation. D0 corresponds Foxf1 IMM 50
V VI 50 50 50
to EpiLCs. (C) Summary
Osr1
of FACS analysis of T-GFP Osr1-GFP 0 20 20 20

0
0
C0

C2

C2
14 14

C2
0 C1 14

4
0B10 0B10

4
10 0B10

C1

C1
B0
8 8

C1
8

C8

C8
1C3 C8

C8
10 B3 10 B3
and PDGFRA. Graphs show 8 3 IR 10 B3 3 3

B1
3

C3

C3
C3
3 B1 IR 3 B1 1 IR 3 B1 1 IR

B3
BMP 3 C14 14 1
CH CH

C1

C1
CH CH

C1
4 10
C2020 BMP 1 B0 BMP 1 B0 BMP 1 B0

0
0 0

C0

C0
0

C0
4 4

B1
4
the percentage of each cell
population differentiated
F H I Nr5a1- NR5A1 Merge
under various concentrations RA 0µM RA3, PD0, SHH0 hCD271 (Endogenous) + DAPI
of BMP4 and CHIR at D2 and 0.2 0.7
Gata4-CFP

D4. The Roman numerals

in the parentheses corre- D4
spond to the cell population
shown in (A). The mean
75.3 23.9
percentage in biologically
Osr1-GFP

triplicate experiments is
Foxl2-tdTomato

0.1 0.5
shown. (D) Marker genes for
monitoring intermediate D5
mesoderm differentiation. RA3, PD1, SHH30 Foxl2- FOXL2 Merge
The diagrams show embry-
G Osr1-GFP GATA4 Merge + DAPI
tdTomato (Endogenous) + DAPI

onic regions expressing Osr1 57.5 42.0

(green) and Foxf1 (orange) 0.2 10.2

(left) and an expected FACS
pattern representing IMM
D6
and LPM (right). (E) Sum-
mary of the FACS analysis of
Osr1-GFP and Foxf1-tdTomato. 43.3 46.3
Graphs show the per-
Nr5a1-hCD271 (BB515)
centage of each cell
population differentiated
under the conditions shown in (B). The mean percentage in biologically triplicate phenylindole (DAPI; blue) in a section of the Osr1-GFP ESC aggregate at 4 days
experiments is shown. (F) Distinct distribution of the Gata4-CFP–positive cell in culture with RA3, 1 mM PD (PD1), and 30 ng/ml SHH (SHH30). Scale bars, 20 mm.
population. The images show cell aggregations at 4 days in the culture with or (H) Differentiation of Nr5a1-hCD271–positive cells. Shown are the FACS profiles of
without 3 mM RA (RA3). Note that the cultures with RA show mutually exclusive Nr271F2T ESCs cultured for the number of days indicated. The numbers in the plot
distribution between Gata4-CFP–positive and Osr1-GFP–positive cells, the border represent the percentages of each cell population. (I) Expression of NR5A1 and
lines of which are indicated by white lines. Scale bars, 200 mm. (G) Expression of FOXL2 in the aggregates. Shown are immunofluorescence images of the reporter
Osr1-GFP and endogenous GATA4 protein. Shown are the results of immunostaining and endogenous protein indicated and merged images with DAPI (blue) in a section
of Osr1-GFP and GATA4 protein and merged images with 4′,6-diamidino-2- of the Nr271F2T ESC aggregate cultured at D6. Scale bars, 20 mm.

addition, Sonic hedgehog (SHH) is involved FGF inhibitor PD0325901 (PD), and SHH by gene for the genital ridge formation (14), in
in specification of the ventromedial coelomic adding each of these reagents at D2 (fig. S4A). female Osr1-GFP ESCs, thereby producing
epithelium in chick embryos (13). Therefore, To monitor differentiation into the precursors Osr1-GFP/Gata4-CFP ESCs (fig. S4B). The
for induction of the anterior ventral IMM, of the genital ridge, we inserted the enhanced addition of RA slightly up-regulated Gata4-
which should contain the precursors of the cyan fluorescent protein (ECFP) gene into the CFP and down-regulated Osr1-GFP, whereas
genital ridge, we tested the effect of RA, the locus of Gata4, the earliest functional marker PD or SHH had no obvious impact on their

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 2 of 8

RES EARCH | R E S E A R C H A R T I C L E

expression (fig. S4C). Q-PCR reinforced the magnetic-activated cell sorting (MACS) (fig. the Nr5a1-hCD271–highly positive cell popula-
slight up-regulation of Gata4, Lhx9, and Wt1, S6A). Comparison of the expression profiles of tion (Fig. 1H). Conversely, Nr5a1 was not de-
functional marker genes for gonadal somatic Nr5a1-hCD271–positive cells at D6 with those tectable in some cells. The heterogeneous level
cell precursors (14–16), in response to RA (fig. of cells in the gonads from E10.5 to E14.5 em- of endogenous NR5A1 protein expression (Fig.
S4D). Mutually exclusive distributions of Gata4- bryos revealed a similar pattern of cell clusters 1I) indicates that the expression of Nr5a1 was
CFP–positive cells and Osr1-GFP–positive cells between E11.5 and E14.5 (Fig. 2A). The num- highly heterogeneous at the transcript and
were observed in the presence of RA (Fig. 1F bers of clusters 0, 1, 2, 4, and 5 were com- protein levels. Because of the substantial con-
and fig. S4E). This pattern was confirmed with parable both in vitro and in vivo, whereas tribution of the cell cycle state to the gene ex-
endogenous GATA4 protein in the Osr1-GFP other clusters were fewer in vitro. Based on pression profile (24, 25), we estimated the cell
ESC aggregates (Fig. 1G). This exclusive pat- the expression of marker genes defined by a cycle stage in each cell population. This analy-
tern is consistent with that in gonadal somatic previous transcriptome study (1), it appears sis suggested that cluster S5 was actively pro-
cell precursors in vivo (fig. S4F). Despite the that clusters 0, 1, 2, 4, and 5 include granulosa liferative and portions of clusters S0 and S3
subtle effects or lack of effect of PD and SHH cell, stromal cell, or early progenitor cell pop- were also proliferative (Fig. 2D). By contrast,
on Gata4, Lhx9, or Wt1 expression, the addi-

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
ulations, and clusters 3, 6, 7, and 8, which were most cells in clusters S2 and S4 were in G1,
tion of these factors resulted in an increased few in vitro, correspond to germ cell, endothe- consistent with previous findings that cells ex-
number of Gata4-CFP–positive/Osr1-GFP–low lial cell, erythrocyte, and megakaryocyte pop- pressing Foxl2 arrested their cell cycle through
cells (fig. S4G). Based on these marker gene ulations, respectively (Fig. 2B and fig. S6B). p27 and CDKN1B (22, 26). Genes involved in
expressions and on the number of Gata4-CFP– The differentiation of germ cell–like cells was epithelial cell function and ovarian epithelial
positive/Osr1-GFP–low cells produced, we fixed confirmed by evidence that cells expressing cancer, such as Krt19, Upk3b, and Itm2a, were
the concentrations of RA, PD, and SHH at Blimp1-mVenus (BV), stella-CFP (SC), and expressed in clusters S1 and S5 (fig. S6H), sug-
3 mM, 1 mM and 30 ng/ml, respectively, for the POU5F1 were sparsely induced under these gesting that these clusters could include the
subsequent culture experiments. conditions (fig. S6C). These germ cell–like surface epithelium of the fetal ovary, known to
Nr5a1 is expressed in all cell lineages in the cells, as well as endothelial cells, erythrocytes, be the source of granulosa cells (26). Based on
genital ridge (1, 17), and its expression is co- and megakaryocytes, could be induced by BMP4 these observations, we designated clusters S2
ordinated by various transcription factors that and WNT signals that promote the differen- and S4 as granulosa cells; clusters S0 and S3 as
are essential for gonadal development, such tiation of PGCs and Flk1-positive common stromal cell progenitors and stromal cells, re-
as GATA4 (14), EMX2 (18), WT1 (15, 16), and progenitors of hematopoietic and endothelial spectively; and clusters S1 and S5 as early pro-
LHX9 (15), therefore indicating that Nr5a1 is cells from ESCs (6, 23). Because Nr5a1 expres- genitors (Fig. 2C). The percentage of granulosa
the most stringent marker for differentiation sion was undetectable in clusters 3, 6, 7, and 8 cells was smaller in the cell population differ-
into gonadal somatic cells. To monitor Nr5a1 (fig. S6D), these minor populations might entiated in vitro than that in vivo (Fig. 2E). This
expression, we derived female ESCs from have been the result of insufficient removal may have been caused by a delay in granulosa
Nr5a1-hCD271 bacterial artificial chromosome by MACS. cell differentiation in culture (see below). Com-
transgenic mice (19) (fig. S1B), in which human To analyze the gonadal somatic cells that parison of the gene expression in each cluster
CD271 gene is driven by the Nr5a1 promoter. directly contribute to the follicle structure, we between the in vivo and in vitro differentiations
Using Nr5a1-hCD271 ESCs, we inserted the compared single-cell profiles between Nr5a1- showed that they were highly similar (R > 0.96)
tdTomato gene into the Foxl2 locus, a marker hCD271–positive cells at D6 and E12.5 gonadal (Fig. 2F and fig. S6I). Based on the similar pat-
gene for granulosa cells (20), thereby produ- somatic cells (fig. S6E) because FOXL2 expres- tern of cell clusters and of gene expression within
cing Nr5a1-hCD271/Foxl2-tdTomato (Nr271F2T) sion was first detectable at those stages in vitro each cluster, we concluded that the Nr5a1-hCD271–
ESCs (fig. S5A). When Nr271F2T ESCs were and in vivo, respectively (Fig. 1H) (22). After positive cell population was similar to the E12.5
cultured under the conditions described above, excluding the germ cell, endothelial cell, and gonadal somatic cell population. We thereafter
Nr5a1-hCD271 was detectable in a group of cells hematopoietic cell populations (fig. S6F), the named the Nr5a1-hCD271–positive cells fetal
at D4 (Fig. 1H). As the culture progressed in remaining populations could be reclassified ovarian somatic cell–like cells (FOSLCs).
the medium containing BMP4 (20 ng/ml) and into six clusters, S0 to S5 (Fig. 2C). Cells ex-
a low dose of FGF9 (2 ng/ml), the percentage pressing the granulosa cell marker genes were FOSLCs support oocyte development
of Nr5a1-hCD271–positive cells increased. From enriched in clusters S2 and S4, and cells ex- To evaluate function, FOSLCs were reaggre-
D6 onward, Foxl2-tdTomato–positive cells ap- pressing the stromal cell marker genes were gated with PGCLCs harboring BV and SC
peared (Fig. 1H and fig. S5B). Immunofluore- enriched mainly in cluster S3 and partially in reporter genes. Considering that PGCLCs cor-
scence analysis confirmed endogenous NR5A1 cluster S0 (fig. S6G). The expressions of some respond to E9.5 PGCs (6), we sorted FOSLCs at
and FOXL2 expression in the cells expressing stromal cell marker genes, such as Wnt5a and D5 by MACS, which yielded 7640 ± 1670 (±SE,
the reporter genes (Fig. 1I). In mouse develop- Tcf21, were detectable in cells belonging to n = 10 replicates) FOSLCs on average from one
ment, the expression of Foxl2 has been shown clusters S1 and S5. In clusters S1 and S5, the aggregation. Because the exact ratio of PGCs
to be detectable in the female gonad from expressions of early progenitor marker genes to gonadal somatic cells in the nascent genital
E12.5 (21, 22). Given that EpiLCs correspond (1) such as Sox11, Ecm1, and Nr2f1 were de- ridge is difficult to define, following the ratio (5
to E5.75 epiblasts (6), the fact that a total of tectable, indicating that these clusters contain to 18%) in the E12.5 gonads (27), 5000 PGCLCs
6 days after EpiLC differentiation was required early progenitors. The close similarity in gene harboring the BV and SC reporter genes were
for the differentiation of Foxl2-tdTomato– expression between the cluster S0/S3 express- reaggregated with 75,000 or 100,000 FOSLCs
positive cells was largely consistent with the ing stromal markers and the cluster S1/S5 and then cultured under in vitro differentia-
time course of development in vivo. expressing early progenitor markers was con- tion culture (IVDi) conditions (2). Many oocytes
sistent with the fact that early progenitors and were formed in the reaggregates (Fig. 3A),
ESC derivatives share similar properties with stromal cell progenitors share a similar gene which were thereafter called reconstituted
gonadal somatic cells in vivo expression profile (1). Nr5a1 was widely ex- ovarioids (rOvarioids) to distinguish them
To analyze the cell populations induced, pressed and enriched in clusters S2 and S4 from the ovarioids containing E12.5 gonadal
we applied single-cell RNA-sequencing analy- (fig. S6G), consistent with the evidence that somatic cells. This oocyte formation relied
sis of Nr5a1-hCD271–positive cells sorted by Foxl2-tdTomato–positive cells appeared from on FOSLCs or gonadal somatic cells because

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 3 of 8

RES EARCH | R E S E A R C H A R T I C L E

A in vivo
in vitro (D6) E10.5 E11.5 E12.5 E13.5 E14.5

0
1
UMAP_2

2
3
4
5
6
7
8

UMAP_1
Granulosa cell markers Early progenitor markers
B Foxl2 Inha Kitl Sox11 Ecm1 Nr2f1

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
Stromal cell markers Germ cell markers
Wnt5a Pdgfra Tcf21 Pou5f1 Ddx4 Dppa5a
E

C E F
E12.5 (in vivo) D6 (in vitro) 100
Granulosa EP EP EP ST ST GR GR Cluster
Stroma
EP S5 (in vivo)
EP
EP
UMAP_2

S5 (in vitro)
Cluster 75 S1 (in vivo)
EP

S0 EP
ST S1 (in vitro)
Early Stromal S1
pro
progenitor progenitor S2 S3 (in vivo)
ST

S3 ST
S4 pro S3 (in vitro)
ST
S5 50 S0 (in vivo)
ST

UMAP_1 S0 (in vitro)

D S2 (in vivo)
GR

ST
Cluster S2 (in vitro)
GR
25 S0 S4 (in vivo)
GR

GR S1
UMAP_2

S2 S4 (in vitro)
Cluster
S5 (in vivo)

S1 (in vivo)

S3 (in vivo)

S0 (in vivo)

S2 (in vivo)

S4 (in vivo)
S5 (in vitro)

S1 (in vitro)

S3 (in vitro)

S0 (in vitro)

S2 (in vitro)

S3 S4 (in vitro)
GR S4
(%) 0 GR S5
G1
vi .5

(in D6

)
tro
(in 12

S
vo

vi
E

G2/M

UMAP_1

Fig. 2. Comparison of gene expression profiles between gonadal somatic Shown are the results of UMAP analysis of somatic cells contributing to the follicle
cells in vivo and in vitro. (A) Two-dimensional uniform manifold approximation structure among E12.5 female gonadal somatic cells and MACS-sorted Nr5a1-
and projection (UMAP) plot of single cells. Shown are the results of UMAP hCD271–positive cells. Cells are clustered by a graph-based clustering. Dotted lines
analysis of MACS-sorted Nr5a1-hCD271–positive cells at D6 and female gonadal designate the cell type in each cluster. (D) Cell cycle phase analysis of single
somatic cells at the embryonic day indicated. The E10.5 sample includes the cells. The cell cycle phase was calculated using canonical cell cycle markers (35).
dorsal mesenchymal tissues around the gonad. Cells are clustered by a graph- (E) Proportion of each cell type in E12.5 female gonads and Nr5a1-hCD271–positive
based clustering. (B) Expression of marker genes for granulosa cells, stromal cells. The percentage of each cell type was calculated by the UMAP plot in (C).
cells, early progenitors, and germ cells. Cell positions are compiled from the (F) Correlation matrix heat map for all clusters in vivo and in vitro. The clusters were
UMAP plots in (A). (C) Comparison of follicular cell precursors in vitro and in vivo. ordered based on the hierarchical clustering.

PGCLCs alone in culture were degraded by pression after D14. This sequence of reporter ber of oocytes in rOvarioids with 75,000 FOSLCs
D14 (fig. S7A). PGCLCs in the reaggregates gene expression in rOvarioids was indistin- was reduced to 56.9%, on average, of the num-
expressed SC and BV at D2, down-regulated guishable from that in the ovarioids containing ber in ovarioids with 75,000 gonadal somatic
these genes at D7, and regained only SC ex- E12.5 gonadal somatic cells (Fig. 3A). The num- cells derived from E12.5 ICR embryos, whereas

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 4 of 8

RES EARCH | R E S E A R C H A R T I C L E

A
D2 D7 D14 D23
E12.5 FOSLCs FOSLCs E12.5 FOSLCs FOSLCs E12.5 FOSLCs FOSLCs E12.5 FOSLCs FOSLCs
BF
F2T
SC
BV (75,000) (75,000) (100,000) (75,000) (75,000) (100,000) (75,000) (75,000) (100,000) (75,000) (75,000) (100,000)

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
B NS D F2T BVSC DDX4 Merge + DAPI
400 100
** (P<0.01)
350
80
No of SCposi oocytes

% of SCposi oocytes

D3
300

250 60
200
F2T BVSC SYCP3 Merge + DAPI
40
150

100
D7
20
50

0 0

E12.5 ICR FOSLCs FOSLCs FOSLCs F2T Nr5a1-hCD271 BVSC Merge + DAPI
(75,000) (75,000) (100,000) (120,000)

C
D11

BF SC Merge
E12.5

D14
FOSLCs

D23

Fig. 3. Oogenesis in the culture system using FOSLCs. (A) IVDi culture using from at least three biologically independent experiments. The bars in the graph
FOSLCs. Images show PGCLCs reaggregated with E12.5 gonadal somatic cells or indicate the mean value. P values relative to the control were determined using
FOSLCs cultured for the number of days indicated. The numbers in parentheses are Student’s t test. (C) Complete elimination of endogenous oocytes in an rOvarioid.
the numbers of E12.5 gonadal somatic cells or FOSLCs cells used. BF, bright field. BF and fluorescence images show an ovarioid using 75,000 E12.5 gonadal somatic
Scale bars, 200 mm. (B) Summary of oocytes yielded in IVDi culture. Shown are cells and an rOvarioid using 100,000 FOSLCs. SC-negative oocytes were observed
the number of oocytes (left) and the percentage of SC-positive oocytes (right) in the in the ovarioid (arrowheads). Scale bars, 200 mm. (D) Immunofluorescence analysis
ovarioid or the rOvarioid at day 23 of culture. The numbers of oocytes were of oocyte development in the rOvarioid. Shown are immunofluorescence images
determined by counting SC-positive cells in the image. The values were obtained of the antigens indicated and merged images with DAPI (blue). Scale bars, 20 mm.

it became comparable in rOvarioids with E12.5 C57BL/6J gonadal somatic cells (fig. S7, which is important to ensure the origin of
100,000 FOSLCs (Fig. 3B). The productivity B and C). All oocytes in the rOvarioids were oocytes. In addition, the uniform SC expres-
was not improved in rOvarioids with 120,000 SC positive, in contrast to ovarioids, which sion also rules out the possibility that cells
FOSLCs. This inferior potential of FOSLCs to contained SC-negative oocytes derived from expressing germ cell markers found in the
become embryonic gonadal somatic cells was residual PGCs mingled in the E12.5 gonadal FOSLCs contribute to oocytes in the rOvarioid.
largely reproducible under the same genetic somatic cells despite depletion by antibodies We next verified the differentiation pro-
background; the number of oocytes in rOvarioids specific for PGCs (Fig. 3, B and C). This result cess in rOvarioids using immunofluorescence
with FOSLCs derived from C57BL/6J ESCs was demonstrated that the rOvarioid completely analyses. At D3, Foxl2-tdTomato–positive cells
reduced to 67.4% of that in ovarioids with eliminated contamination of residual oocytes, started to surround PGCLCs that appeared to

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 5 of 8

RES EARCH | R E S E A R C H A R T I C L E

proliferate to form germ cell cysts with ex- the plasticity to differentiate into granulosa were heterozygous for the reporter genes. Two
pression of DDX4, a later germ cell marker cells and the capability to form follicle struc- independent pairs of these mice produced 10
(Fig. 3D and fig. S8, A and B). At D7, PGCLCs tures. This plasticity is consistent with the ob- and 14 pups by their intercrosses (Fig. 4H),
entered meiosis, thereby differentiating oo- servation that the granulosa cell population and 17 out of 21 pups tested had the BV and/or
cytes, with typical alignments of the SYCP3 continuously increased after D6 (fig. S9A). Given SC reporter genes (Fig. 4I), demonstrating their
protein. Individual follicle structures with that the percentage of granulosa cells was smaller fertility in both males and females. These
Foxl2-tdTomato–positive cells were formed in the cell population differentiated in vitro at results demonstrated that mouse oocytes pro-
by D11. At this stage, oocytes degrading in D6 than in that in vivo (Fig. 2E), the differenti- duced in the ovarian environment entirely re-
the rOvarioids were frequently observed (fig. ation of granulosa cells may be delayed in the constituted by pluripotent stem cells acquired
S8C), consistent with our previous report that culture system because of an unknown condi- the competence for fertilization followed by
oocyte loss accompanied with apoptosis was tion that was not fully recapitulated in culture. development to term.
observed in ovarioids at D11 (28). As the culture
progressed, Foxl2-tdTomato–positive granulosa- Oocytes acquire developmental competence in Applicability of FOSLCs
like cells became stratified and Nr5a1-hCD271 the culture system using FOSLCs

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
In our system, purification of FOSLCs is great-
expression was more prominent in the cells Developmental competence of FOSLCs was ly dependent on the Nr5a1-reporter construct.
surrounding the follicle structure than in further validated by in vitro growth culture This requirement may compromise the ap-
Foxl2-tdTomato–positive granulosa-like cells (IVG), in which secondary follicles grow up to plicability of this system because of the time-
(Fig. 3D and fig. S8B). This is consistent with a stage equivalent to pre-ovulatory follicles (2). consuming and laborious processes needed
evidence in vivo that NR5A1 becomes prom- In the IVG culture, FOSLC-derived granulosa for the production of the reporter cell line.
inent in theca and stromal cells but is down- cells proliferated and formed cumulus-oocyte Therefore, we tried to provide an rOvarioid
regulated in granulosa cells during follicle complexes (COCs) by D12 with the formation system without the need for a reporter gene.
development (29, 30). At D23, the formation of transzonal projections (TZPs), which are The main reason for requiring a reporter sys-
of secondary follicle structures composed of essential for juxtacrine interaction to support tem was the appearance of massively prolifer-
SC-positive oocytes with a multilayer of Foxl2- oocyte growth (31) (Fig. 4, A and B). Under in ative cells, which severely disturbed oogenesis,
tdTomato–positive granulosa-like cells and the vitro maturation culture (IVM) conditions (2), in reaggregations without purification of
far surrounding Nr5a1-hCD271–positive theca- FOSLC-derived cumulus cells were expanded, FOSLCs (fig. S11A). Because such proliferative
like cells was observed. as is typically observed in maturation of cu- cells were observed in the aggregates contain-
To investigate changes in the competence of mulus cells (Fig. 4C). These cumulus cells were ing undifferentiated cells (32, 33), they were
FOSLCs during differentiation, PGCLCs were readily dispersed by treatment with hyaluroni- likely to lie in the Nr5a1-hCD271–negative cell
reaggregated with FOSLCs at day 4, 5, 6, 7, or 8 dase, and 28.4% (33/116) of the isolated oocytes population. Indeed, the proliferative cells al-
of culture. FACS analysis during the differen- proceeded to the MII stage with extrusion most exclusively appeared in aggregates with
tiation period showed an increase in the per- of the first polar body (Fig. 4D and table S1). Nr5a1-hCD271–negative cells (fig. S11B). There-
centage of Foxl2-tdTomato–positive cells after This developmental rate to the MII stage in fore, we tried to remove the source of the
D6 (fig. S9A). The analysis also showed that rOvarioids was comparable to that derived proliferative cells by using antibodies against
the Nr5a1-hCD271–positive/PDGFRA–negative from reaggregates using E12.5 gonadal somatic endogenous SSEA1 and CD31 because pluri-
cells, which was the major population at D5, cells in our previous report (2) (28.9%, 923/ potent stem cells express SSEA1 and CD31
differentiated into either Nr5a1-hCD271–highly 3198; P = 0.994 by Pearson’s chi-square test). (34), and these markers were indeed expressed
positive/PDGFRA-negative cells or Nr5a1-hCD271– We then used mature COCs from rOvarioids in a small subset of Nr5a1-hCD271–negative
positive/PDGFRA-positive cells, which, based for in vitro fertilization (IVF) using wild-type cells (fig. S11C). rOvarioids with the cell frac-
on the marker gene expression (fig. S6G), are sperm from ICR mice. In IVF followed by tion in the flow-through after the depletion
granulosa and stromal cells, respectively. When in vitro culture, oocytes were fertilized, and with SSEA1 and CD31 antibodies yielded a
aggregated with PGCLCs, FOSLCs at D5 and 30.2% (301/996) of oocytes used in the IVF number of follicle structures without a prolif-
D6 showed a high potential for supporting became two-cell embryos (Fig. 4D and table erative cell clump (fig. S11, D and E). These
oogenesis (fig. S9B), suggesting that FOSLCs S2). Then, 25.8% (24/93) of the two-cell em- results demonstrate that the depletion meth-
interact with PGCLCs in a timely fashion. Based bryos developed to blastocysts (Fig. 4D and od eliminates the origin of proliferative cells,
on marker gene expression, FOSLCs at D6 can table S3). This developmental rate from two- and therefore a reporter construct is dispens-
be divided into three subpopulations: Foxl2- cell embryos to blastocysts was comparable to able for the rOvarioid system, which would
tdTomato–positive (F2T+) cells, Foxl2-tdTomato– that observed in embryos derived from reag- help to expand the applicability of this sys-
negative and PDGFRA-positive (F2T-P+) cells, gregates using E12.5 gonadal somatic cells in tem to production of oocytes from pluripotent
and Foxl2-tdTomato–negative and PDGFRA- our previous report (2) (31.8%, 44/138; P = stem cells without embryonic tissues.
negative (F2T-P–) cells (fig. S10A). Genes for 0.397 by Pearson’s chi-square test). When the
granulosa cells were enriched, as expected, in two-cell embryos were transferred into pseu- Outlook
the F2T+ cell population (fig. S10B). Although dopregnant females, 5.2% (11/212) of the em- Here, we have established a culture system
the F2T-P+ and F2T-P– cell populations could bryos gave rise to offspring and all of them that reconstitutes functional ovarian follicles,
not be clearly distinguished, genes for stromal– developed to adult mice (Fig. 4, E and F, and including oocytes, from pluripotent stem cells.
stromal progenitor cells were slightly enriched table S4). This developmental rate to offspring This system provides several insights for en-
in the F2T-P+ cell population. When aggregated was comparable to that derived from reaggre- hancing our understanding and reconstitu-
with PGCLCs, F2T-P+ and F2T-P– cells restored gates using E12.5 gonadal somatic cells in our tion of oogenesis. First, this culture system
Foxl2-tdTomato expression by D7 and formed a previous report (2) (3.5%, 11/316; P = 0.459 by would be an efficient tool for understanding
number of follicle structures, whereas the F2T+ Pearson’s chi-square test). All offspring had the molecular mechanisms underlying the dif-
cell population formed a significantly smaller dark eyes and some of them had the BV or SC ferentiation of gonadal somatic cells. Second,
number of follicle structures (fig. S10, C and reporter gene (Fig. 4, F and G), consistent with it enables us to address the interaction be-
D). These results indicate that the F2T-P+ and the fact that the ESCs used were derived from tween PGCs and/or oocytes and gonadal so-
F2T-P– populations contain cells that still have the F1 blastocyst (129X1/Svj × C57Bl/6J) and matic cells. Because this system can separately

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 6 of 8

RES EARCH | R E S E A R C H A R T I C L E

Fig. 4. Full-term development of embryos A BF F2T SC Merge B Phalloidin FOXL2

derived from rOvarioids. (A) IVG culture
using FOSLCs. Shown are BF and fluores-

D3
cence images of reconstituted follicles
cultured for the number of days indicated
after isolation of individual follicles.
Abbreviations are as shown in Fig. 3A. Scale

D6
bars, 100 mm. (B) Formation of TZPs. SC Merge+DAPI
Shown are fluorescence images of a COC
stained with phalloidin and the antibodies
indicated. TZPs were formed between the
granulosa cells and the oocyte (arrowheads).

D12
Scale bars, 10 mm. (C) IVM culture using

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
FOSLCs. Shown are BF and fluorescence
images of COCs before or after IVM.
Note that cumulus cells after IVM were C BF F2T SC Merge
expanded. Scale bars, 100 mm. (D) MII
oocytes and preimplantation embryos
Before IVM

derived from rOvarioids. Scale bars, 100 mm.

(E) Newborn pups derived from rOvarioids.
Shown are pups with placentae obtained
by transferring two-cell embryos.
(F) Adult mice from the newborn pups
4 weeks after birth. (G) Detection of
reporter genes in the mice derived from
After IVM

rOvarioids. Shown are gel electrophoresis

of PCR products to detect BV or SC in the
genome. (H) Fertility of the adult mice
obtained. Shown are pups derived from
the intercrossing of the adult mice derived
from rOvarioids. (I) Detection of reporter
genes in the pups derived from two pairs D MII oocytes Fertilized eggs 2-cell embryos Blastocysts
of the adult mice. Shown are the results of
gel electrophoresis of PCR products to
detect BV or SC in the genome.

E G
M NC 1 2 3 4 5 6 7 8 91011PC
BV

SC

I Parent#1 Parent#2
M NC 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 11
BV

SC

produce FOSLCs and PGCLCs, which are equiv- interactions between these cell types. Last, this somatic cells is an obstacle to the application
alent to nascent gonadal somatic cells and culture system opens the possibility of apply- of in vitro gametogenesis to mammalian spe-
premigratory PGCs (6), respectively, it becomes ing in vitro gametogenesis to other mamma- cies other than mice. FOSLCs could be an op-
possible to investigate the first and subsequent lian species. The scarcity of embryonic gonadal timal substitute for embryonic tissue because

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 7 of 8

RES EARCH | R E S E A R C H A R T I C L E

they have full potency to support oogenesis. to an Osr1-dependent nephron progenitor compartment and c-Kit. Dev. Growth Differ. 51, 567–583 (2009).
For these reasons, this system will be a use- within the mammalian kidney. Dev. Biol. 324, 88–98 (2008). doi: 10.1111/j.1440-169X.2009.01119.x; pmid: 21314674
doi: 10.1016/[Link].2008.09.010; pmid: 18835385 28. N. Hamada et al., Germ cell-intrinsic effects of sex
ful tool for solving key issues in reproductive 9. A. Rojas et al., Gata4 expression in lateral mesoderm is chromosomes on early oocyte differentiation in mice.
biology and regenerative medicine. downstream of BMP4 and is activated directly by Forkhead PLOS Genet. 16, e1008676 (2020). doi: 10.1371/
and GATA transcription factors through a distal enhancer [Link].1008676; pmid: 32214314
Methods summary element. Development 132, 3405–3417 (2005). doi: 10.1242/ 29. K. Takasawa et al., FOXL2 transcriptionally represses Sf1
dev.01913; pmid: 15987774 expression by antagonizing WT1 during ovarian development in
For FOSLC induction, Nr5A1-hCD271 reporter 10. S. Takada et al., Wnt-3a regulates somite and tailbud formation mice. FASEB J. 28, 2020–2028 (2014). doi: 10.1096/
ESCs were differentiated into EpiLCs with in the mouse embryo. Genes Dev. 8, 174–189 (1994). fj.13-246108; pmid: 24451388
doi: 10.1101/gad.8.2.174; pmid: 8299937 30. K. Miyabayashi et al., Heterogeneity of ovarian theca and
activin A and basic FGF. The EpiLCs were 11. T. P. Yamaguchi, K. Harpal, M. Henkemeyer, J. Rossant, interstitial gland cells in mice. PLOS ONE 10, e0128352 (2015).
cultured in a low-cell-binding, U-bottomed, fgfr-1 is required for embryonic growth and mesodermal doi: 10.1371/[Link].0128352; pmid: 26039146
96-well plate with 14 µM CHIR99021, 1 ng/ml patterning during mouse gastrulation. Genes Dev. 8, 3032–3044 31. R. Li, D. F. Albertini, The road to maturation: Somatic cell
(1994). doi: 10.1101/gad.8.24.3032; pmid: 8001822 interaction and self-organization of the mammalian oocyte.
BMP4, and 50 ng/ml epidermal growth factor
12. A. Iulianella, B. Beckett, M. Petkovich, D. Lohnes, A molecular Nat. Rev. Mol. Cell Biol. 14, 141–152 (2013). doi: 10.1038/
(EGF) for 2 days; then with 3 µM RA, 30 ng/ml basis for retinoic acid-induced axial truncation. Dev. Biol. 205, nrm3531; pmid: 23429793

Downloaded from [Link] at Instituto de Investigacion En Recursos Cinegeticos on November 27, 2024
Shh, 1 µM PD0325901, 50 ng/ml EGF, and 33–48 (1999). doi: 10.1006/dbio.1998.9110; pmid: 9882496 32. K. Hayashi, O. Hikabe, Y. Obata, Y. Hirao, Reconstitution of
1 ng/ml BMP4 for another 2 days; and then 13. T. Yoshino, H. Murai, D. Saito, Hedgehog-BMP signalling mouse oogenesis in a dish from pluripotent stem cells.
establishes dorsoventral patterning in lateral plate mesoderm Nat. Protoc. 12, 1733–1744 (2017). doi: 10.1038/
with 20 ng/ml BMP4 and 2 ng/ml FGF9 for to trigger gonadogenesis in chicken embryos. Nat. Commun. 7, nprot.2017.070; pmid: 28796232
another 1 or 2 days. Nr5A1-hCD271–positive 12561 (2016). doi: 10.1038/ncomms12561; pmid: 27558761 33. N. Hamazaki et al., Reconstitution of the oocyte transcriptional
FOSLCs were purified by MACS using anti- 14. Y. C. Hu, L. M. Okumura, D. C. Page, Gata4 is required for network with transcription factors. Nature 589, 264–269
formation of the genital ridge in mice. PLOS Genet. 9, (2021). doi: 10.1038/s41586-020-3027-9; pmid: 33328630
hCD271 antibody conjugated with micro- e1003629 (2013). doi: 10.1371/[Link].1003629; 34. T. Furusawa, K. Ohkoshi, C. Honda, S. Takahashi, T. Tokunaga,
beads and an MS column. For the production pmid: 23874227 Embryonic stem cells expressing both platelet endothelial
of MII oocytes in rOvarioids, FOSLCs were re- 15. D. Wilhelm, C. Englert, The Wilms tumor suppressor WT1 cell adhesion molecule-1 and stage-specific embryonic antigen-
regulates early gonad development by activation of Sf1. 1 differentiate predominantly into epiblast cells in a chimeric
aggregated with PGCLCs in a low-cell-binding, Genes Dev. 16, 1839–1851 (2002). doi: 10.1101/gad.220102; embryo. Biol. Reprod. 70, 1452–1457 (2004). doi: 10.1095/
U-bottomed, 96-well plate. The rOvarioids were pmid: 12130543 biolreprod.103.024190; pmid: 14736812
placed on Transwell-COL membranes and then 16. M. Chen et al., Wt1 directs the lineage specification of sertoli 35. M. S. Kowalczyk et al., Single-cell RNA-seq reveals changes in
and granulosa cells by repressing Sf1 expression. Development cell cycle and differentiation programs upon aging of
cultured under IVDi conditions for 21 days. 144, 44–53 (2017). pmid: 27888191 hematopoietic stem cells. Genome Res. 25, 1860–1872 (2015).
Individual follicles were isolated from the 17. X. Luo, Y. Ikeda, K. L. Parker, A cell-specific nuclear receptor is doi: 10.1101/gr.192237.115; pmid: 26430063
rOvarioids and then cultured under IVG con- essential for adrenal and gonadal development and sexual 36. Data for: T. Yoshino et al., Generation of ovarian follicles
differentiation. Cell 77, 481–490 (1994). doi: 10.1016/ from mouse pluripotent stem cells. Zenodo (2021);
ditions for 12 days. Cumulus-oocyte complexes
0092-8674(94)90211-9; pmid: 8187173 [Link]
were collected by a fine glass capillary and then 18. M. Kusaka et al., Abnormal epithelial cell polarity and ectopic
cultured under IVM conditions for 16 hours. epidermal growth factor receptor (EGFR) expression induced AC KNOWLED GME NTS
MII oocytes obtained after IVM culture were in Emx2 KO embryonic gonads. Endocrinology 151, 5893–5904 We thank F. Arai for technical support, H. Sasaki for providing
(2010). doi: 10.1210/en.2010-0915; pmid: 20962046 T-GFP mice, B. Roelen and S. M. C. de Sousa Lopes for comments
subjected to IVF. 19. S. Kuroki et al., Epigenetic regulation of mouse sex on the manuscript, and the Research Support Center, Kyushu
determination by the histone demethylase Jmjd1a. Science University Graduate School of Medical Sciences, for technical
RE FE RENCES AND N OT ES 341, 1106–1109 (2013). doi: 10.1126/science.1239864; assistance. Funding: This work was supported in part by KAKENHI
1. I. Stévant et al., Dissecting cell lineage specification and sex pmid: 24009392 Grants-in-Aid from MEXT, Japan (nos. 17H01395, 18H05544, and
fate determination in gonadal somatic cells using single-cell 20. D. Schmidt et al., The murine winged-helix transcription factor 18H05545 to K.H.; nos. 19K06678 and 20H04926 to T.Y.; no.
transcriptomics. Cell Rep. 26, 3272–3283.e3 (2019). Foxl2 is required for granulosa cell differentiation and ovary 17H06177 to R.N.; nos. 17H06427 and 20H03436 to K.M.; no.
doi: 10.1016/[Link].2019.02.069; pmid: 30893600 maintenance. Development 131, 933–942 (2004). doi: 10.1242/ 19K07378 to Y.S.; and no. 17H06424 to M.T.); by the Takeda
2. O. Hikabe et al., Reconstitution in vitro of the entire cycle of dev.00969; pmid: 14736745 Science Foundation (K.H.); by the Luca Bella Foundation (K.H.);
the mouse female germ line. Nature 539, 299–303 (2016). 21. A. Auguste et al., Loss of R-spondin1 and Foxl2 amplifies and by a Grant-in-Aid from The Open Philanthropy Project, Silicon
doi: 10.1038/nature20104; pmid: 27750280 female-to-male sex reversal in XX mice. Sex Dev. 5, 304–317 Valley Community Foundation (K.H.). Author contributions: T.Y.
3. D. G. Wilkinson, S. Bhatt, B. G. Herrmann, Expression pattern (2011). doi: 10.1159/000334517; pmid: 22116255 and K.H. conceived and designed the project. T.Y., G.N., H.K.,
of the mouse T gene and its role in mesoderm formation. 22. S. E. Gustin et al., WNT/b-catenin and p27/FOXL2 differentially and K.H performed the cellular and embryonic experiments. T.S.,
Nature 343, 657–659 (1990). doi: 10.1038/343657a0; regulate supporting cell proliferation in the developing ovary. H.Y., M.I., M.K., T.I., and K.N. performed single-cell sequencing
pmid: 1689462 Dev. Biol. 412, 250–260 (2016). doi: 10.1016/j. analysis. R.N., M.T., G.N., M.I., Y.S., and K.M. provided materials.
4. H. Kataoka et al., Expressions of PDGF receptor alpha, c-Kit ydbio.2016.02.024; pmid: 26939755 K.H. and T.Y. wrote the manuscript, incorporating feedback from
and Flk1 genes clustering in mouse chromosome 5 define 23. M. C. Nostro, X. Cheng, G. M. Keller, P. Gadue, Wnt, activin, all the authors. Competing interests: The authors declare no
distinct subsets of nascent mesodermal cells. Dev. Growth and BMP signaling regulate distinct stages in the competing financial interests. Data and materials availability:
Differ. 39, 729–740 (1997). doi: 10.1046/j.1440-169X.1997. developmental pathway from embryonic stem cells to blood. The RNA-sequencing data have been deposited at the Gene
t01-5-00009.x; pmid: 9493833 Cell Stem Cell 2, 60–71 (2008). doi: 10.1016/ Expression Omnibus (GEO) database under accession number
5. Y. Imuta, H. Kiyonari, C. W. Jang, R. R. Behringer, H. Sasaki, [Link].2007.10.011; pmid: 18371422 GSE151143. R scripts generated for the analysis are available on
Generation of knock-in mice that express nuclear enhanced 24. Y. Sasagawa et al., Quartz-Seq: A highly reproducible and GitHub ([Link] and
green fluorescent protein and tamoxifen-inducible Cre sensitive single-cell RNA sequencing method, reveals Zenodo (36). All materials are available from the corresponding
recombinase in the notochord from Foxa2 and T loci. Genesis 51, non-genetic gene-expression heterogeneity. Genome Biol. 14, authors upon request.
210–218 (2013). doi: 10.1002/dvg.22376; pmid: 23359409 R31 (2013). doi: 10.1186/gb-2013-14-4-r31; pmid: 23594475
6. K. Hayashi, H. Ohta, K. Kurimoto, S. Aramaki, M. Saitou, 25. F. Buettner et al., Computational analysis of cell-to-cell SUPPLEMENTARY MATERIALS
Reconstitution of the mouse germ cell specification pathway in heterogeneity in single-cell RNA-sequencing data reveals
[Link]/content/373/6552/eabe0237/suppl/DC1
culture by pluripotent stem cells. Cell 146, 519–532 (2011). hidden subpopulations of cells. Nat. Biotechnol. 33, 155–160
Materials and Methods
doi: 10.1016/[Link].2011.06.052; pmid: 21820164 (2015). doi: 10.1038/nbt.3102; pmid: 25599176
Figs. S1 to S11
7. M. Mahlapuu, M. Ormestad, S. Enerbäck, P. Carlsson, The 26. L. Mork et al., Temporal differences in granulosa cell
Tables S1 to S6
forkhead transcription factor Foxf1 is required for specification in the ovary reflect distinct follicle fates in mice.
References (37–43)
differentiation of extra-embryonic and lateral plate mesoderm. Biol. Reprod. 86, 37 (2012). doi: 10.1095/
MDAR Reproducibility Checklist
Development 128, 155–166 (2001). doi: 10.1242/dev.128.2.155; biolreprod.111.095208; pmid: 21976597
pmid: 11124112 27. Y. Morita-Fujimura, Y. Tokitake, Y. Matsui, Heterogeneity of View/request a protocol for this paper from Bio-protocol.
8. J. W. Mugford, P. Sipilä, J. A. McMahon, A. P. McMahon, Osr1 mouse primordial germ cells reflecting the distinct status of
expression demarcates a multi-potent population of their differentiation, proliferation and apoptosis can be 27 July 2020; accepted 28 May 2021
intermediate mesoderm that undergoes progressive restriction classified by the expression of cell surface proteins integrin a6 10.1126/science.abe0237

Yoshino et al., Science 373, eabe0237 (2021) 16 July 2021 8 of 8