Article

Ex utero mouse embryogenesis from

pre-gastrulation to late organogenesis

https://doi.org/10.1038/s41586-021-03416-3 Alejandro Aguilera-Castrejon1,10,11 ✉, Bernardo Oldak1,10, Tom Shani1, Nadir Ghanem2,

Chen Itzkovich3, Sharon Slomovich4, Shadi Tarazi1, Jonathan Bayerl1, Valeriya Chugaeva1,
Received: 30 June 2020
Muneef Ayyash1, Shahd Ashouokhi1, Daoud Sheban1, Nir Livnat1, Lior Lasman1, Sergey Viukov1,
Accepted: 4 March 2021 Mirie Zerbib1, Yoseph Addadi5, Yoach Rais6, Saifeng Cheng6, Yonatan Stelzer6,
Hadas Keren-Shaul7, Raanan Shlomo8, Rada Massarwa1,11,12, Noa Novershtern1,11, Itay Maza4,9,11 ✉
Published online: 17 March 2021
& Jacob H. Hanna1,11 ✉
Check for updates

The mammalian body plan is established shortly after the embryo implants into the
maternal uterus, and our understanding of post-implantation developmental
processes remains limited. Although pre- and peri-implantation mouse embryos are
routinely cultured in vitro1,2, approaches for the robust culture of post-implantation
embryos from egg cylinder stages until advanced organogenesis remain to be
established. Here we present highly effective platforms for the ex utero culture of
post-implantation mouse embryos, which enable the appropriate development
of embryos from before gastrulation (embryonic day (E) 5.5) until the hindlimb
formation stage (E11). Late gastrulating embryos (E7.5) are grown in three-
dimensional rotating bottles, whereas extended culture from pre-gastrulation stages
(E5.5 or E6.5) requires a combination of static and rotating bottle culture platforms.
Histological, molecular and single-cell RNA sequencing analyses confirm that the
ex utero cultured embryos recapitulate in utero development precisely. This culture
system is amenable to the introduction of a variety of embryonic perturbations and
micro-manipulations, the results of which can be followed ex utero for up to six days.
The establishment of a system for robustly growing normal mouse embryos ex utero
from pre-gastrulation to advanced organogenesis represents a valuable tool for
investigating embryogenesis, as it eliminates the uterine barrier and allows
researchers to mechanistically interrogate post-implantation morphogenesis and
artificial embryogenesis in mammals.

Understanding the developmental processes that lead to the for- after culture initiation. Thus, stable and efficient protocols for extended
mation of tissues and organs represents a fundamental question in culturing of pre-gastrulating mouse embryos until advanced organo-
developmental biology. In mammals, this process takes place after the genesis are still required11.
embryo implants into the uterus, which makes it relatively inaccessible
for observation and manipulation3,4. Consequently, the sequence of
developmental events that takes place from pre-gastrulation to organo- Enhanced ex utero roller culture platform
genesis remains to be fully understood and is difficult to manipulate. We set out to test whether some of the newly established cell cul-
Establishing culture conditions that sustain the proper long-term ture supplements or biomechanical principles that have arisen in
development of post-implanted mouse embryos outside the uterine stem cell research could be helpful for addressing this challenge
environment remains a challenge. Several useful culture techniques (for example, pressure control, supplements or synthetic sera12–14).
have been proposed since the 1930s, including culturing the embryos We used the roller culture system on a drum and integrated it with a
in conventional static conditions5,6, in rotating bottles on a drum (‘roller customized, in-house-developed electronic gas regulation module
culture systems’)7 or in circulator systems8. However, these platforms that not only allows the precise and sensitive control of O2 and CO2
remain highly inefficient for normal embryo survival, use already gas- levels, but also allows atmospheric gas pressure to be controlled
trulating embryos and are limited to short periods of time9,10, as the (Fig. 1a, b, Extended Data Fig. 1, Supplementary Video 1). The latter
embryos begin to display developmental anomalies as early as 24 h was motivated by the ability of pressure to enhance oxygen delivery

1
Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel. 2Department of Obstetrics and Gynecology, Rambam Health Care Campus, Haifa, Israel. 3The Clinical
Research Institute at Ramban (CRIR), Rambam Health Care Campus, Haifa, Israel. 4Bruce Rappaport Faculty of Medicine, Israel Institute of Technology - Technion, Haifa, Israel. 5Department of
Life Sciences Core Facilities, Weizmann Institute of Science, Rehovot, Israel. 6Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel. 7Department of Life
Sciences Core Facilities, Weizmann Institute of Science, Rehovot, Israel. 8Arad Technologies Ltd, Ashdod, Israel. 9Gastroenterology Unit, Rambam Health Care Campus, Haifa, Israel. 10These
authors contributed equally: Alejandro Aguilera-Castrejon, Bernardo Oldak. 11These authors jointly supervised this work: Alejandro Aguilera-Castrejon, Rada Massarwa, Noa Novershtern, Itay
Maza, Jacob H. Hanna. 12Deceased: Rada Massarwa. ✉e-mail: alejandroac@weizmann.ac.il; i_maza@rambam.health.gov.il; jacob.hanna@weizmann.ac.il

Nature | Vol 593 | 6 May 2021 | 119

Article
a Roller culture E7.5 +Day 1 +Day 2 +Day 3 +Day 4

Ex utero 5% O2 13% O2 18% O2 21% O2

culture medium 5% CO2 5% CO2 5% CO2 5% CO2
6.5 psi 6.5 psi 6.5 psi 6.5 psi
+3 mg ml–1 glucose +3.5 mg ml–1 glucose +4 mg ml–1 glucose

b c
E7.5 EPC E8.5 E9.5 E10.5 E11.5

Intact yolk sac

YS Am
Ch
Gas mixing Roller culture
box incubator Am
Fg

In utero
Epi 6–8 s 23–26 s 35–38 s >45 s
Rho Ms Ms

Dissected yolk sac

Mt D
Gas regulation Al Mx OP My T
OtP LV
module NF Md Pro OlP
Rotating PN
Sc TB
drum H
FL HL
S S

E7.5 EPC +Day 1 +Day 2 +Day 3 +Day 4

Intact yolk sac

YS
Ch Am UC
d 100 Am
YSV
Embryos developed

Ex utero

80 Fg
Epi 6–8 s 23–26 s 35-38 s 44 s
normally (%)

Ms Mt Ms

Dissected yolk sac

60 Rho D
My
NF OtP Mx OP T LV
n = 230, 15x

n = 163, 12x
n = 230, 15x

n = 185, 13x

40 Al Md Pro OlP
H Sc PN TB
20 HL
S
S FL

1 2 3 4
ay ay ay ay
+D +D +D +D f
NS E7.5 +Day 1 +Day 2 +Day 3 +Day 4
P = 0.1431 SOX2
e DAPI
**
P = 0.0048
6,000
Length (μm)

NS
P = 0.9321
4,000 NS
P = 0.9507 E7.5 +Day 1 +Day 2 +Day 3 +Day 4
2,000 SOX9
DAPI
E8.5 +Day 1 E9.5 +Day 2 E10.5 +Day 3 E11 +Day 4

Fig. 1 | Ex utero culture system for growing mouse late-gastrulating Whitney test. f, SOX2 and SOX9 whole-mount immunofluorescence of embryos
embryos until advanced organogenesis. a, Schematic of the E7.5 embryo ex developed ex utero from E7.5. Images represent a minimum of three biological
utero culture platform. b, Electronic gas and pressure regulation module replicates. Am, amnion; Al, allantois; Ch, chorion; D, diencephalon; Epi,
connected to the roller culture incubator system. c, Bright-field images of epiblast; EPC, ectoplacental cone; Fg, foregut pocket; FL, forelimb bud; H,
embryos developing in utero from E7.5 to E11.5 and equivalent embryos heart; HL, hindlimb bud; LV, lens vesicle; Md, mandibular arch; Ms,
cultured ex utero. d, Percentage of developmentally normal embryos per mesencephalon; Mt, metencephalon; Mx, maxillary arch; My, myelencephalon;
culture day. n, total number of embryos; x, number of experiments. All data are NF, neural folds; OlP, olfactory placode; OP, optic pit; OtP, otic pit; PN, posterior
mean ± s.e.m. e, Quantification of embryonic length for in utero and cultured neuropore; Pro, prosencephalon; Rho, rhombencephalon; S, somites; Sc,
embryos. Dots represent individual embryos. Numbers of embryos, left to spinal cord; T, telencephalon; TB, tail bud; UC, umbilical cord; YS, yolk sac; YSV,
right: in utero, 13, 19, 15, 38; ex utero, 32, 15, 43, 41; NS, not significant; Mann– yolk sac vessel. Scale bars, 500 μm.

to tissues and by demonstrations that atmospheric pressure can alter of 6–7 psi (Extended Data Fig. 2b). This protocol yielded approximately
cell growth13,14. We established conditions that support the growth 77% normal embryo development after four days in culture in differ-
of E7.5 late-gastrulating embryos (neural plate and headfold stage15) ent mouse strains (Fig. 1d, Extended Data Fig. 2c). After four days, the
until the hindlimb formation stage (around E11) with high efficiency embryos started to show abnormalities, yolk sac circulation abrup-
(Fig. 1a–c, Supplementary Video 2, Extended Data Fig. 2a, b). First, a tion and pericardial effusion, and quickly died overnight. The latter
mixture of 25% Dulbecco’s modified Eagle’s medium (DMEM), 50% rat effects are consistent with the development of hydrops fetalis due to
serum and 25% human umbilical cord blood serum (HCS) consistently insufficient oxygenation and nutrient supply by the ex utero system
supports embryo growth with much higher efficiency than rat serum (given the lack of maternal blood supply in this setting) that no longer
only (Extended Data Fig. 2b), and we designated this medium as ex utero matches the increased body size at E11.
culture medium (EUCM). Notably, supplementing EUCM with extra We evaluated previously defined morphological landmarks16 to
glucose every 24 h and until the end of the culture period was essential assess appropriate embryo development ex utero (Fig. 1c, Supple-
for overcoming developmental abnormalities after two days in culture mentary Video 3, Supplementary Methods). On the last day of culture,
(Extended Data Fig. 2b). The application of sequential increases in O2 maximum embryo growth was reached at about 44 somites, equivalent
levels every 24 h, from 5% O2 at E7.5, 13% at E8.5, 18% at E9.5, to 21% O2 at to approximately E11 (Theiler stage 18). The length of the cultured
E10.5 was most optimal (Extended Data Fig. 2b). In addition, normal and embryos was comparable to matched in utero embryos (Fig. 1e). We
efficient development depended on the maintenance of a gas pressure analysed eleven developmental markers, all of which showed consistent

120 | Nature | Vol 593 | 6 May 2021

 a Day 0 +Day 1 +Day 2 b E6.5 +Day 1 +Day 2
E6.5 YS
EPC

Bright field
Al
Ch NF
ExE
AB H
VE PS Am
Static culture AC Fg S
Epi Epi
Ibidi plate EUCM 21% O2 98% 97%
c 15 d 15
Ex utero +Day 2 Extraembryonic Foregut
Mid-hindgut
In utero E8.5 endoderm

10 10
Extraembryonic
ectoderm
5 5 Blood Blood Somitic mesoderm
Amnion Placodes

UMAP-2
UMAP-2

0 0 Endothelial Amnion
Pharingeal
mesoderm Amnion
–5 –5 Endothelial
Extraembryonic
mesoderm
Cardiac
Presomitic/mixed
–10 –10 Neural tube
mesoderm
Neural tube
Mid-hindbrain
–10 –5 0 5 10 –10 –5 0 5 10
UMAP-1 UMAP-1

Fig. 2 | Defining conditions for recapitulating mouse gastrulation ex utero. dots). UMAP plot displaying individual cells (n = 6,358 ex utero +Day 2; n = 4,349
a, Static culture protocol for growing gastrulating embryos until in utero E8.5). d, Cell lineage annotation of clusters based on marker genes of
somitogenesis. b, Bright-field images of embryos developing ex utero from the major cell types identified in E8.5 mouse embryos19. Points are coloured
E6.5 until E8.5; bottom right, percentage of properly developed embryos. according to their assigned cell cluster. AB, allantoic bud; ExE, extraembryonic
n = 421 at +Day 1, n = 399 at +Day 2. c, scRNA-seq analysis of in utero E8.5 ectoderm; PS, primitive streak; VE, visceral endoderm. Scale bars, 100 μm.
embryos (purple dots) versus E6.5 +Day 2 ex utero developing embryos (green

spatio-temporal gene expression patterns between embryos that devel- distribution of cell states overlapped strongly between in utero and
oped in utero or ex utero (Fig. 1f, Extended Data Figs. 3, 4, Supplemen- ex utero embryos (Fig. 2c). The identity of each cluster was annotated
tary Video 4). Mouse transgenic lines expressing the GFP reporter for using specific marker genes of cell lineages that have been previously
imprinting erasure of the Dlk1–Dio3 intergenic differentially meth- defined by single-cell transcriptomics of early mouse embryos19,20
ylated region (DMR)17 in migrating primordial germ cells, or under (Extended Data Fig. 8c, e). We identified derivatives of the three germ
tissue-specific promoters (Wnt1-Cre and Isl1-Cre), showed that the layers as well as extraembryonic tissues, and the profile of cell types
GFP expression patterns in the cultured transgenic embryos resemble found in embryos developing ex utero was equivalent to that in utero
those of in utero embryos (Extended Data Fig. 5). These data suggest (Fig. 2c, d). In summary, the alternative static conditions generated
that the ex utero cultured embryos recapitulate development properly herein faithfully recapitulate embryo development ex utero from the
until approximately the 44-somite stage. onset of gastrulation until somitogenesis (E6.5 to E8.5).

Capturing authentic gastrulation ex utero Extended ex utero embryogenesis

We aimed to expand the ex utero culture protocol by establishing We next tested the ability to bridge mouse pre-gastrulation develop-
conditions to grow the mouse embryo from pre-gastrulation stages ment to advanced organogenesis in culture by combining our static and
(E5.5–6.5). Explanted E6.5 embryos grown in rotating bottles either roller culture protocols. After two days of static culture from E6.5, early
in 5% or 21% O2, or under previously described static conditions18, did somite-stage embryos (E8.5) were transferred into the roller culture
not develop beyond the early somite stage (Extended Data Fig. 6d, e, protocol, which allowed normal development of embryos isolated at
o, Supplementary Discussion). Thus, we set out to devise alternative early-streak stages until the hindlimb formation stage (E11), with 40%
culture parameters for growing embryos from E6.5 to E8.5 in static efficiency (Fig. 3a–c, Extended Data Fig. 6b, Supplementary Video 6).
conditions (Fig. 2a, Extended Data Fig. 6). The following conditions Only those embryos cultured in normoxia from E6.5 to E8.5 were able
were optimized: 25% DMEM, 50% rat serum and 25% HCS in 21% O2. to continue growing after being moved to the roller culture system,
The latter supported the development of early-streak embryos (E6.5) which indicates that these conditions support appropriate embryo
in static culture until the early somite stage (48 h) with up to 97% effi- development (Extended Data Fig. 6a, b, g, h). Culturing the embryos in a
ciency (Fig. 2a, b, Extended Data Fig. 6a, b). Embryos grown in utero constant atmosphere of 21% oxygen through 5 days of culture increased
or ex utero were equivalent morphologically and in the expression of the efficiency of development to 55% (Extended Data Fig. 9a–d). The
all lineage markers analysed (Extended Data Fig. 7). The robustness of same protocol and conditions were able to support the development
these culture conditions allowed in toto imaging of the gastrulating of pre-primitive streak (E5.5) mouse embryos for a total of 6 days ex
mouse embryo for 58 h (Supplementary Video 5). utero, with 46% of embryos reaching E8.5 and about 20% reaching the
To characterize the various lineages present in the embryos, and to 42-somite stage (Fig. 3d, Extended Data Fig. 9e–h, Supplementary
identify to what extent the global transcriptional profile of embryos Video 7). Even though a delay of 1–2 pairs of somites seemed to arise
developing in culture mimics their in utero counterparts, we performed in the timing of developmental events from E6.5, and of 2–4 pairs of
single-cell RNA sequencing (scRNA-seq) on embryos grown ex utero somites from E5.5, embryo and tissue morphogenesis proceeded prop-
at day two (E6.5 + 2 days) and compared the results to those from cells erly (Fig. 3b, d, Extended Data Fig. 9). As with in vivo development, the
obtained from equivalent embryos developing in utero (Extended embryos increased in size from about 200 μm at E6.5 to about 5.4 mm
Data Fig. 8a, b). Clustering analysis based on differentially expressed at the 44-somite stage (Extended Data Fig. 9a–d). Immunofluores-
genes revealed 19 different cell states (Extended Data Fig. 8c). The cence analyses of developmental genes confirmed that these markers

Nature | Vol 593 | 6 May 2021 | 121

Article
a E6.5 Static culture Roller culture
Day 0 +Day 1 +Day 2 +Day 3 +Day 4 +Day 5

+3 mg ml–1 +3.5 mg ml–1 +4 mg ml–1

E6.5 E7.5 E8.5 glucose E9.5 glucose E10.5 glucose E11
Day 0 +Day 1 +Day 2 +Day 3 +Day 4 +Day 5 +Day 6
E5.5 Static culture Roller culture

21% O2 Dynamic or 21% O2

EUCM EUCM 5% CO2
5% CO2
6.5 psi
b
E6.5 +Day 1 +Day 2 +Day 3 +Day 4 +Day 5
Bright field

4–8 somites 21–25 somites 33–36 somites 44 somites

c d
GATA4 MHC-II SOX2

+Day 3 +Day 4 +Day 5 E5.5 +Day 6

42 somites

e f Neural progenitor cells

Ex utero +Day 4
In utero E10.5 Radial glia
Schwann cell
10 10 Glial cells precursors
Radial glia Erythroid
Notochord lineage
Cholinergic
Limb mesenchyme
neurons
Chondrocytes Primitive erythroid
UMAP-2
UMAP-2

and osteoblasts lineage

0 0 Chondrocytes
and
Erythroid
osteoblasts
lineage
Jaw and tooth
progenitors Hepatocytes
Osteoblasts Myocytes

–10 –10
Osteoblasts
Epithelial cells

White blood cells

White blood cells

–10 –5 0 5 10 –10 –5 0 5 10
UMAP-1 UMAP-1

Fig. 3 | Extending the mouse embryo ex utero culture system from cultured for 6 days until the 42-somite stage. n ≥ 5 embryos. Scale bars, 50 μm
pre-gastrulation to advanced organogenesis. a, Schematic protocol for (E5.5), 500 μm (all others). e, Comparative scRNA-seq analysis of E6.5 +Day 4 ex
culturing mouse embryos from pre-gastrulation to organogenesis. utero embryos (green dots) and equivalent E10.5 embryos developing in utero
b, Bright-field images of embryos grown for five days ex utero from E6.5 to the (purple dots). UMAP plot depicting all cells considered in the analysis
44-somite stage. Embryos cultured beyond day two are shown without the (n = 39,374 ex utero; n = 24,107 in utero). f, Cell lineage annotation of clusters
yolk sac. The variation in somite number is indicated. n ≥ 65 embryos. based on the expression of marker genes described in the mouse
c, Immunostaining of early-gastrulating embryos grown ex utero for 3, 4 and organogenesis cell atlas21. Points are coloured according to their assigned cell
5 days. Images are representative of at least three biological replicates. cluster.
d, Representative bright-field images of pre-gastrulating E5.5 embryos

were located according to their expected expression patterns (Fig. 3c, organs and tissues derived from all three germ layers, consistent with
Extended Data Figs. 7, 9h). HCS could be replaced with serum isolated advanced organogenesis stages (Fig. 3f). Analysis of differentially
from human adult blood (HBS) to allow ex utero development until expressed genes revealed a high correlation (about 0.9) between ex
the hindlimb formation stage (44 somites), starting from E6.5 and E7.5 utero and in utero embryos for all cell states, with the most variable
(Extended Data Fig. 10). cluster showing only 0.4% (8 out of 2,000) differentially expressed
We characterized the transcriptional profiles of cells isolated from genes (Extended Data Fig. 8g). This minimal difference in blood and
embryos grown ex utero and in utero matched embryos by scRNA-seq cardiac gene expression signature at E10.5 (Extended Data Fig. 8g)
(Extended Data Fig. 8a, b). The cells profiled were grouped into 20 could be consistent with early signs of hydrops fetalis in the embryos.
clusters21 (Extended Data Fig. 8d, f). Our analysis confirmed that the Comparison of the relative cell proportions across cell types showed
composition of cell transcriptional states in the embryos that had devel- no significant differences in the majority of clusters, while minor dif-
oped ex utero until advanced organogenesis (E6.5 + 4 days) was equiva- ferences were found in only three clusters (Extended Data Fig. 8h).
lent to that of their in vivo counterparts (Fig. 3e). The annotated cell Collectively, these results show that embryos developing ex utero from
clusters identified represent lineage-committed cell types comprising pre-gastrulation stages, by a combination of static and rolling bottle

122 | Nature | Vol 593 | 6 May 2021

 a E7.5 24 h c E6.5

Embryo
culture Electroporation Back to culture Analysis Lentivirus injection Ex utero culture Analysis

Days after electroporation Days after lentiviral infection

b d
+Day 1 +Day 2 +Day 3 +Day 1 +Day 2 +Day 3 +Day 4 +Day 5
GFP

GFP
GATA4 SOX2 GFP
TUJ1 SOX2 GFP

e In vitro f Chimerism level in positive embryos +Day 1

>10,000 * +Day 2
mPS cells
**

Integrated cells per

E7.5
Injection >1,000
*** *** +Day 3
+Day 4

embryo
100
A P
D Ex utero culture Analysis 10
E7.5
h E7.5 E9.5–E11

epiblast EpiS EpiL E7.5

cells cells epiblast Human
microglia
progenitors

g Mouse EpiS cells Mouse EpiL cells Mouse E7.5 epiblast

+Day 1 +Day 2 +Day 1 +Day 2 +Day 1 +Day 4

GFP TUJ1 DAPI

GATA4 SOX9 +Day 3
GFP SOX2 GATA4

tdT tdT
SOX2 TUJ1
Ventral
ntral

rsal
sal
Dorsal
a
r

Dors
rs
Vent
ent
en
e

Ventral Dorsal

Fig. 4 | Measuring functional outcomes of perturbations introduced into individual embryos; data are mean ± s.e.m. Embryos with no contribution are
ex utero whole-embryo culture platform. a, Schematic illustration of the not represented in the graph. Numbers of embryos analysed: EpiS cells, 21, 39,
ex utero electroporation protocol at E8.5. b, Immunofluorescence of 18, 8; EpiL cells, 22, 9; E7.5 epiblasts, 34, 12, 5, 7 (left to right); Mann–Whitney
electroporated embryos stained for GFP, SOX2 and TUJ1. n = 17, 15 and 11 test. ***P < 0.0001; **P = 0.001; *P = 0.0025. g, Immunostaining of chimeric
embryos (left to right). c, Lentiviral transduction of E6.5 mouse embryos. embryos injected at E7.5 with GFP–EpiS cells, GFP–EpiL cells or tdTomato
d, GFP, SOX2 and GATA4 immunostaining of E6.5 embryos transduced with GFP (tdT)–E7.5 epiblasts and cultured ex utero for 1–4 days. h, Top, generation of
using lentivirus and grown ex utero for 1–5 days. n = 15, 24, 19, 16 and 20 embryos human–mouse microglia chimeras. n = 11 embryos. Bottom, whole-mount
(left to right). e, Generation of post-implantation chimeras by microinjection immunostaining of chimeric embryos 3 days after injection. Scale bars,
of primed EpiS cells, EpiL cells and E7.5 in vivo epiblasts. A, anterior; D, distal; 500 μm.
P, posterior. f, Quantification of GFP+ cells in chimeric embryos. Dots represent

cultures, are capable of proper symmetry breaking, establishment of The ability to perform genetic modifications by lentiviral transduc-
the germ layers and embryonic axis, and subsequently differentiation tion23 was also shown in E6.5 embryos by microinjecting lentivirus
and patterning of tissues and organs without maternal interaction harbouring a gene encoding enhanced GFP (EGFP) (Fig. 4c). Lentiviral
over a period of six days from the symmetric pluripotent epiblast to transduction yielded an embryo survival rate similar to that of con-
the advanced organogenesis stages. trols and did not affect morphology or tissue differentiation (Fig. 4d,
Extended Data Fig. 11d). After 24 h, GFP was detected throughout the
epiblast and extraembryonic tissues, and by the last culture day, GFP
Perturbation of post-implantation development expression was extensively spread over the embryo and yolk sac in more
One major advantage of this ex utero culture platform is the ability to than 90% of the embryos (Fig. 4d, Extended Data Fig. 11e).
apply manipulations in post-implanted mouse embryos, and to follow Next, we harnessed the ex utero culture platform to analyse chimeric
their effects on the same embryos after several days of further ex utero mouse embryos obtained after microinjection of primed pluripotent
development. We performed whole-embryo electroporation4,22 of a stem cells (PS cells) at post-implantation stages24. The ability to evaluate
fluorescent marker at early E8.5 (before neural tube closure) followed the chimeric potential of primed mouse PS cells has been limited by the
by long-term ex utero culture (72 h). Embryos at E7.5 were dissected lack of protocols that enable transfer of post-implantation embryos
and cultured for 24 h. Afterwards, a GFP plasmid vector was injected in utero or their ex utero culture for prolonged periods, which has now
into the neural tube and electroporated to label a population of neural been achieved herein. Thus, we microinjected clusters of GFP-labelled
cells. Electroporated embryos were then put back in culture for up to mouse epiblast stem cells (EpiS cells) or epiblast-like stem cells (EpiL
three days (Fig. 4a). Sixty-eight per cent of the embryos developed cells) into the anterior, distal or posterior epiblast of E7.5 embryos,
properly until the hindlimb stage after electroporation (Extended Data which were subsequently cultured ex utero (Fig. 4e, Extended Data
Fig. 11a). Cells that expressed GFP were widely distributed in the neural Fig. 11f, g, i–k). After 24 h, we observed chimerism efficiency of 50–60%
tissues after 1–3 days of culture in 75% of embryos (Fig. 4b, Extended for both EpiS cells (27 of 49 embryos) and EpiL cells (44 of 69 embryos)
Data Fig. 11b, c). injected into the posterior epiblast, with an estimated number of

Nature | Vol 593 | 6 May 2021 | 123

Article
transplant-derived cells of 10 to 100 cells distributed along the embryo and set the stage for expanding ex utero embryo research from different
body axis (Fig. 4f, g, Extended Data Fig. 11i, j), consistent with previous mammalian species and from stem cell aggregated synthetic embryos30.
studies25–27. Co-immunostaining for SOX2 and GATA4 confirmed that the
cells integrated into embryonic tissues (Fig. 4g, Extended Data Fig. 11i).
However, the number of EpiS cell-derived GFP+ cells decreased over the Online content
subsequent three days and these cells were outcompeted by the cells Any methods, additional references, Nature Research reporting sum-
of the host (Fig. 4f, g, Extended Data Fig. 11i, j). Low integration was also maries, source data, extended data, supplementary information,
evident when microinjecting EpiS cells and EpiL cells into the anterior acknowledgements, peer review information; details of author con-
and distal epiblasts (Extended Data Fig. 11k). Isogenic naive embryonic tributions and competing interests; and statements of data and code
stem (ES) cells microinjected into blastocysts that were subsequently availability are available at https://doi.org/10.1038/s41586-021-03416-3.
transferred and re-isolated at E7.5 and subjected to ex utero culture
yielded high-contribution chimeras (Extended Data Fig. 11h). This find- 1. Bedzhov, I. & Zernicka-Goetz, M. Self-organizing properties of mouse pluripotent cells
ing excludes genetic background or ex utero culture as the underlying initiate morphogenesis upon implantation. Cell 156, 1032–1044 (2014).
cause of limited chimeric integration of in vitro-derived primed PS 2. White, M. D. et al. Long-lived binding of Sox2 to DNA predicts cell fate in the four-cell
mouse embryo. Cell 165, 75–87 (2016).
cells. Finally, we injected primed cell clusters isolated directly from 3. New, D. A. T. Whole-embryo culture and the study of mammalian embryos during
tdTomato+ E7.5 embryonic epiblasts into recipient embryos (Fig. 4e). organogenesis. Biol. Rev. Camb. Philos. Soc. 53, 81–122 (1978).
4. Huang, Q. et al. Intravital imaging of mouse embryos. Science 368, 181–186 (2020).
Unlike EpiS cells or EpiL cells, in vivo-derived E7.5 epiblast orthotopic
5. Nicholas, J. S. & Rudnick, D. The development of rat embryos in tissue culture. Proc. Natl
grafts contributed extensively and adequately to chimeric embryos Acad. Sci. USA 20, 656–658 (1934).
across different tissues (more than 10,000 integrated cells at the last 6. New, D. A. T. & Stein, K. F. Cultivation of mouse embryos in vitro. Nature 199, 297–299
(1963).
day of culture; Fig. 4f, g, Extended Data Fig. 11j, l). These results suggest
7. New, D. A. T., Coppola, P. T. & Terry, S. Culture of explanted rat embryos in rotating tubes.
that, in relation to their in vivo counterparts, in vitro primed PS cells J. Reprod. Fertil. 35, 135–138 (1973).
possess a limited capacity to expand and incorporate into host tissues 8. New, D. A. T. Development of explanted rat embryos in circulating medium. Development
17, 513–525 (1967).
even when they are injected into post-implantation stages and allowed
9. Beddington, R. S. Induction of a second neural axis by the mouse node. Development
to undergo advanced organogenesis ex utero. 120, 613–620 (1994).
In toto confocal live imaging can be applied to ex utero developed 10. Parameswaran, M. & Tam, P. P. L. Regionalisation of cell fate and morphogenetic
movement of the mesoderm during mouse gastrulation. Dev. Genet. 17, 16–28 (1995).
embryos, as shown by imaging of neural tube closure in tdTomato+
11. Tam, P. P. Postimplantation mouse development: whole embryo culture and
mouse embryos, which were maintained ex utero from E7.5 and sub- micro-manipulation. Int. J. Dev. Biol. 42, 895–902 (1998).
sequently mounted for live confocal imaging at E9.0 to visualize the 12. Cantor, J. R. et al. Physiologic medium rewires cellular metabolism and reveals uric acid
as an endogenous inhibitor of UMP synthase. Cell 169, 258–272 (2017).
dynamics of convergence and closure of the neural folds for about 9 h
13. Nagamatsu, G., Shimamoto, S., Hamazaki, N., Nishimura, Y. & Hayashi, K. Mechanical
(Extended Data Fig. 11m, Supplementary Video 8). Furthermore, the stress accompanied with nuclear rotation is involved in the dormant state of mouse
teratogenic effects of valproic acid on neural tube closure could be oocytes. Sci. Adv. 5, eaav9960 (2019).
14. Ueda, Y. et al. Intrauterine pressures adjusted by Reichert’s membrane are crucial for early
recapitulated by supplying this teratogen to the embryo environment mouse morphogenesis. Cell Rep. 31, 107637 (2020).
ex utero (Extended Data Fig. 11n). 15. Downs, K. M. & Davies, T. Staging of gastrulating mouse embryos by morphological
Finally, we derived GFP-labelled primitive microglia progenitors landmarks in the dissecting microscope. Development 118, 1255–1266 (1993).
16. van Maele-Fabry, G., Delhaise, F. & Picard, J. J. Evolution of the developmental scores of
from human PS cells28 (Extended Data Fig. 12a, b), and microinjected sixteen morphological features in mouse embryos displaying 0 to 30 somites. Int. J. Dev.
them into mouse embryos at E7.5 followed by ex utero culture (Fig. 4h, Biol. 36, 161–167 (1992).
Extended Data Fig. 12c). Analysis of integrated human cells revealed that 17. Stelzer, Y. et al. Parent-of-origin DNA methylation dynamics during mouse development.
Cell Rep. 16, 3167–3180 (2016).
microglia precursors robustly integrated, proliferated and migrated 18. McDole, K. et al. In toto imaging and reconstruction of post-implantation mouse
into the host brain (Fig. 4h, Extended Data Fig. 12d–g). The microglial development at the single-cell level. Cell 175, 859–876 (2018).
identity of the injected cells was confirmed by the presence of cells that 19. Ibarra-Soria, X. et al. Defining murine organogenesis at single-cell resolution reveals a
role for the leukotriene pathway in regulating blood progenitor formation. Nat. Cell Biol.
were double-positive for GFP and TMEM119 (Extended Data Fig. 12f). 20, 127–134 (2018).
GFP+ human cells were also detected circulating through the yolk sac 20. Pijuan-Sala, B. et al. A single-cell molecular map of mouse gastrulation and early
and yolk sac vessels, indicating that human microglia progenitors can organogenesis. Nature 566, 490–495 (2019).
21. Cao, J. et al. The single-cell transcriptional landscape of mammalian organogenesis.
migrate through the mouse embryonic circulation (Extended Data Nature 566, 496–502 (2019).
Fig. 12g). These results demonstrate the usability of the platform 22. Saito, T. & Nakatsuji, N. Efficient gene transfer into the embryonic mouse brain using
described herein to shed light on the development of human cells in in vivo electroporation. Dev. Biol. 240, 237–246 (2001).
23. Beronja, S., Livshits, G., Williams, S. & Fuchs, E. Rapid functional dissection of genetic
the context of cross-species embryonic chimeras29. networks via tissue-specific transduction and RNAi in mouse embryos. Nat. Med. 16,
821–827 (2010).
24. Morgani, S., Nichols, J. & Hadjantonakis, A. K. The many faces of pluripotency: in vitro
adaptations of a continuum of in vivo states. BMC Dev. Biol. 17, 7 (2017).
Discussion 25. Kojima, Y. et al. The transcriptional and functional properties of mouse epiblast stem cells
We have established a robust embryo culture system that faithfully resemble the anterior primitive streak. Cell Stem Cell 14, 107–120 (2014).
recapitulates mouse in utero post-implantation development from 26. Wu, J. et al. An alternative pluripotent state confers interspecies chimaeric competency.
Nature 521, 316–321 (2015).
pre-gastrulation to advanced organogenesis stages, enabling the 27. Huang, Y., Osorno, R., Tsakiridis, A. & Wilson, V. In vivo differentiation potential of epiblast
application and monitoring of external and internal manipulations in stem cells revealed by chimeric embryo formation. Cell Rep. 2, 1571–1578 (2012).
mouse embryos for up to six days of post-implantation development. 28. van Wilgenburg, B., Browne, C., Vowles, J. & Cowley, S. A. Efficient, long term production
of monocyte-derived macrophages from human pluripotent stem cells under
The ability to remove a mammalian embryo from the uterine environ- partly-defined and fully-defined conditions. PLoS One 8, e71098 (2013).
ment and grow it normally under controlled conditions constitutes a 29. Gafni, O. et al. Derivation of novel human ground state naive pluripotent stem cells.
Nature 504, 282–286 (2013).
powerful tool for characterizing the effects of different perturbations
30. Harrison, S. E., Sozen, B., Christodoulou, N., Kyprianou, C. & Zernicka-Goetz, M. Assembly
on development during the period from gastrulation to organogenesis, of embryonic and extraembryonic stem cells to mimic embryogenesis in vitro. Science
and can be combined with genetic modification, chemical screens, 356, eaal1810 (2017).
tissue manipulation and microscopy methods. This work establishes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
that the processes of gastrulation and organogenesis in a mammalian published maps and institutional affiliations.
species can be jointly recapitulated entirely and adequately in vitro. The
latter findings underscore the self-organizing properties of embryos © The Author(s), under exclusive licence to Springer Nature Limited 2021

124 | Nature | Vol 593 | 6 May 2021

Methods the adequate voltage on the pressure transmitter (component 10 in
Extended Data Fig. 1d, e), set at 5–6 V to obtain pressures of 6–7 psi in
Data reporting the gas mixing box of this specific model. The gases are injected into
No statistical methods were used to predetermine sample size. Embryos the incubator at a pressure of about 6–7 psi (which was found to be the
were randomly allocated when placed in the different growth condi- optimal level) by a pump that builds pressure. The bubble rate (which
tions. Other experiments were not randomized. The investigators were indicates the speed of gas flowing into the bottles) can be adjusted as
not blinded to allocation during experiments and outcome assessment, needed by the user by closing or opening the valve on the lid of the
as there was no relevant scientific reason to do so. water bottle (bubbler unit). Gas flows from the mix box through the
inlet into the water bottle, and the speed of gas flowing into the bottle
Animals can be controlled with a valve (Extended Data Fig. 1f). Humidified gas
Female mice (5–8-week-old ICR, C57BL/6 or BDF1) were mated with circulates to a glass test tube and then to the inside of the bottles in the
BDF1 male studs (Harlan). For experiments using transgenic reporter rotating drum; gas flow speed can be monitored by the rate of bubbles
lines mTmG (Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo) ( Jackson #007576) created inside an outlet water-filled test tube. Ideally, the flow of bub-
females were mated with either Wnt1-Cre ( Jackson #022137) or Isl1-Cre bles should be such to allow the formation of individual bubbles at a
( Jackson #024242) males. For live imaging and post-implantation graft- rate of 1–4 bubbles per second as measured in the outlet water-filled
ing experiments, Ai14(RCL-tdT)-D (B6.Cg-Gt(ROSA)26Sor(CAG-tdTomato)Hze) test tube in the right back corner of the incubator. The main compo-
( Jackson #007914) mice were crossed with Stra8-iCre mice; F1 males nents of the system are the following: O2 and CO2 controller, pressure
were mated with ICR females, and Td-Tomato+ embryos were selected. pump, vacuum pump, O2 and CO2 sensors, power supply, check valve,
For imprinting experiments, Dlk1-Dio3 IG-DMR-Snrpn-GFP ( Jackson mix gas box, pressure transmitter, limit flow, adaptor control for gases,
#030539) males were mated with ICR or C57BL/6 females. Insemination 1-μm filters, centrifugal blower (Extended Data Fig. 1). The gas control
was verified the next morning by the presence of a copulatory plug, and unit and gas mixing box established here were assembled and main-
this day was defined as E0.5 days post coitum (dpc). All animal experi- tained by Arad Technologies Ltd. Monthly testing of sensors, tubing
ments were performed according to the Animal Protection Guidelines and system parameters (pressure and gas concentration) was routinely
of Weizmann Institute of Science, and were approved by the relevant performed on site by Arad Technologies Ltd to ensure system stability
Weizmann Institute IACUC (#01390120-1, 01330120-2, 33520117-2). All and adequate performance.
mice were housed under standard 12-h light–dark cycle conditions in a
specialized and certified animal facility. Additional embryo processing Isolation of human umbilical cord blood serum and adult blood
protocols are provided in the Supplementary Methods. serum
For the purpose of this study, healthy pregnant women who were
Ex utero whole embryo roller culture and gas regulation module referred to the prenatal clinic at the Rambam Medical Center were
A step-by-step detailed protocol accompanying this manuscript is asked to give their informed consent to have blood collected from their
available on Protocol Exchange31. For cultures starting at E7.5 or later, umbilical cord, as approved by a Rambam Medical Center Helsinki com-
the embryos were kept on a rotating bottle culture unit inside a Preci- mittee (#RMB-0452-15). The source of each collection underwent full
sion Incubator system (BTC01 model with gas bubbler kit by B.T.C. anonymization and was not identified by name or other designation,
Engineering, Cullum Starr Precision Engineering Ltd) for the duration and the extracted serum was used only as described herein. Healthy
of culture. A ‘rotator’ culture method provides a continuous flow of women over the age of 18 and under 40, who gave their consent and
oxygenating gas to cultures in rotating bottles (BTC Rotating Bottle were scheduled for caesarean section delivery by their obstetrician
Culture Unit BTC02 model by B.T.C. Engineering, Cullum Starr Preci- following a prenatal clinic visit, were eligible for cord blood collection.
sion Engineering Ltd). Culture bottles (Glass Bottles (Small) BTC 03 We excluded women who gave birth vaginally as well as women with
and Glass Bottles (Large) BTC 04) are plugged into the hollow rotating any chronic illness or active medical conditions, including gestational
drum. Oxygenating gas flows along the axis and is distributed to the diabetes or hypertension. On the day of scheduled caesarean delivery
culture bottles by a baffle plate within the drum. The system maintains and in order to ensure fresh isolation and processing of serum, a team
a more stable pH than systems with sealed culture bottles. The rotator stood by for cord blood collection and serum extraction. Immediately
is supplied complete with gas filter, bubbler and leads by the manufac- upon delivery of the infant, the umbilical cord was double clamped 5–7
turer. The BTC Precision Incubator uses a thyristor-controlled heater cm from the umbilicus and transected between the clamps. Blood was
and high flow-rate fan to give a highly stable and uniform temperature collected for the purpose of this study, only after the infant had been
throughout the easily accessible working volume. The incubator has a removed from the field of surgery and umbilical blood had been drawn
working volume of 370 × 350 × 200 mm, which is accessed through the for clinical tests as needed. To avoid any traces of haemolysis, blood was
hinged Perspex top. The heater element is rated at 750 W. Bung (Hole) manually drawn by the obstetric surgeon, using a large-bore 14-gauge
BTC 06 is used to seal the bottles and Bung (Solid) BTC 07 is used to needle and a 50-ml syringe, directly from the umbilical vein while the
seal the drum (B.T.C. Engineering, Cullum Starr Precision Engineering placenta remained in situ. This was done to avoid any coagulation of
Ltd). The embryo incubator should be shielded from light by exter- blood before collection, which could lead to traumatic haemolysis,
nal covering with a dark cloth or diaper. and also to take advantage of the enhanced blood flow generated by
To achieve constant O2 and CO2 levels in the culture medium uterine contraction. Next, the collected blood was freshly and quickly
throughout the incubation period, the incubator module was linked distributed into 5-ml pro-coagulant sterile test tubes (Greiner Bio-One,
to an in-house-designed and customized gas and pressure control Z Serum Sep Clot Activator, #456005) and cooled to 4 °C for 15 min, to
unit (model HannaLab1, assembled and maintained by Arad Technolo- allow full coagulation. Then, coagulated test tubes were centrifuged
gies Ltd). Carbon dioxide and oxygen concentrations are regulated by at 2,500g for 10 min in a cooled 4 °C centrifuge. Any tube that showed
specific controllers located inside the regulation module. N2 and CO2 signs of haemolysis (such as pinkish-red serum) was discarded. The
are then injected into the gas mixer box. The mixing of the gases in the separated serum (yellowish) was collected using a pipette and filtered
gas box is homogeneous and mixed by a centrifugal blower. A pressure through a 0.22-μm filter (Nalgene, Ref # 565-0020) and then inactivated
transmitter allows control of the gas pressure between 5 and 10 psi in the in a 55 °C bath for 45 min. The inactivated serum was next distributed
gas mixing box (positive pressure over ambient external atmospheric into aliquots and placed in a freezer at −80 °C for storage for up to six
pressure), which is transmitted to the sealed embryo bottle apparatus. months. Shipping temperature was kept at −70 °C using dry ice and
The pressure generated by the pressure pump is regulated by setting any thawed serum was not refrozen. Human adult blood serum was
Article
collected from healthy adults and freshly prepared using the same 1 mM valproic acid (Sigma-Aldrich, P4543) diluted in water was added
protocol as for umbilical cord blood serum. We noted that in-house directly to the culture medium during pre-heating.
freshly prepared human umbilical cord serum and adult human serum During the process of media optimization, to reach a meaningful
gave notably superior results to commercially available ones tested conclusions regarding each serum or tissue culture supplement, at least
in this study. three different lots (batches) of reagent from the same vendor were
used. The following supplements were tested: KSR (KnockOut Serum
E7.5 embryo dissection and ex utero culture Replacement, GIBCO 10828010), HPLM (human plasma-like medium,
Mouse embryos were obtained from non-hormone primed pregnant kindly provided by J. Cantor12), N-2 Supplement (GIBCO, 17502048),
mice that were killed by cervical dislocation at E7.5. Subsequently, and B-27 Supplement (GIBCO, 17504044).
embryos were dissected out from the uterus in dissection medium
pre-equilibrated at 37 °C for 1 h, consisting of DMEM (GIBCO 11880; Ex utero culture of E5.5 and E6.5 embryos
includes 1 mg/ml d-glucose and pyruvate, without phenol red and with- Cultures starting with pre-gastrulation (E5.5) and early gastrulation
out l-glutamine) supplemented with 10% fetal bovine serum (FBS; Bio- (E6.5) embryos were done in static culture conditions until the early
logical Industries; 040131A), sterilized with a 0.22-μm filter ( JetBiofil; somite stage. Embryos were dissected out of the uterus as described
FCA-206-250). The embryos were carefully dissected from the decidua above and individual embryos were transferred into each well of an
and parietal yolk sac, leaving the intact ectoplacental cone attached 8-well glass bottom/ibiTreat μ-plate (iBidi; 80827/80826) filled with
to the egg cylinder. In brief, the decidua was isolated from the uterine 250 μl EUCM. Medium was pre-heated for an hour in an incubator with
tissue and the tip of the pear-shaped decidua was cut. The decidua 5% CO2 at 37 °C. Pre-primitive streak stage embryos (distal and anterior
was then opened into halves by introducing the forceps adjacent to visceral endoderm stage) were chosen for culture in the case of E5.5,
the embryo in parallel to its long axis and subsequently opening the and early-primitive streak stage embryos were selected for cultures
forceps. Afterwards, the embryo was grasped from the decidua and beginning at E6.5. Only embryos with no evident damage and without
the parietal yolk sac was peeled off the embryo using two forceps. Reichert’s membrane were cultured. The total volume of medium was
Embryo dissection was performed on a microscope equipped with a replaced every 24 h. Embryos were transferred into the rotating culture
Tokai Hit thermo plate at 37 °C, within a maximum of 30 min to avoid at the 4–7-somite stage (three days for cultures started at E5.5 and two
affecting the developmental potential of the embryo. Embryos in the days for cultures started at E6.5) using the same conditions described
neural plate/early head-fold stage that showed no evidence of damage previously for E8.5, with the difference that embryos were maintained
in the epiblast were selected for culture. The developmental stage of in a constant atmosphere of 21% oxygen and 5% CO2. Transfer of the
the embryos was determined as described15. EUCM consisted of 25% embryos at earlier or later somite stages resulted in failure of further
DMEM (GIBCO 11880; includes 1 mg/ml d-glucose and pyruvate, without development. Dynamic oxygen conditions yielded slightly less effi-
phenol red and without l-glutamine) supplemented with 1× glutamax ciency for expanding E6.5–E11 than using constant 21% O2 (Extended
(GIBCO, 35050061), 100 units/ml penicillin and 100 μg/ml streptomycin Data Fig. 6a, b). This difference could result from oxygen diffusion in
(Biological Industries; 030311B) and 11 mM HEPES (GIBCO, 15630056), static conditions being less efficient than in roller conditions, and that
plus 50% rat serum (rat whole embryo culture serum, ENVIGO Bioprod- might be why higher oxygen is needed to be delivered in protocols that
ucts B-4520) and 25% HCS prepared in-house. DMEM (GIBCO 11880) include static conditions. For testing static cultures using higher gas
supplemented with glutamax, penicillin/streptomycin and HEPES pressures we used the Avatar (XCell Biosciences) pressurized incubator
was stored at 4 °C in aliquots and used within 2 months. Rat serum at 2.5 and 5 psi in 21% O2. HCS could be replaced by HBS with a minor
was stored at −80 °C, heat-inactivated at 56 °C for 30 min and filtered decrease in survival efficiency. For live imaging methodology, see Sup-
through a 0.22-μm PVDF filter (Millipore; SLGV033RS) before use. HCS plementary Methods.
was collected at Rambam Medical Center in Haifa, Israel, and stored
as heat-inactivated and filtered aliquots at −80 °C. HCS was freshly Whole-mount immunostaining of E5.5–E8.5 mouse embryos
thawed and used immediately before experimentation. HCS could Embryos grown ex utero and in utero were dissected, removing the
be replaced by freshly collected HBS. HBS was collected and stored Reichert’s membrane for E6.5–E7.5 embryos, or the yolk sac and amnion
as heat-inactivated and filtered aliquots at −80 °C. Rat serum, HCS for E8.5 embryos, washed once with 1× PBS, then transferred to iBidi
and HBS could be thawed and refrozen once. Culture medium was glass bottom 8-well slides (iBidi) and fixed with 4% PFA EM grade (Elec-
pre-heated for at least one hour by placing it inside a glass bottle on the tron Microscopy Sciences, 15710) in PBS at 4 °C overnight. Embryos
rotating culture. Immediately after dissection, groups of 5–6 embryos were then washed in PBS for 5 min three times, permeabilized in PBS
were transferred into glass culture bottles (B.T.C. Engineering, Cullum with 0.5% Triton X-100/0.1 M glycine for 30 min, blocked with 10% nor-
Starr Precision Engineering Ltd) containing 2 ml EUCM. The bottles mal donkey serum/0.1% Triton X-100 in PBS for 1 h at room temperature
were placed on a rotating bottle culture system, rotating at 30 rpm at (RT), and incubated overnight at 4 °C with primary antibodies, diluted in
37 °C, and continuously gassed with an atmosphere of 5% O2, 5% CO2 at blocking solution. Next, embryos were rinsed three times for 5 min each
6.5 psi. After 24 h, groups of three embryos were moved to a new bottle in PBS/0.2% TritonX-100, incubated for 2 h at room temperature with
containing 2 ml freshly prepared medium supplemented with an extra 3 secondary antibodies diluted 1:200 in blocking solution (all secondary
mg/ml d-glucose ( J.T. Baker) (in addition to the 1 mg/ml glucose found in antibodies were from Jackson ImmunoResearch), counterstained with
the base DMEM), and a gas mixture of 13% O2, 5% CO2. At 48 h of culture, DAPI (1 μg/ml in PBS) for 10 min, and washed with PBS for 5 min three
embryos were transferred to a new bottle (two embryos per bottle) with times. If necessary, yolk sacs separated from the embryos were fixed
fresh medium supplemented with 3.5 mg/ml glucose and cultured a gas and stained following this protocol. For details of primary antibodies
atmosphere of 18% O2 and 5% CO2. After 72 h of culture, each embryo was and imaging see Supplementary Methods.
moved to an individual bottle with 1.5 ml fresh medium plus 4 mg/ml
glucose, with a gas supply of 21% O2 and 5% CO2. For medium exchange, iDISCO immunostaining of E9.5–E11.5 mouse embryos
culture medium was pre-heated for at least one hour by placing it inside Clearing of embryos from E9.5 to E11.5 was performed as described32
a glass bottle on the rotating culture with an adequate gas atmosphere with some modifications. After blocking, embryos were incubated with
depending on the stage of the cultured embryos. Embryos were imaged primary antibodies diluted in PBS/0.2%Tween-20 with 10 μg/ml heparin
each day using a Discovery V.20 stereoscope (Carl Zeiss). For pater- (PTwH)/5% DMSO/3% donkey serum at 37 °C (E9.5/E10.5, 24 h; E11.5, 48
nal imprinting experiments, littermate embryos lacking the reporter h (72 h for SOX17 and FOXA2 antibodies)). Afterwards, samples were
allele were used as negative controls. For teratogenic experiments, washed in PTwH for 24 h (15 min, 30 min, 1 h, 2 h, and overnight washes),
and incubated with adequate secondary antibodies (1:200) diluted method. The first 15 dimensions showed the majority of data variability.
in PTwH/3% donkey serum at 37 °C for 48 h. For human cell-specific Therefore, UMAP dimensional reduction was performed on the first 15
NUMA staining, donkey anti-rabbit biotin and streptavidin-Cy3 (each dimensions in all samples. The parameters for fold-change-threshold
incubated overnight) were used for signal enhancement. Next, embryos used for identification of differentially expressed genes were log(0.25)
were incubated for 30 min with DAPI (1 μg/ml) diluted in PTwH, washed and min.pct = 0.25 for embryos at both E8.5 and E10.5. For cluster anno-
in PTwH for one day (5 min, 15 min, 30 min, 1 h, 2 h, and overnight tation, we used the area under the curve (AUC) methodology to identify
washes) and dehydrated in a methanol/H2O series (1 h each), and then the enrichment of each annotated gene-set to each individual single
incubated overnight in 100% methanol. Embryos were incubated in cell. The annotations were based on published gene annotations19 for
66.6% DCM/33.3% methanol on a shaker for 3 h, followed by 100% DCM E8.5 (day 2) embryos and on the Mouse Organogenesis Cell Atlas21 for
(Sigma; 270997) for 5 min, and finally cleared and stored in benzyl E10.5 (day 4) embryos, and performed using the R package AUCELL
ether (Sigma; 108014). For details of primary antibodies and imaging 1.10.035, using parameters: aucMaxRank = 100 (5% of the total gene
see Supplementary Methods. count) under the AUCell_calcAUC function. Each cell was then anno-
tated to a single tissue based on its highest AUC score prediction. Each
Statistical analysis tissue was then cross-tabulated with each cluster to assess cluster–tis-
All statistical analyses were performed using GraphPad Prism 8 software sue overlap, and additionally normalized by z-score and ranged to 0–1
(La Joya, California). Data on graphs are shown as mean ± s.e.m. of a for plotting purposes. Next, to evaluate the probability of a certain
minimum of two independent experiments, unless otherwise stated. A cluster being enriched in a certain tissue, we used the annotated AUC
Kolmogorov–Smirnov test was performed to check normal distribution predictions of each cell to a tissue to compare to our observed cluster
of data before each statistical test. Significant differences between two annotation of each cell, thus producing a P value based on Mann–Whit-
samples were evaluated by unpaired two-sided Student’s t-test if data ney U statistics. This was calculated using the R package roc.area v1.42
were normally distributed or Mann–Whitney test for non-normally (CRAN.R-project.org). Integration of both the predicted annotation
distributed data. P < 0.05 was considered as statistically significant. overlap and its statistical enrichment to each cluster resulted in a single
predicted tissue per cluster. Differentially expressed genes (DEGs)
Single-cell RNA-seq of compatible clusters between in utero and ex utero embryos were
Ex utero cultured embryos dissected from the maternal uterus at E6.5 identified using parameters: fold-change-threshold of log(0.5) and
were sequenced at two time points (after two days and four days of with min.pct = 0.25. DEGs with significant values were also enriched
culture). In utero and ex utero developmentally matched embryos were using the gene ontology database via R package limma 3.42.236 using
dissociated using Trypsin–EDTA solution A 0.25% (Biological Industries; function ‘goana’. To assess significant changes in the proportional size
030501B) for 10 min and 15 min, respectively, at 37 °C. E8.5 embryos of each cluster between in utero and ex utero at E10.5, a t-test of the
were processed including the yolk sac but after removal of the ectopla- proportional size of each cluster was evaluated and corrected using
cental cone, and for E10.5 only the embryo proper was processed with Bonferroni correction, comparing the two groups.
the extraembryonic membranes removed. Trypsin was neutralized with
medium including 10% FBS and cells were washed and resuspended Reporting summary
in 1× PBS (calcium- and magnesium-free) with 400 μg/ml BSA. The Further information on research design is available in the Nature
cell suspension was filtered with a 100-μm cell strainer to remove cell Research Reporting Summary linked to this paper.
clumps. A cell viability percentage higher than 90% was determined
by trypan blue staining. Cells were diluted to a final concentration of
1,000 cells per μl. Each group of embryos at E8.5 (four ex utero and four Data availability
in utero) was run into two independent channels of the Chromium 10x Bulk and single-cell RNA-seq data have been deposited in the Gene
Genomics chip, the first channel containing an independent embryo Expression Omnibus (GEO) database under accession number
while the second channel consisted of three embryos pooled together. GSE149372. Source data are provided with this paper.
All E10.5 embryos (seven ex utero and five in utero) were run as inde-
pendent samples. scRNA-seq libraries were generated using the 10x 31. Aguilera-Castrejon, A. & Hanna, J. H. Highly conducive ex utero mouse embryogenesis
from pre-gastrulation to late organogenesis. Protoc. Exch. https://doi.org/10.21203/
Genomics Chromium v3 system (5,000-cell target cell recovery) and rs.3.pex-1372/v1 (2021).
sequenced on an Illumina NovaSeq 6000 platform according to the 32. Renier, N. et al. iDISCO: a simple, rapid method to immunolabel large tissue samples for
manufacturer’s instructions. volume imaging. Cell 159, 896–910 (2014).
33. Stuart, T. et al. Comprehensive integration of single-cell data. Cell 177, 1888–1902
(2019).
Single cell RNA-seq data processing 34. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell
transcriptomic data across different conditions, technologies, and species. Nat.
10x Genomics data analysis was performed using Cell Ranger 3.1.0
Biotechnol. 36, 411–420 (2018).
software (10x Genomics) for pre-processing of raw sequencing data, 35. Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering. Nat.
and Seurat 3.033,34 for downstream analysis. The mm10-3.0.0 gene Methods 14, 1083–1086 (2017).
36. Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing
set downloaded from 10x was used for gene reference requirements.
and microarray studies. Nucleic Acids Res. 43, e47 (2015).
To filter out low-expressing single cells, possible doublets produced
during the 10x sample processing, or single cells with extensive mito-
Acknowledgements This work was funded by Pascal and Ilana Mantoux; the European
chondrial expression, we filtered out cells with under 200 expressing Research Council (ERC-CoG-2016 726497-Cellnaivety); the Flight Attendant Medical
genes, over 4,000 expressing genes and over 15% or 10% mitochondrial Research Council (FAMRI); an Israel Cancer Research Fund (ICRF) professorship; BSF; the
gene expression in E8.5 (day 2) and E10.5 (day 4), respectively. Filtering Helen and Martin Kimmel Institute for Stem Cell Research; the Helen and Martin Kimmel
Award for Innovative Investigation; the Israel Science Foundation (ISF); Minerva; the
from E8.5 and E10.5 (accumulated samples of in utero and ex utero) Sherman Institute for Medicinal Chemistry; the Nella and Leon Benoziyo Center for
reduced the cell count from 16,317 to 10,707 cells and from 64,543 Neurological Diseases; the David and Fela Shapell Family Center for Genetic Disorders
to 63,481 cells, respectively. Seurat integrated analysis and anchor- Research; the Weizmann–U. Michigan program; the Kekst Family Institute for Medical
Genetics; the Dr. Beth Rom-Rymer Stem Cell Research Fund; the Edmond de Rothschild
ing of all individual samples was performed and then normalized by Foundations; the Zantker Charitable Foundation; and the Estate of Zvia Zeroni. We thank
log-normalization using a scale-factor of 10,000. The top 2,000 vari- O. Reiner and T. Sapir for help with mouse embryo electroporations; the Crown Genomics
institute of the Nancy and Stephen Grand Israel National Center for Personalized Medicine
able genes were identified by the variance stabilizing transformation
at the Weizmann Institute for support with scRNA-seq; and the Weizmann Institute
method, and subsequently scaled and centred. Principal components management and board for providing critical financial and infrastructural support.
analysis was performed for dimensional examination using the ‘elbow’ We dedicate this paper to the memories of R. Massarwa and H. Garty.
Article
Author contributions A.A.-C. designed and conducted most of the wet lab, embryology, J.H.H. conceived the idea for this project, conceptually designed the gas regulator module,
sequencing and imaging experiments, established the ex utero culture protocol and established the ex utero culture protocol, supervised execution of experiments and
co-wrote the manuscript. B.O. conducted embryo injections, performed human microglia adequate analysis of data, and wrote the manuscript.
cultures and generated human–mouse chimeras, assisted in culture condition testing and
processed cryosections. T.S. conducted bioinformatics analysis with N.N. supervising.
Competing interests J.H.H is an advisor to Biological Industries Ltd, and submitted a patent
R.M. helped to reproduce previously published protocols for ex utero culture and taught
application that covers the roller and static culture conditions described herein (filed by
immunohistochemical protocols to our team. I.M. submitted Helsinki approval, collected
J.H.H. and the Weizmann Institute of Science). R.S. is CEO of Arad Technologies Ltd. All
cord blood and calibrated human cord serum production. N.G. recruited donors and
other authors declare no competing interests.
performed cord blood extraction during caesarean sections. C.I. and S.S. assisted with
human cord serum production. S.T. generated lentiviruses and assisted in embryo
immunostaining and lentiviral infection of embryos. J.B., D.S. and S.V. performed tissue Additional information
culture and bulk RNA sequencing of mouse pluripotent stem cells. V.C., S.A. and L.L. Supplementary information The online version contains supplementary material available at
assisted with embryo immunostaining. N.L. performed characterization of cultured cells by https://doi.org/10.1038/s41586-021-03416-3.
qPCR. M.A. and H.K.-S. assisted with library preparation and single-cell RNA sequencing. Correspondence and requests for materials should be addressed to A.A.-C., I.M. or J.H.H.
Y.A. assisted with light sheet microscopy and live imaging. Y.R., S.C. and Y.S. generated Peer review information Nature thanks Hiromitsu Nakauchi, Magdalena Zernicka-Goetz and
tdTomato reporter embryos and assisted with allele imprinting experiments. M.Z. assisted the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
with embryo injections. R.S. assembled and maintained the gas-pressure regulator module. Reprints and permissions information is available at http://www.nature.com/reprints.
Extended Data Fig. 1 | Optimized gas-regulating module for roller culture generated by the pressure pump in the gas mixing box. Lph, litres per hour.
incubators. a, Diagram depicting the configuration of the gas-mixing box for b, c, Top and front views of the gas controller module. d, e, Picture displaying
gas concentration and pressure regulation. N2 and CO2 enter the gas-mixing the localization of the main components in the gas regulation module (d; listed
box and are mixed by a centrifugal blower. Gases are then injected into a water in e). f, Interior of the precision incubator system (by B.T.C. Engineering,
bottle inside the incubator by a pressure pump that allows control of the gas Cullum Starr Ltd) showing the direction of the gas flow (white arrowheads).
pressure in the gas-mixing box that is transmitted to the sealed embryo bottle g, Image of day 3 (E10.5) embryos cultured in rotating bottles (yellow
apparatus. The voltage on the pressure transmitter controls the pressure arrowheads).
Article

Extended Data Fig. 2 | Establishment and optimization of a mouse embryo point. Embryos that were dissected, fixed or moved to other conditions are
ex utero culture from late gastrulation (E7.5) until advanced subtracted from the total. Representative bright-field images of embryos
organogenesis (E11). a, E7.5 embryo dissection overview (see Methods). cultured under certain conditions are shown to the right. c, Efficiency of
b, Percentage of normally developed embryos under different gas pressures normal embryonic development evaluated in different mouse genetic
and glucose or oxygen concentrations. Blue numbers indicate the conditions backgrounds. Parental mouse lines are indicated on the left (female: male).
that yielded the highest efficiency of embryo survival. Values in parentheses Values in parentheses show the numbers of embryos evaluated. PYS, parietal
denote the number of embryos assessed per condition at every sampled time yolk sac; RS, rat serum. Scale bars, 500 μm.
Extended Data Fig. 3 | Spatio-temporal expression patterns of ectoderm- SOX9, Brachyury, CDX2 and MHC-II (myosin heavy chain-II) at the indicated
and mesoderm-related lineage markers are recapitulated in ex utero stages. Blue, DAPI. Images are representative of a minimum of three
cultured embryos. Maximum intensity projections of embryos developed biological replicates. Scale bars, 100 μm (E7.5), 200 μm (E8.5, E9.5), and
in utero and ex utero, fixed and immunostained for SOX2, OTX2, TUJ1, PAX6, 500 μm (E10.5, E11.5).
Article

Extended Data Fig. 4 | In vivo spatio-temporal expression patterns of Representative immunohistochemistry (mid-section, sagittal plane) images
endoderm-related lineage markers are recapitulated in cultured embryos. are shown for FOXA2 and GATA4 at the last time point (far-right panels). Images
Maximum intensity projections of embryos developed in utero and ex utero, represent a minimum of three biological replicates. Scale bars, 100 μm (E7.5),
fixed and immunostained for SOX17, FOXA2 and GATA4 at the indicated stages. 200 μm (E8.5, E9.5), and 500 μm (E10.5, E11.5).
Blue, DAPI. For SOX17, insets are enlargements of the dashed boxes.
Extended Data Fig. 5 | Ex utero culture of GFP-reporter transgenic confocal images of in utero E11.5 and ex utero +Day 4 transgenic mouse
embryos. a, Bright-field and GFP fluorescence images of ex utero embryos in embryos expressing GFP following activation by Wnt1-Cre and Isl1-Cre reporter
culture at the specified times expressing the GFP reporter following activation alleles. Scale bars, 1 mm. c, GFP fluorescence and bright-field images of in utero
by Wnt1-Cre and Isl1-Cre lineage-specific reporter alleles. n = 7 and 10 embryos E10.5 and ex utero +Day 3 IG-DMR–GFP reporter embryos. n = 7 in utero; n = 7 ex
for Wnt1-Cre and Isl1-Cre, respectively. Embryos dissected out of the yolk sac at utero. Scale bars, 500 μm.
+Day 4 are shown in the far-right panel. Scale bars, 500 μm. b, Representative
Article

Extended Data Fig. 6 | Devising a platform for culturing mouse embryos Embryos that were dissected, fixed or moved to other conditions are
from the onset of gastrulation until advanced organogenesis. subtracted from the total. Representative bright-field images of embryos
a–o, Schematic protocols indicating the percentages of E6.5 embryos that had cultured under certain conditions are shown to the right. Numbers in blue
developed properly per day in each condition. The medium composition, static indicate the protocol that yielded the highest efficiency of embryo survival and
or roller culture, and oxygen concentration are specified for each protocol. was subsequently used throughout the study. Scale bars, 500 μm.
Values in parentheses denote the number of embryos evaluated per condition.
Extended Data Fig. 7 | Embryos grown ex utero since early gastrulation time points. Blue, DAPI. Images are representative of a minimum of three
recapitulate the spatio-temporal expression profiles of lineage markers biological replicates. Scale bars, 50 μm (E6.5), 100 μm (+Day 1), 200 μm (+Day
seen in utero. a–c, Maximum intensity projections of embryos developed ex 2/3), 500 μm (+Day 4/5).
utero, fixed, and immunostained for eleven specific markers at the indicated
Article

Extended Data Fig. 8 | See next page for caption.

Extended Data Fig. 8 | Single-cell transcriptomic analysis of ex utero +Day 2 magnitude of enrichment. Colours indicate P value (calculated from AUC).
and +Day 4 cultured embryos compared to in utero E8.5 and E10.5 e, f, UMAP-based plots illustrating the normalized AUC assigned value of all
embryos. a, Schematic illustration of the embryo culture protocol and individual cells for each lineage at culture days +2 (e) and +4 (f). g, Correlation
sequenced time points. Early-gastrulating (E6.5) embryos grown ex utero were of gene expression of the top 2,000 most variable genes per cluster between
processed for 10x Genomics scRNA-seq after 2 or 4 days of culture. b, Violin in utero E10.5 and ex utero +Day 4 embryos. Differentially expressed genes are
plot indicating the number of unique molecular identifiers (UMIs) and genes named and shown as red dots. Clusters with the highest number of variable
obtained per condition at each time point. E8.5, median of 9,787 UMIs and genes (range of 2–8 genes only per cluster) are encased in a red box. h, Pie
2,989 genes detected per cell; E10.5, median of 4,795 UMIs and 1,789 genes charts depicting the proportional abundance of each cell cluster in both
detected per cell. c, d, Lineage annotation at culture days +2 (c) and +4 (d). Dot in utero and ex utero developed embryos at +Day 4/E10.5. Asterisks denote
plots illustrating the area under the curve (AUC) enrichment value of clusters with statistically significant differences between the two groups.
overlapping cells across clusters and tissue lineages. Circle size denotes the Cluster 7, P = 0.004; cluster 8, P = 0.009; cluster 15, P = 0.001.
Article

Extended Data Fig. 9 | Changes in morphology and size in embryos since E5.5 exhibit a mild developmental delay of about 2–4 pairs of somites
developing ex utero from pre-gastrulation to the hindlimb formation when compared to those developed in utero; however, overall morphological
stage. a, Proportional increase in size of ex utero embryos grown from the development seemed to occur correctly. f, Percentages of normal embryos in
onset of gastrulation (E6.5) to the 44-somite stage. Representative bright-field cultures started at E5.5. g, Representative increase in size of embryos cultured
images of embryos cultured for 5 days are shown at each specific stage. from E5.5 to the hindlimb stage (6 days of culture). Embryos dissected at the
Embryos without yolk sac are shown from day 3 to day 5. n ≥ 119. b, Percentages beginning and end of culture are shown. h, Immunostaining of pre-gastrulating
of normal embryos in cultures started at E6.5. c, Diagram depicting the (E5.5) embryos cultured for 6 days until the 42-somite stage. LEFTY1 and OCT4
embryonic axis measured at each stage (length of the antero-posterior axis immunostaining on a section of an E5.5 embryo (left); GATA4, MHC-II and SOX2
(A-P) for E6.5 to E8.5 and crown–rump length for later stages). d, Measurements maximum intensity projection of an embryo at culture day 6 and stained
of embryonic length at the indicated time points. Dots represent individual (right). Scale bars, 50 μm (E5.5 embryos), 500 μm (all others). n, total number of
embryos; in utero, n = 72, 25, 13, 19, 15, 38 (left to right); ex utero, n = 68, 29, 8, 19, embryos; x, number of experiments; all data represent mean ± s.e.m. Images
24; **Mann–Whitney test; ns, not significant. e, Bright-field images of E5.5 are representative of a minimum of three embryos.
embryos grown ex utero for 6 days until the 42-somite stage. Embryos cultured
Extended Data Fig. 10 | Ex utero culture medium supplemented with HBS in-house-prepared and freshly isolated adult HBS. c, Percentages of normal and
supports embryo development from early/late gastrulation until the defective embryos in cultures started at E7.5 and E6.5. n, total number of
hindlimb stage (E11). a, b, Bright-field microscopy images of mouse embryos cultured embryos; x, number of experiments. Data represent mean ± s.e.m.
grown ex utero from E7.5 (a) or E6.5 (b) with HCS replaced by Scale bars, 500 μm.
Article

Extended Data Fig. 11 | Ex utero manipulation of mouse embryonic cells or EpiL cells at E7.5, cultured ex utero for 1–4 days and stained for GFP,
development. a, b, Percentages of developmentally normal (a) and SOX2 and GATA4. Insets are enlargements of the dashed boxes. n ≥ 8 embryos.
GFP-expressing embryos (b) at 1–3 days after electroporation. c, Quantification j, Percentages of chimeric embryos (GFP+ or tdT+) after micro-injection and ex
of GFP+ cells in electroporated embryos at the indicated times. Dots represent utero culture. k, Immunostaining of +Day 1 cultured embryos injected with
individual embryos. d, e, Percentages of normally developed (d) and GFP+ EpiS cells and EpiL cells in the anterior or distal epiblast. Images represent a
embryos (e) after lentiviral transduction. Data represent mean ± s.e.m. minimum of three biological replicates. l, Representative confocal images of
f, Representative qPCR data showing the relative expression levels of mouse mouse post-implantation chimeras generated by tdT+ E7.5 in vivo epiblast
naive and primed markers in V6.5 mouse EpiS cells and formative EpiL cells, orthotopic transplantation followed by ex utero culture for 1–4 days, stained
normalized to isogenic naive 2i/Lif ES cells. n = 3. g, Overlap in the for tdTomato, GATA4 or SOX9 and SOX2 or TUJ1. n ≥ 10 embryos. m, tdT+
transcriptional signature of differentially expressed genes measured by bulk embryos explanted at E7.5 and subjected to in toto live imaging of neural tube
RNA-seq in EpiS cells and ES cells used herein, compared to previously closure at E9.0. n = 3. n, Embryos cultured ex utero since E7.5 and exposed to
published datasets26. n = 2. h, Top, generation of mouse chimeras using vehicle or 1 mM valproic acid (VPA) from E8.5 to E9.5. n = 6. Inset shows
isogenic naive ES cells. Bottom, GFP, SOX2 and GATA4 immunofluorescence magnification of the dashed box. Arrowheads, neural tube closure defects.
images of chimeric embryos generated with naive ESCs. i, Whole-mount Scale bars, 100 μm (m), 500 μm (all others). n, total number of embryos
immunostaining of GFP+ cells detected in embryos injected with mouse EpiS assessed; x, number of experiments.
Extended Data Fig. 12 | Generation of human–mouse microglia interspecies magnification of the dashed box at day 4 identifying human nuclei (hNUMA),
chimeric embryos. a, Protocol for differentiation of microglia progenitors GFP and TUJ1. n = 11 embryos (day 3); n = 8 embryos (day 4). e, Quantification of
from humans ES cells as previously described28. b, Flow cytometry dot plot to GFP+ cells detected in human–mouse microglia chimeric embryos (excluding
validate the identity of obtained microglia cells by co-expression of the GFP+ cells found in the yolk sac). Dots represent individual embryos; n = 11 and 8
microglia progenitor cell markers CD34+ and CD43+. n = 3 independent embryos for day 3 and day 4, respectively. f, Immunostaining for GFP and
experiments. c, Merged bright-field and fluorescence images of E7.5 embryos human TMEM119 in chimeric embryos. n = 3. g, Representative GFP
injected with GFP+ human microglia progenitors at day 0. d, Representative immunofluorescence of a human microglia chimeric embryonic yolk sac and
immunofluorescence images of ex utero human microglia chimeric embryos 3 yolk sac vessel with circulating human GFP+ cells. n = 3. Scale bars, 50 μm (f),
and 4 days after injection, labelled for GFP and TUJ1. The inset shows a 500 μm (all others).
 nature research | reporting summary
Corresponding author(s): Jacob Hanna, Alejandro Aguilera, Itay Maza
Last updated by author(s): 17/02/21

Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested

A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.

Software and code

Policy information about availability of computer code
Data collection Single cell sequencing data were collected using Novaseq platform. Bulk RNA seq were collected with TrueSeq Illumina. qPCR data were
obtained with the Viia7 platform (Applied Biosystems). Microscopy images were acquired with a Zeiss LSM 700 inverted confocal microscope
(Carl Zeiss). 3D images of embryos were acquired with a light-sheet microscope (Ultramicroscope II, LaVision Biotec)

Data analysis All statistical analysis besides single cell and bulk RNA-seq analysis were performed using the GraphPad Prism 8 software (La Joya, California).
Fiji/Image J (version 1.52p) was used to count cell numbers. 10X Genomics data analysis was performed with the Cell Ranger 3.1.0 software
(10x Genomics) and Seurat 3.0. The annotations were performed using the R package AUCELL 1.10.041, using parameters: aucMaxRank =100
(5% of the total gene count) under the AUCell_calcAUC function. To evaluate the probability of a certain cluster to be enriched to a certain
tissue, we utilized the annotated AUC predictions of each cell to a tissue to compare to our observed cluster annotation of each cell, thus
producing a p-value based on Mann-Whitney U statistics, this was performed using the R package roc.area v1.42 (CRAN.R-project.org).
Differential expression (DEGs) of compatible clusters between in utero and ex utero was performed using parameters : fold-change-threshold
of log(0.5) and with min.pct=0.25. DEGs with significant values were also enriched using the gene ontology database via R package limma
3.42.2 using function “goana”. For bulk RNA seq, reads were trimmed with TrimGalore 0.6.5 (flags --stringency 3 --paired) and aligned to
GRCm38 genome using STAR aligner (flags --runThreadN 64 --genomeLoad). Counts were estimated using HTSeq-count 0.7.2 (flags -q -f bam -
April 2020

r pos -s no -t exon -i gene_name). Normalization and differentially expressed genes were calculated using DESeq2 R package, with default
parameters. External gene signatures, based on mouse microarray data, were calculated from GSE60603 (PMID 25945737). Flow cytometry
was analyzed using FlowJo v10.7. Morphometric measurements were performed using the CellSens Entry 1.18 software (Olympus).
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and
reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

1
Data

nature research | reporting summary

Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability

The single cell RNA-seq and bulk RNA-seq data sets generated in this study have been deposited and publicly-available in the NCBI Gene Expression Omnibus (GEO;
http://www.ncbi.nlm.nih.gov/geo) under accession number GSE149372. Source data are provided with this paper.

Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design

All studies must disclose on these points even when the disclosure is negative.
Sample size No statistical methods were used to predetermine sample size. The number of embryos used in each experiment was determined taking into
account dat consistency, resources available and ethical reduction of animal use. For single cell RNA-Seq, the sample size was determined
when the main cell lineages at each developmental stages were captured.

Data exclusions For scRNA-seq, to filter out low expressing single cells, possible doublets produced during the 10X sample processing or single cells with
extensive mitochondrial expression, we filtered out cells with under 200 expressing genes, over 4000 expressing genes or over 10%
mitochondrial gene expression.

Replication The exact number of replicate independent experiments and embryos is indicated in the figure or in the figure legend. All data refer to
biological replicates. Single cell RNA-seq was performed in 2 biological replicates for E8.5/Day 2 and 5 &7 biological replicates for E10.5/+Day
4. All the attempts at replication were successful.

Randomization Embryos were chosen randomly when placed in different culture conditions. Other experiments were not randomized.

Blinding The investigators were not blinded to allocation during experiments and outcome assessment. We had no relevant scientific reasons to
conduct blinding.

Reporting for specific materials, systems and methods

We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Materials & experimental systems Methods

n/a Involved in the study n/a Involved in the study
Antibodies ChIP-seq
Eukaryotic cell lines Flow cytometry
Palaeontology and archaeology MRI-based neuroimaging
Animals and other organisms
Human research participants
Clinical data
Dual use research of concern
April 2020

Antibodies
Antibodies used Rabbit monoclonal anti-Brachyury (D2Z3J) (Cell Signaling Cat# 81694); Rabbit polyclonal anti-Cdx2 (Cell Signaling Cat# 3977); Mouse
monoclonal anti-Cdx2 (Biogenex Cat# MU392A-UC); Goat polyclonal anti-Gata4 (Santa Cruz Cat# SC-1237); Rabbit polyclonal anti-
Gata4 (Abcam Cat# Ab84593); Rabbit monoclonal anti-Foxa2 (Abcam Cat# Ab108422); Mouse monoclonal anti-Myosin Heavy Chain II
(clone MF-20) (R&D Cat# MAB4470); Goat polyclonal anti-Otx2 (R&D Cat# AF1979); Rabbit polyclonal anti-Pax6 (Covance; Cat#
PBR-278P); Goat polyclonal anti-Sox2 (R&D Cat# AF2018); Rabbit polyclonal anti-Sox9 (Millipore Cat# AB5535); Goat polyclonal anti-
Sox17 (R&D Cat# AF1924); Mouse monoclonal anti-TUBB3 (Tuj1) (Covance Cat# MMS-435P); Chicken polyclonal anti-GFP (Abcam

2
 Cat# Ab13970); Mouse monoclonal anti-Oct4 (clone C-10) (Santa Cruz Cat# SC-5279); Goat polyclonal anti-Lefty 1 (R&D Cat# AF746);
Goat polyclonal anti-mCherry/Tomato (SiCGEN Cat# AB0040-200); Rabbit anti-hNUMA (Abcam Cat# ab84680); Rabbit polyclonal anti-

nature research | reporting summary

human TEMEM119 (Invitrogen Cat# PA562505); Mouse monoclonal anti-human CD34-PE (BioLegend, Cat# 561), Mouse monoclonal
anti-human CD43-APC (ebioscience, Cat# 84-3C1).

Validation All the antibodies have been validated by the companies from which they were offered. Details of the validation statements,
antibody profiles and relevant citations can be found on the manufacturer's website.
Antibodies for immunofluorescence: All antibodies were validated herein by immunofluorescence on mouse embryos, based on their
well described expression patterns.
Rabbit monoclonal anti-Brachyury (Cell Signaling Cat# 81694). The antibody has been referenced in 5 publications: https://
www.cellsignal.com/products/primary-antibodies/brachyury-d2z3j-rabbit-mab/81694
Rabbit polyclonal anti-Cdx2 (Cell Signaling Cat# 3977). The antibody has been referenced in 13 publications: https://
www.cellsignal.com/products/primary-antibodies/cdx2-antibody/3977
Mouse monoclonal anti-Cdx2 (Biogenex Cat# MU392A-UC). The antibody has been referenced in more than 49 publications: https://
store.biogenex.com/us/anti-cdx-2-clone-cdx2-130.html
Goat polyclonal anti-Gata4 (Santa Cruz Cat# SC-1237). The antibody has been referenced in 176 publications: https://www.scbt.com/
p/gata-4-antibody-c-20
Rabbit polyclonal anti-Gata4 (Abcam Cat# Ab84593). The antibody has been referenced in 45 publications: https://www.abcam.com/
gata4-antibody-ab84593.html
Rabbit monoclonal anti-Foxa2 (Abcam Cat# Ab108422). The antibody has been referenced in 42 publications: https://
www.abcam.com/foxa2-antibody-epr4466-ab108422.html
Mouse monoclonal anti-Myosin Heavy Chain II (clone MF-20) (R&D Cat# MAB4470). The antibody has been referenced in 50
publications: https://www.rndsystems.com/products/myosin-heavy-chain-antibody-mf20_mab4470
Goat polyclonal anti-Otx2 (R&D Cat# AF1979). The antibody has been referenced in 37 publications: https://www.rndsystems.com/
products/human-otx2-antibody_af1979
Rabbit polyclonal anti-Pax6 (Covance; Cat# PBR-278P). The antibody has been referenced in 169 publications: https://
www.biolegend.com/en-us/products/purified-anti-pax-6-antibody-11511
Goat polyclonal anti-Sox2 (R&D Cat# AF2018). The antibody has been referenced in 93 publications: https://www.rndsystems.com/
products/human-mouse-rat-sox2-antibody_af2018
Rabbit polyclonal anti-Sox9 (Millipore Cat# AB5535). Anti-Sox9 Antibody is a well characterized affinity purified Rabbit Polyclonal
Antibody that reliably detects Transcription Factor Sox-9. This highly published antibody has been validated in IHC & WB: https://
www.merckmillipore.com/INTL/en/product/Anti-Sox9-Antibody,MM_NF-AB5535?ReferrerURL=https%3A%2F%2Fwww.google.com%
2F&bd=1
Goat polyclonal anti-Sox17 (R&D Cat# AF1924). The antibody has been referenced in 163 publications: https://
www.rndsystems.com/products/human-sox17-antibody_af1924
Mouse anti-TUBB3 (Tuj1) (Covance Cat# MMS-435P). The antibody has been referenced in 436 publications. https://
www.biolegend.com/en-us/products/purified-anti-tubulin-beta-3-tubb3-antibody-11580
Chicken polyclonal anti-GFP (Abcam Cat# Ab13970). The antibody has been referenced in 2036 publications: https://
www.abcam.com/gfp-antibody-ab13970.html
Mouse monoclonal anti-Oct4 (clone C-10) (Santa Cruz Cat# SC-5279). The antibody has been referenced in 1905 publications:
https://www.scbt.com/p/oct-3-4-antibody-c-10
Goat polyclonal anti-Lefty 1 (R&D Cat# AF746). Validated in for IHC in mouse tissues by a previous study. The antibody has been
referenced in 1 publication: https://www.rndsystems.com/products/human-mouse-lefty-antibody_af746
Goat polyclonal anti-mCherry/Tomato (SiCGEN Cat# AB0040-200). This antibody (AB0040) recognizes very well tdTomato and does
not recognize GFP (green fluorescent protein). The antibody has been referenced in 15 publications: https://www.origene.com/
catalog/antibodies/primary-antibodies/ab0040-200/mcherry
Rabbit anti-hNUMA (Abcam Cat# ab84680). The antibody has been referenced in 6 publications: https://www.abcam.com/numa-
antibody-ab84680.html
Rabbit polyclonal anti-human TEMEM119 (Invitrogen Cat# PA562505). This antibody has been validated by the manufacturer to react
with human human cerebral cortex microglia. https://www.thermofisher.com/antibody/product/TMEM119-Antibody-Polyclonal/
PA5-62505
Antibodies for immunohistochemistry:
Rabbit monoclonal anti-Foxa2 (Abcam Cat# Ab108422). Validated by the manufacturer using Formalin/PFA-fixed paraffin-embedded
sections of mouse liver tissue. The antibody has been referenced in 42 publications: https://www.abcam.com/foxa2-antibody-
epr4466-ab108422.html
Goat polyclonal anti-Gata4 (Santa Cruz Cat# SC-1237). Validated by the manufacturer for IHC in mouse tissues. The antibody has
been referenced in 176 publications: https://www.scbt.com/p/gata-4-antibody-c-20
Antibodies for flow cytometry: All the antibodies guarantee covers the use of the antibody for flow cytometry applications.
Mouse monoclonal anti-human CD34-PE (BioLegend, Cat# 561). Each lot of this antibody is quality control tested by
immunofluorescent staining with flow cytometric analysis. The antibody has been referenced in 8 publications: https://
www.biolegend.com/en-us/products/pe-anti-human-cd34-antibody-6036
Mouse monoclonal anti-human CD43-APC (ebioscience, Cat# 84-3C1). This eBio84-3C1 (84-3C1) antibody has been pre-titrated and
tested by flow cytometric analysis of human peripheral blood monocytes. The antibody has been referenced in 7 publications:
https://www.thermofisher.com/antibody/product/CD43-Antibody-clone-eBio84-3C1-84-3C1-Monoclonal/17-0439-42

Eukaryotic cell lines

April 2020

Policy information about cell lines

Cell line source(s) Mouse V6.5 and WIS2 human embryonic stem cell lines were previously reported in Gafni et al. Nature 2013 and Geula et al.
Science 2015. HEK293 was obtained from ATCC.

Authentication Karyotype and sequencing data cofirmed expected sex, karyotype, gene reporters and SNPs.

3
 Mycoplasma contamination All cell lines tested negative for mycoplasma contamination by using the MycoAlert plasma Detection Kit (Lonza, Cat#

nature research | reporting summary

LT07-318) and were routinely screened every 2 months.

Commonly misidentified lines None were used herein

(See ICLAC register)

Animals and other organisms

Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research
Laboratory animals All mice were housed in a standard 12-hour light/12-hour dark cycle conditions in a specialized and certified animal facility. 5-8 weeks
old female mice (mus musculus) of the strains ICR, C57BL/6, 129 and BDF1 were mated with male (5-40 weeks old) ICR, BDF1 or C57/
B6 mice. Established transgenic mouse strains used were: Jackson #007576, Jackson #022137, Jackson #024242, Jackson #007914,
Jackson #030539

Wild animals The study did not involve wild animals

Field-collected samples The study did not involve samples collected from the field

Ethics oversight All animal experiments were performed according to the Animal Protection Guidelines of Weizmann Institute of Science, Rehovot,
Israel. All animal experiments described herein were approved by relevant Weizmann Institute IACUC (#01390120-1, 01330120-2,
33520117-2).
Note that full information on the approval of the study protocol must also be provided in the manuscript.

Human research participants

Policy information about studies involving human research participants
Population characteristics Healthy pregnant women who were asked to give their informed consent to have blood collected from their umbilical cord,
as approved by a Rambam Medical Center Helsinki committee (#RMB-0452-15). The source of each collection underwent full
anonymization and was not identified by name or other designation, and the extracted serum was only used as described
herein. Healthy women over the age of 18 and under 40, who gave their consent and were scheduled for caesarian section
delivery by their obstetrician following a prenatal clinic visit, were eligible for cord blood collection. We excluded women who
gave vaginal birth as well as women with any chronic illness or active medical conditions, including gestational diabetes or
hypertension. All adult blood samples were collected from healthy donors.

Recruitment Research donors in the study were recruited from the prenatal clinic at the Rambam Medical Center were asked to give their
informed consent to have blood collected from their umbilical cord. Adult blood samples were collected from the authors of
the study.

Ethics oversight Rambam Medical Center Helsinki committee (#RMB-0452-15)

Note that full information on the approval of the study protocol must also be provided in the manuscript.

Flow Cytometry
Plots
Confirm that:
The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.

Methodology
Sample preparation Cells were incubated for half an hour with CD34-PE (BioLegend, 561) and CD43-APC (ebioscience, 84-3C1) antibodies (1:50)
on PBS/0.5% BSA.

Instrument BD FACS-Aria III

April 2020

Software FlowJo v10.7

Cell population abundance Only one cell population was analyzed and the purity was verified

Gating strategy FSC and SSC singlets were gated to remove debris and aggregated cells, and only single cells were considering for all analyses.
To determine the gating for positive of negative populations, an unstained control was employed, making sure that 100% of
the unstained population was allocated on the negative area of the histogram/dot plot. Gating strategies are shown in

4
 Extended Data Fig. 12b.

nature research | reporting summary

Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information.

April 2020