Trends in

OPEN ACCESS Cell Biology

Review

Human embryonic development: from

peri-implantation to gastrulation
Jinglei Zhai, 1,2,3,5 Zhenyu Xiao, 1,2,3,5 Yiming Wang, 1,2,4,5 and Hongmei Wang 1,2,3,4,*

The basic body plan of the mammalian embryo is established through gastrulation, Highlights
a pivotal early postimplantation event during which the three major germ layers In vitro culture of mammalian embryos,
(endoderm, ectoderm, and mesoderm) are specified with cellular and spatial diver- especially primate embryos, greatly en-
riches our knowledge of the dynamic
sity. Despite its basic and clinical importance, human embryo development from
characteristics of human early embryo
peri-implantation to gastrulation remains shrouded in mystery. Recent advances development.
in the elongated in vitro culture of rodent and non‐primate embryos and the con-
struction of embryo-like structures have helped to improve understanding of the Culturing non-human primate embryos
beyond the gastrulation stage is one
mechanisms of human early embryonic development. Here, we review the recent of the best strategies to understand
advances and possible future directions in the development of in vitro models to human gastrulation.
better understand human embryogenesis from peri-implantation to gastrulation.
Reconstructions of stem cell–based em-
bryo models provide insight to decipher
the mechanisms underlying human
Human embryogenesis from peri-implantation to gastrulation is critical to a embryogenesis.
healthy birth
Advancements in single-cell multiomic
Early human embryonic development from peri-implantation to gastrulation has always been a very
studies and imaging techniques con-
active yet difficult field of developmental biology (Box 1 and Figure 1A). Errors in this process may tribute to clarifying the cell lineage
lead to adverse pregnancy outcomes, including miscarriage and fetal defects. Understanding the trajectory and unveiling the underpin-
mechanisms underlying early human development is of great importance not only for basic develop- ning molecular mechanisms of human
embryogenesis and gastrulation.
mental biology but also for regenerative medicine. However, scientists have been puzzling over this
Pandora’s box because the embryos are hardly accessible. Most of our knowledge about human
embryogenesis from peri-implantation to gastrulation is derived from studies on existing anatomical
and histological collections of natural human embryos. Lessons from mouse models also enrich our
understanding of human embryonic development from a mammalian point of view (Figure 1A,B).

Recent progress in the in vitro studies of embryogenesis, including the development of elongated
in vitro culture systems for mouse and primate embryos and the construction of stem cell–based
embryo models, assisted by single-cell–based multiomic studies and imaging techniques, opens
1
The State Key Laboratory of Stem Cell
new avenues for studying human embryonic development and has significantly improved our un-
and Reproductive Biology, Institute of
derstanding of the characteristics and mechanisms of human embryos from peri-implantation to Zoology, Chinese Academy of Sciences,
gastrulation. In this context, it is necessary to provide an overview of these in vitro studies cover- Beijing 100101, P. R. China
2
Institute for Stem Cell and
ing their recent progress, limitations, and potential future directions (Figure 1C).
Regeneration, Chinese Academy of
Sciences, Beijing 100101, P. R. China
Prolonged in vitro culture of rodent and primate embryos holds the key to 3
Beijing Institute for Stem Cell and
Regenerative Medicine, Beijing 100101,
opening the Pandora’s box of human early embryogenesis
P. R. China
In vitro culture of human and non-human primate embryos reveals the features of natural human 4
University of Chinese Academy of
embryogenesis from peri-implantation to gastrulation Sciences, Beijing 100049, P. R. China
5
These authors contributed equally to
Long-term in vitro culture studies on mouse embryos (Box 2 and Table S1 in the supplemental
this work
information online) have shown that mammalian blastulas (see Glossary) have self-
organization potential beyond gastrulation and into the organogenesis stage, providing the
most valuable theoretical and technical references for the in vitro culture of human or non- *Correspondence:
human primate embryos. [email protected] (H. Wang).

18 Trends in Cell Biology, January 2022, Vol. 32, No. 1 https://doi.org/10.1016/j.tcb.2021.07.008

© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Trends in Cell Biology
OPEN ACCESS

Box 1. Major events of human early embryonic development from peri-implantation to gastrulation
Glossary
Human embryonic development starts with the formation of the totipotent zygotes during fertilization. The zygotes develop
Blastocyst: the developmental stage
into a blastula through several rounds of cleavage division, and the blastula hatches from the zona pellucida and attaches
when a mammalian embryo develops
to the maternal endometrium for embryo implantation at approximately E6–E7 (CS4). At approximately E7–E9 (CS5a),
into a blastula. For humans, the
the epiblast gradually exits from its pluripotent state and changes into a group of radially oriented cells surrounding a small
blastocyst is approximately E5–E6.
primordial cavity, the proamniotic cavity. Primitive endoderm (PE) cells grow underneath the trophectoderm (TE) to form
Blastulas: consist of cells forming an
parietal PE (parPE) and arrange beneath the ventral side of the epiblast to construct the ventral endoderm (VE). At
outer trophectoderm (TE), an inner cell
E10–E11 (CS5b), with the formation of the amniotic cavity (AC) through proamniotic cavity expansion, epiblast cells
mass (ICM), and a blastocoel (fluid-filled
on the dorsal part of the proamniotic cavity form a layer of squamous amnion epithelium [43,44], whereas the cells
cavity).
on the ventral part keep their goblet morphology and develop into late epiblasts [6]. At the same time, VE cells form
Carnegie stage (CS): criteria for the
the primary yolk sac with distal PE cells. At E12–E13 (CS5c), the epiblast expands into an asymmetric plate shape.
identification of embryonic
The VE rearranges to form the secondary yolk sac, which also consists of a plate shape on the ventral part of the epiblast.
developmental state based on the age,
Hence, this embryonic structure is called the ‘bilaminar disc’ (see Figure 1A in main text). At E14 (CS6), the human embryo
size, morphology, and somite number of
enters the gastrulation journey. Gastrulation is a milestone event in early embryonic development because it sets the foundation
the embryos. Based on the Carnegie
for three germ layers and the body plan. An embryo at the gastrulation stage with three germ layers is called a ‘gastrula.’
stages, the human embryonic period
Human embryonic gastrulation begins with the emergence of the PS at the posterior region of epiblasts based on
covering the first 8 weeks of
the epithelial–mesenchymal transition of T (brachyury)-expressed gastrulating cells (Gast) and covers CS7–CS9.
postfertilization is divided into 23 stages.
The PS defines the midline of the body and acts as a reference for the convergence of epiblast cells. Moreover, the
The staging criteria can be applied to all
emergence of Gast and PS represents the asymmetric patterning of epiblasts along the anteroposterior axis (see Figure 1A in
vertebrates.
main text). The major anatomic features of human middle and late gastrulation have been well summarized [45,46]. Altogether,
Embryo implantation: a process in
the biology of human embryo development from peri-implantation to gastrulation is far from being uncovered.
which a blastula establishes a direct
connection with the maternal
The developmental events and molecular features of mouse embryos at similar stages [E4.5–E6.5; Theiler stage (TS)
endometrium. For humans, embryo
6–8) (see Figure 1A,B in main text) have been well summarized.
implantation occurs at E7–E8 (CS4).
Germ layers: the cell layers formed
during the process of gastrulation. The
In vitro culture of human embryos followed the ethical restrictions established in 1985 in the layers consist of the endoderm,
ectoderm, and mesoderm. Each germ
Warnock report, which states that the development of human embryos in vitro should be not
layer eventually develops into certain
beyond the 14th day after fertilization or not beyond the appearance of primitive streak (PS) tissues and organs in the body. In
[1,2]. With the rapid development of techniques, scientists have pushed the culture of human humans, germ layers begin to form in
embryos [3–6]. In particular, human embryos [embryonic days 6–14 (E6–E14)] cultured in an week 3 of embryonic development. This
term should not be confused with ‘germ
extracellular matrix (ECM)-filled 3D culture system have been shown to be able to recapitulate
cells.’
the natural embryos at corresponding developmental stages, especially the formation of the PS Inner cell mass: a cluster of cells at the
anlage [6] (Figure 1A and Table S1 in the supplemental information online). Key events of inner side of a blastula adhering to one
normal human embryogenesis from peri-implantation to pregastrulation, including the specialization side of the TE. The inner cell mass will
develop into a layer of primitive
of epiblast/hypoblast lineages, emergence of the proamniotic cavity, formation of the embryonic disc endoderm and a cluster of cells called
and yolk sac, and differentiation of the trophoblast lineages, have all been well documented (Box 1 the ‘epiblast’ and eventually form the
and Figure 1A). embryo proper.
KnockOut Serum Replacement
(KSR): a defined fetal bovine serum
The in vitro culture system to support non-human primate embryos beyond gastrulation can fill (FBS)-free formulation designed to
the knowledge gap about human embryogenesis after E14. Cynomolgus macaque embryos replace FBS in feeder-dependent ESC
have been successfully cultured in vitro to E20 stage (Carnegie stage; CS8) (Table S1 in the and iPSC cultures.
Microfluidic chips: inexpensive and
supplemental information online) [7,8]. The in vitro cultured macaque embryos highly recapitu-
rapid analytical technology used to
lated the key developmental events of their in vivo counterparts, including the early development create an effective tool for
of primordial germ cells (PGCs), the establishment of the anteroposterior axis, characteriza- manipulation, monitoring, and
tion of PS cells, and even the emergence of the early neural fold. Recently, this in vitro long- assessment of cells and investigating
drug discovery, which enables the
term culture system has been successfully used to identify chimeric competency of human ex-
culture of various cells in a small
tended pluripotent stem cells (hEPSCs) [9]. In the future, this unique culture system can be amount of fluid. Thus, these chips
used to test the chimeric competency of various other stem cells, including human pluripotent have the ability to overcome the
stem cells (hPSCs) and human trophoblast stem cells (hTSCs), among others. mentioned restrictions of 2D and 3D
cell cultures, as well as animal models.
Primordial germ cells (PGCs): a
Long-term in vitro culture systems need to be optimized to satisfy the need to better understand population of cells that will form either
human embryogenesis from peri-implantation to gastrulation stages spermatozoa (sperm) progenitors in the
Although promising, the long-term in vitro culture of mouse and primate embryos still poses embryonic testis or oocyte progenitors
in the embryonic ovary.
challenges. The overall low success rate of obtaining high-fidelity in vitro cultured embryos

Trends in Cell Biology, January 2022, Vol. 32, No. 1 19

 Trends in Cell Biology
OPEN ACCESS

Theiler stage (TS): criteria for the

staging of mouse embryos based on the
age, cell number, size, somite number,
and other features of the embryos.
These criteria were first described by
Karl Theiler in The House Mouse: Atlas of
Mouse Development [42]. According to
TS, mouse embryonic development is
divided into 28 stages covering the
whole gestational period and the first
week after birth.
Trophectoderm (TE): the enveloping
layer of a blastula, by which the embryo
implants into the maternal endometrium
to establish a connection with the
maternal uterus. It will develop into a
placenta.

Trends in Cell Biology

Figure 1. Overview of human and mouse embryo development from preimplantation to late gastrulation
and a graphical abstract of this review. (A) Schematic of human and mouse embryo development from
preimplantation to late gastrulation. Abbreviations: AC, amnion cavity; Al, allantois; A-P, anteroposterior axis; AVE,
anterior visceral endoderm; DE, definitive endoderm; E, embryonic day; Ect, ectoderm; emVE, embryonic VE;
Endo, endoderm; EPI, epiblast; ExE, extraembryonic ectoderm; exVE, extraembryonic VE; HF, headfold; Meso,
mesoderm; PE, primitive endoderm; PN, primitive node; Prox.-Dist., proximal–distal axis; PS, primitive streak; TB:
trophoblast; TE, trophectoderm; VE, ventral endoderm; YS, yolk sac; *, notochordal process. Black broken lines
indicate the inner cell mass and the blue broken frame indicates the bilaminar disc of human embryo and ‘egg
cylinder’ structure of mouse embryo. (B) A correspondence of the developmental stage of human, rhesus

(Figure legend continued at the bottom of the next page.)

20 Trends in Cell Biology, January 2022, Vol. 32, No. 1

Trends in Cell Biology
OPEN ACCESS

indicates that the current conditions for extended culturing need to be optimized in terms of
reproducibility and efficiency.

Notably, high-quality embryos per se and the refined cell culture environments, including
the culture medium, supplements, gas regulation module, and integrated roller culture
system, are key steps toward establishing stable and efficient protocols for the long-term
culture of mammalian embryos. Although it has been reported that KnockOut Serum Re-
placement (KSR) allows the in vitro development of mouse and monkey embryos to the
early gastrulation stage [8,10] (Box 2), their function in nourishing human embryos
beyond the early gastrulation stage has not been confirmed due to ethical limitations.
Current in vitro culture methods for more advanced developmental embryos are still heavily
dependent on rat and human sera [7,11]. However, rat and human sera support elongated
embryonic development at the cost of high reproducibility due to considerable batch-to-
batch variations. Determining the specific functional factors in the sera that support embryonic
development during and beyond gastrulation is thus instrumental for providing more stable,
efficient, and reproducible in vitro culture systems.

Human embryo development relies highly on the simultaneous growth of extraembryonic

tissues, such as the placenta (Box 3), which provide essential signals to help extend
embryo development to certain stages. However, the interactions, especially the signal-
ing communications, between the extraembryonic tissues and the embryos per se that
drive human early embryonic development are far from being elucidated. In the current
long-term culture system, extraembryonic tissues tend to grow vigorously with the de-
velopment of the embryo, and the overwhelming growth of extraembryonic tissues
greatly hinders the development of the inner embryo by absorbing nourishment from
the culture medium. In this regard, strategies to balance extraembryonic growth and
embryonic development can be a good starting point to enable long-term embryo
culture.

It has been reported that mouse embryos can be cultured in vitro to the E11.5 stage [11], which
is comparable to the E30 stage of macaque embryos (Figure 1B; https://embryology.med.
unsw.edu.au/embryology/index.php/Animal_Development; http://www.emouseatlas.org/
emap/ema/theiler_stages/StageDefinition/stagedefinition.html). At this stage, the placenta
starts functioning as an endocrine organ to distribute nutrients to the fetal body and to
remove waste (Box 3). For culturing embryos to more advanced developmental stages,
such as beyond mouse E11.5 or macaque E30, it is indispensable to establish a functional
placenta and a functional vascularized umbilical cord to connect the placenta to the fetus.
Nutrition perfusion based on artificial umbilical cord and placenta has proved effective in
supporting extremely premature lamb infants to develop to nearly mature status [12]. In
the future, culture systems incorporated with a 3D-printed vascularized umbilical cord
and an in vitro artificial functional placenta will aid in elongating the in vitro culture of the
embryos to advanced developmental stages.

monkey, and mouse embryos based on Carnegie stages and Theiler stages, respectively. (C) A graphical abstract
showing the major content of the review. The studies on in vitro culture of rodent and primate embryos and studies on
reconstruction of stem cell–based embryo models serve as the main integrated and reliable models for elucidating
human embryo development in vitro. The in vitro systems should be optimized by considerably applying biomaterials,
microfluidic chips, dynamic culture conditions, bioengineering, and so forth. Besides, multiomics studies and imaging
techniques substantially drive the in vitro studies of human embryos forward. Finally, the in vitro studies of human
embryogenesis would be applied to gene editing, virus infection, transplantation, toxicology, and so forth within the
scope of ethics.

Trends in Cell Biology, January 2022, Vol. 32, No. 1 21

 Trends in Cell Biology
OPEN ACCESS

Box 2. In vitro culture of mouse embryos provides a great foundation for the culture of primate embryos
Mice are one of the most commonly used animal models in developmental biology studies. Following pioneering research
on the development of the limb bud stage embryos from mouse E8.5-like embryos cultured in chick plasma clots, a group
of studies on in vitro culture of mouse embryos has broadened our current understanding of mouse early embryonic
development. Many dated platforms could not efficiently support the normal embryo development for a prolonged period.
Some of them were used only in supporting the already gastrulating embryos. Although it is challenging to reproduce
the reported systems, many key factors of these platforms could be further applied in the optimization of new systems
(see Table S1 in the supplemental information online). The addition of estradiol, progesterone, pyruvic acid, and insulin-
transferrin, among others, has proved to be efficient in supporting in vitro embryonic development before/at the
pregastrulation stage, and rat serum and human cord blood serum have been demonstrated to be robust for the cultiva-
tion of gastrulating embryos. The replacement of serum from rat and human cord blood by KSR can bypass the batch-to-
batch variations in serum derived from different individuals and also can ensure a constant culture condition. However,
KSR might have limitations in supporting embryos beyond early gastrulation. To mimic the in vivo ECM-filled microenviron-
ment for the implantation and development of blastulas, biomaterials such as Matrigel and collagen have been introduced
to the in vitro culture system for providing a soft and suitable surface for trophoblast attachment and increasing the
development efficiency of cultured embryos (see Table S1 in the supplemental information online). In addition, dynamic
culture conditions such as shaker or rotator systems have been tested and proved effective in supporting in vitro cultured
embryos to develop beyond gastrulation stage (see Table S1 in the supplemental information online). All these efforts have
led to the successful culturing of mouse blastulas to the E9.5-like stage with limb bud formation and mouse E5.5 embryos
to the E11.5-like stage. The in vitro culture systems for mouse embryos have been widely used to identify various transient
features in dynamic morphogenesis during mammalian embryo development, such as cellular convergent extension or
apical constriction [10,47,48]. Despite their limited efficiency, long-term in vitro culture studies on mouse embryos have
proved in principle that mammalian blastulas own self-organization potential beyond gastrulation and into the organogen-
esis stage.

Construction techniques of stem cell–based embryo models are an emerging

crucial approach to unravel the mysteries of the early embryonic development
in humans
Various modes of stem cell–based embryo models can mimic different developmental events of
early embryos
The stem cell–based embryo model, a simplified model to mimic the definitive features of natural
embryos, is constructed using single or multiple types of embryonic or extraembryonic stem cells
(see Figure I in Box 4). The knowledge gained through the experimentation on the stem cell–
based embryo models, including in vitro culture, gene editing, screening, and so forth, will greatly

Box 3. Key developmental landscape of human primitive placentation

Human placentation begins from the TE of a blastula at approximately E5, when the preimplantation embryo is segregated
into two lineages: the ICM and TE. Following implantation, the polar TE adjacent to the ICM attaches to the uterine endo-
metrium and fuses to form a primary syncytium. The primary syncytium quickly invades the surface of the endometrium.
Fluid-filled lacunae appear within the primary syncytium at approximately E10. The enlarged lacunae gradually divide the
primary syncytium into multiple trabeculae. The trophoblasts [termed ‘cytotrophoblasts’ (CTBs)] beneath the primitive
syncytiotrophoblast (STB) then undergo rapid proliferation and penetration through the primary syncytium along the
trabeculae to form primary villi. Some CTBs penetrate through the primitive STB to form a continuous CTB shell between
the villi and decidua at approximately E14. Soon afterward, at approximately E17, the extraembryonic mesenchymal cells
invade the primary villi to form secondary villi. Fetal capillaries and Hofbauer cells (HBCs; placenta macrophages) appear
within the villi at approximately E18, marking the beginning of tertiary villus formation. Cells on the CTB shell migrate away
from the shell and invade the decidua as extravillous trophoblasts (EVTs). Based on the complex developmental progress,
the placenta villus atlas forms, which includes a continuous STB membrane; a layer of trophoblast epithelium; and the inner
side villous core with mesenchymal cells, capillaries, and HBCs [49,50]. During primitive placentation, different cell types
perform distinct functions. The CTB always functions as a stem cell–like progenitor that fuses to the STB [38,51]. The
STB directly immerses in maternal blood to transport growth-promoting substances to the embryo/fetus and acts as a
natural barrier to prevent pathogen infections. The EVT invades the maternal decidua to either anchor the placenta or
remodel maternal vessels to facilitate fetal–maternal nutrient transport [52]. The placenta mesenchymal cells are respon-
sible for maintaining the morphology of villi, supporting vascular network construction, and function as a placental stem
cell niche. Endothelial cells of placental capillaries might provide a vascular niche for the development of hematopoietic
progenitor stem cells and HBCs. The HBCs play a defensive role and support placental angiogenesis and remodeling.
Based on the formation of villi with multiple functional cell types, the placenta gradually develops and guarantees human
embryogenesis during the whole gestation [53,54].

22 Trends in Cell Biology, January 2022, Vol. 32, No. 1

Trends in Cell Biology
OPEN ACCESS

Box 4. Embryonic or extraembryonic stem cell lines derived from human and mouse embryos at different
stages of development
Mammalian embryonic or extraembryonic stem cell lines are simplified but persuasive models for deciphering the complex
cellular and molecular mechanisms underlying mammalian embryonic development. The first mESC line was successfully
established from blastocysts (Figure I). Based on many lines of evidence, including chimeric experiments, mESCs have pluripotency
similar to that of the epiblast in the blastula. The first human primed ESC line was established later. Strikingly, mouse or human ma-
ture fibroblast cells could be reprogrammed into pluripotent stem cells, iPSCs, which can differentiate into various types of stem
cells. Subsequently, naïve hESCs were generated by reprogramming iPSCs or primed hESCs, and this line of cells is validated
to retain molecular characteristics and functional properties similar to those of mESCs. The mouse expanded potential stem cells
(mEPSCs), the totipotent stem cells that potentially give rise to epiblasts, hypoblasts, and trophectoderm, were established from
eight-cell stage embryos (Figure I). However, it is worth noting that some mEPSC lines might contribute in a limited way to the
trophectoderm [55]. Spliceosomal repression in mouse ESCs can also drive a pluripotent-to-totipotent state transition.
The totipotent ESCs established by splicing inhibitors share comparable molecular levels with two- and four-cell blastomeres
[56]. Human EPSCs were obtained by induction from primed ESCs and iPSCs [57,58]. More strikingly, a precious stem cell
resource pool, including mouse epiblast-like cells (EpiLCs) or epiblast stem cells (EpiSCs), mouse and human formative plu-
ripotent stem cells (fPSCs) [59,60], and XPSCs [61] have been gradually developed over recent years (Figure I). Specifically,
both fPSCs and XPSCs can effectively differentiate into three germ layers and PGCs.

Extraembryonic stem cells, including mouse and human trophoblast stem cells and extraembryonic endoderm cells (XENs
or Ends), have also been established (Figure I). hTSCs can be derived from both blastocysts and early placental villi or from
induced human naïve PSCs [62,63], and human Ends can only be obtained via the induction of hPSCs.

Trends in Cell Biology

Figure I. Human and mouse embryonic or extraembryonic stem cell lines. The stem cell lines derived from
embryonic or extraembryonic human and mouse tissues or induced from adult tissues. The pluripotent nature of these
stem cells can indicate different developmental stages. Abbreviations: EpiLCs, epiblast-like cells; EpiSCs, epiblast stem
cells; EPSCs, expanded potential stem cells; ESCs, embryonic stem cells; fPSCs, formative pluripotent stem cells;
TSCs, trophoblast stem cells; XENs or Ends, extraembryonic endoderm stem cells; XPSCs, pluripotent stem cells
represent a pluripotency between naïve and primed states; iPSCs, induced pluripotent stem cells.

enhance the understanding of the mechanisms of human early embryonic development from
blastocyst to gastrulation because these models are more scalable, versatile, and accessible
than natural embryos (Figure 2 and Table S2 in the supplemental information online). The stem
cell–based embryo models, either the nonintegrated or the integrated ones [13], can be pro-
duced via 2D micropatterning techniques, microfluidic devices, or 3D aggregation methods
[14], and they exhibit self-organization similar to that of natural embryos during the in vitro culture
process (Figure 2 and Table S2 in the supplemental information online).

Compared with the integrated models, which consist of both embryonic and extraembryonic
tissues, the nonintegrated ones have been constructed to mimic a specific organ or system of

Trends in Cell Biology, January 2022, Vol. 32, No. 1 23

 Trends in Cell Biology
OPEN ACCESS

Trends in Cell Biology

Figure 2. Stem cell–based embryo models established to recapitulate various developmental events of human
and mouse embryos. For each model, the morphology and the corresponding embryonic stages are shown.
Abbreviations: E, embryonic day; ETX embryo, stem cell–based embryo model established by the assembly of embryonic
stem cells (ESCs); trophoblast stem cells (TSCs); and extraembryonic endoderm stem cells (XENs); PS, primitive streak.
See also Table S2 in the online supplemental information.

the embryo. The first defined mouse nonintegrated models have been established from 3D-cultured
mouse embryonic stem cells (mESCs), which display a polar distribution of WNT signaling and an
unexpected degree of self-organization, similar to the natural mouse early gastrula in the PS region
(Table S2 in the supplemental information online). The first human nonintegrated stem cell–based
embryo models have been constructed via 2D micropatterning of human embryonic stem
cells (hESCs), and they exhibit gastrulation-like characteristics in a region resembling the natural
PS (Table S2 in the supplemental information online). Neuruloids can also be established by analo-
gous methods [15]. Microfluidic chips allow gradient chemical signals and contribute to the
construction of more asymmetrically patterned stem cell–based embryo models, which provide
precious models for understanding early embryonic developmental events [16]. For example, the
human postimplantation amniotic sac has been constructed using microfluidic devices that provide

24 Trends in Cell Biology, January 2022, Vol. 32, No. 1

Trends in Cell Biology
OPEN ACCESS

asymmetric BMP4 signals along the dorsal–ventral axis [17]. This model can be used to study the
lumenogenesis of epiblasts and the differentiation of PGCs and PS cells, and it can unravel the
mechanisms underlying the formation of amnion, a signal center to guide embryonic development
from peri-implantation to gastrulation. Furthermore, with microfluidic control of the WNT-activating
gradient along the rostrocaudal axis, stem cell–based embryo models that have been generated
to mimic neural tissue from the forebrain to the hindbrain [18] function as great models to
reveal the organizational mechanisms of the rostrocaudal axis of the gut tube or neural crest.
Three-dimensional systems built with various microwells or biomaterials enable the construction of
more nonintegrated models in mice and humans that are more similar to their natural compartments,
such as gastruloids [19,20], foregut–midgut boundary organoids [21], trunk-like structures [22], and
somite-like structures [23] (Figure 2 and Table S2 in the supplemental information online). Aggregated
hESCs can be induced to form gastruloids after transient treatment with a Wnt agonist [19]. In
particular, the 3D aggregated gastruloids [19] display a spatially elongated morphology of the ante-
rior–posterior axis compared with the 2D gastruloids [24]. More interestingly, various 3D aggregated
gastruloids have been navigated to form somite-like structures [19] or even heart organoids [25],
suggesting that the tractable 3D gastruloids could be used to reveal the regulatory processes that
occur during early to late gastrulation. Microwell-based 3D aggregation also allows the construction
of more complicated embryo-like structures. For example, the model of hepato-biliary-pancreatic
anlage has been established by the aggregation of hESC-derived anterior gut and posterior gut
spheroids [21]. Considering that the natural embryos consist of both cells and ECM, biomaterials
mimicking ECM can be incorporated into the assembly of specific 3D models to provide specific
biophysical niches and deliver signals. Accordingly, 3D human postimplantation amniotic sac
or neural cyst–like structures with dorsal-ventral asymmetric patterns have been generated using
different biophysical niche factors embedded within biomaterials such as basal lamina Matrigel and
a soft gel bed of Geltrex [26] (Table S2 in the supplemental information online). The mouse
somitogenesis clock and rostrocaudal somite patterning can be obtained using Matrigel-enveloped
gastruloids [23]. Matrigel compounds also help mESCs to aggregate and develop into a highly orga-
nized trunk-like structure with neural tube- and somite-like structures.

The integrated embryo models are mainly established by microwell-based aggregation of embryonic
stem cells and extraembryonic stem cells or by the aggregation of totipotent stem cells. The aggre-
gation of mouse trophoblast stem cells (mTSCs) and mESCs generated the first mouse blastoid,
which consists of an inner cell mass–like structure, a blastocoel-like cavity, and a trophectoderm
(TE)-like outer layer, which can develop into the E6.5-like stage in vivo [27]. Human blastoids are
reassembled by aggregating a mixed population of incompletely defined stem cells (with epiblast,
trophectoderm, or hypoblast-like signatures) reprogrammed from human fibroblast cells, which
not only display morphologically comparable features as blastulas but also can attach to the tissue
culture plates and develop into E10-like embryos with the lumenogenesis of proamniotic sac–like
structure [28] (Box 1). Naïve pluripotent stem cells or hEPSCs can be also used in the construction
of blastoids because of their potent differentiation capacity [29–32] (Box 4). Even though both
blastoids stop their development at the E10-like stage, the significant events during this stage,
especially the implantation, could be sufficiently elucidated. To construct the integrated models at
the postimplantation stage, the aggregation of representative cell types from epiblast (ESCs),
trophectoderm (TSCs), and visceral endoderm [extraembryonic endoderm stem cells (XENs)] is a
potential method. Mouse ETX (mETX) embryos were generated using this strategy. Owing to the
high levels of BMP, WNT, and FGF from the artificial signal center at the boundary of the
trophectoderm and visceral endoderm, the cells at the posterior epiblast of the embryo models
differentiate into gastrulating cells, and the mETX embryos can self-organize to form a gastrulation
stage model [33–35]. This model can be used to study the crosstalk between embryonic and extra-
embryonic tissues and major embryonic development events during early gastrulation (Box 4,

Trends in Cell Biology, January 2022, Vol. 32, No. 1 25

 Trends in Cell Biology
OPEN ACCESS

Figure 2, and Table S2 in the supplemental information online). So far, there have been no human
stem cell–based embryo models established by similar methods.

The development of totipotent stem cells that have the capacity to differentiate into both embryonic
and extraembryonic cell types allows the construction of integrated blastoids from a single cell.
Mouse integrated blastoids have been constructed by the aggregation of mouse expanded poten-
tial stem cells, and they can induce decidualization of the endometrium and further implant into the
uterus, although they exhibit limited developmental capacity and disorganize structures at the
putative gastrulation stage [36]. A similar method has been applied recently to establish human
blastoids by hESPCs, iPSCs, or naïve ESCs, which are morphologically and transcriptomically
akin to natural blastulas and can develop into an E10-like stage [29,31,32] (Box 1).

Developmental capacity of current stem cell–based embryo models needs to be improved to

build more powerful platforms for understanding early embryogenesis in humans
Even though stem cell–based embryo models have been generated to mimic the natural embryos
at different developmental stages (Figure 2), establishment of an ideal embryonic model is still
pending. All the existing models are limited in terms of developmental capacity, which greatly
impedes their application.

The very limited knowledge available on the regulatory mechanisms of cell fate determination
during early human embryogenesis dramatically hinders the design of strategies to construct
stem cell–based embryo models. This huge knowledge gap can only be filled by further research
on the human natural embryos per se, the in vitro cultured non-human primate embryos, and the
embryo models.

The cells or ‘building blocks’ used to construct the embryo models determine the developmental
capacity of the models to some extent. Some types of ‘bricks’ still cannot completely model the
features of natural embryonic or extraembryonic cells. For instance, human blastoids that were
built through the aggregation of the derivatives of the fibroblast cells after reprogramming exhibited
morphological and transcriptomic similarities with natural blastocysts [28]. However, the
‘trophectoderm lineages’ in this model were reported to share more similarities with amnion epithe-
lial cells than with natural trophectoderm cells [37]. The inclusion of real functional hTSCs will help
generate blastoids or other models with a better developmental capacity. The hTSCs derived from
blastula or early placenta villi [38] are ideal ‘building blocks’ for the establishment of hETS or hETX
embryos (Box 4). At the same time, more types of stem cell lines that can serve as the origins of
different embryonic compartments should be established and used to assemble embryo models.

To construct embryo models that can almost completely mimic the natural embryos in terms of
both morphology and function, the current construction techniques need to be further optimized.
For example, the microfluid chips designed with multichannels will allow better asymmetric signal
stimulation along multibody axes [16]. The biomaterials used in the establishment of embryo
models, which contribute to feasible biophysical niches to support their self-organization, should
be optimized to be more comparable to natural microenvironment conditions. Development of
the strategies to combine all the conditions benefiting the construction of an ideal model will be
the ultimate solution to build a better stem cell–based embryo model.

Multiomics studies combined with imaging techniques dramatically propel the

study of human embryonic development
Improvements in sequencing techniques, ranging from single-cell–based single-omic to
multiomic studies and from indiscriminate RNA-sequencing to spatiotemporal transcriptome

26 Trends in Cell Biology, January 2022, Vol. 32, No. 1

Trends in Cell Biology
OPEN ACCESS

studies, have contributed to the mapping of the cell atlas and the understanding of the molecular Outstanding questions
mechanisms underlying early human embryo development [39] (Table S3 in the supplemental in- How long or to which embryonic stage
formation online). The computational embryos generated by the omic strategy not only help to can primate embryos be cultured
in vitro?
optimize the in vitro culture system to grow embryos or embryo models better but also provide
an excellent reference for the identification of in vitro cultured embryos or embryo models How can an in vitro artificial uterus be
(Figure 1C). In the future, the techniques in the spatial transcriptome will further improve and established to better support the
lead to the establishment of more complete computational embryos with higher resolution. development of mammalian embryos?

How can a human stem cell–based

Advances in microscopic imaging have tremendously contributed to the study of embryonic embryo model be constructed to the
development, both morphologically and mechanically. By combining advanced time-lapse greatest extent to mimic a natural em-
imaging captured by computable data management with intelligent software, many dynamic bryo under ethical limitations?

developmental processes of embryos have been observed. For example, via a computational Do we need to extend ethical
framework based on light-sheet microscopy and an in vitro culture system for reconstructing limitations for the in vitro culture of
long-term cell tracks, cell divisions and dynamic fate maps of tissue morphogenesis across an natural human embryos, and do we
need to set new ethical limitations for
entire fluorescent gastrulating embryo can be brightly visualized [40], and mouse embryos
stem cell–based embryo models?
developed from E9.5 to birth can be visualized vividly by using a window implanted into the uterus
of a pregnant mouse [41].

Concluding remarks
The rise of in vitro culture systems for rodent and primate embryos and the construction of
embryo-like structures have not only vastly expanded our knowledge of human early embryonic
development, especially gastrulation, but also hastened the development of potential models for
drug screening and gene editing in embryos with potential birth defects. However, many aspects
of human early embryonic development are still black boxes that will demand intensive research
efforts in the future (see Outstanding questions).

Optimizing the in vitro culture systems of natural embryos and embryo models to make the
conditions closer to the in vivo system is a recurring theme in this field. Considering that the
development of in vivo mammalian embryos relies on the maternal decidua at the early stage of
gestation and on the placenta from the early stage of gestation until term, the related biomaterial
and bioengineering techniques will help to provide a functional artificial uterus and placenta to
further support the growth of embryos. In addition, a stem cell–based embryo model is
constructed to mimic the development of natural embryos. To make embryo models that are
more like their natural counterparts or at more advanced stages of development, various strate-
gies need to be adopted using different cell types, medium supplemented with specific factors,
culture systems including microfluidic devices and 3D culturing systems, and biomaterials and
bioengineering technologies (Figure 1C).

The discussion about the in vitro studies on human embryogenesis gradually becomes heated
because the ‘14-day rule’ limits further progress in this field. Do we need to revisit this rule and
extend it to an advanced stage of embryonic development? Recently, the International Society
for Stem Cell Research (ISSCR) moved the 14-day rule from Category 3 to Category 2 (ISSCR
Guidelines for Stem Cell Research and Clinical Translation, Version 1.0, May 2021) and suggested
that studies proposing to grow human embryos beyond the formation of the PS or 14 days can
be considered on a case-by-case basis and can be subjected to several phases of review to
determine the point at which the experiments must be stopped. Due to this ethical restriction
and limited embryo resources, it is still challenging to thoroughly reveal key events in human
embryonic development by the in vitro culture of human natural embryos. Therefore, studies
based on in vitro culture of monkey embryos, which theoretically can be cultured in vitro until any
developmental stage as long as the culture conditions allow, will continue to serve as one of the

Trends in Cell Biology, January 2022, Vol. 32, No. 1 27

 Trends in Cell Biology
OPEN ACCESS

most integrated and reliable models for optimizing the culture conditions and elucidating human
embryo development. Stem cell–based embryo models are experimental models that should not
be given the same ethical considerations as natural embryos [14]. In the updated guidance,
ISSCR proposed that nonintegrated and integrated models should be considered differently
from an ethical point of view. The generation of nonintegrated stem cell–based embryos is permis-
sible after several prereviews under existing mandates. However, integrated human stem cell–
based embryo models such as blastoids should be maintained in culture for the minimum time
necessary to achieve the scientific objective. With technological advancements, the integrated
models may almost completely mimic natural embryos. When that happens, will the study of inte-
grated embryo models need to follow stricter rules? All these efforts are needed to contextualize
the development of human embryos in a physiologically meaningful manner and to apply those
principles in disease treatment. It is very urgent but discreet to amend the legislation or limitation
on in vitro studies of human embryogenesis.

Acknowledgments
We thank Dr. Qi Zhou for his invaluable comments and guidance. We apologize that some excellent work could not be cited
because of word count limitations. This work was supported by the National Key R&D Program of China (2017YFA0103800,
2020YFA0112201, 2018YFE0201100); the Strategic Priority Research Program of the Chinese Academy of Sciences
(XDA16020700); the Strategic Collaborative Research Program of the Ferring Institute of Reproductive Medicine, Ferring
Pharmaceuticals, and the Chinese Academy of Sciences (FIRMA180305); the National Natural Science Foundation of
China (NASF 82001562); and a grant from the Chinese Academy of Sciences to J.Z. and Z.X.

Declaration of interests
The authors declare no competing interests.

Supplemental information
Supplemental information associated with this article can be found online at https://doi.org/10.1016/j.tcb.2021.07.008.

References
1. Appleby, J.B. and Bredenoord, A.L. (2018) Should the 14-day 14. Nicolas, P. et al. (2021) The ethics of human-embryoids model: a
rule for embryo research become the 28-day rule? EMBO Mol. call for consistency. J. Mol. Med. (Berl). 99, 569–579
Med. 10, e9437 15. Haremaki, T. et al. (2019) Self-organizing neuruloids model de-
2. Warnock, M. (1985) Warnock – report of the committee of inquiry velopmental aspects of Huntington’s disease in the ectodermal
into human fertilisation and embryology. Ir. Nurs. News 5, 7–8 compartment. Nat. Biotechnol. 37, 1198–1208
3. Zhou, F. et al. (2019) Reconstituting the transcriptome and DNA 16. Manfrin, A. et al. (2019) Engineered signaling centers for the spa-
methylome landscapes of human implantation. Nature 572, tially controlled patterning of human pluripotent stem cells. Nat.
660–664 Methods 16, 640–648
4. West, R.C. et al. (2019) Dynamics of trophoblast differentiation in 17. Zheng, Y. et al. (2019) Controlled modelling of human epiblast and
peri-implantation-stage human embryos. Proc. Natl. Acad. Sci. amnion development using stem cells. Nature 573, 421–425
U. S. A. 116, 22635–22644 18. Rifes, P. et al. (2020) Modeling neural tube development by
5. Lv, B. et al. (2019) Single-cell RNA sequencing reveals regula- differentiation of human embryonic stem cells in a microfluidic
tory mechanism for trophoblast cell-fate divergence in human WNT gradient. Nat. Biotechnol. 38, 1265–1273
peri-implantation conceptuses. PLoS Biol. 17, e3000187 19. Moris, N. et al. (2020) An in vitro model of early anteroposterior
6. Xiang, L. et al. (2020) A developmental landscape of 3D-cultured organization during human development. Nature 582, 410–415
human pre-gastrulation embryos. Nature 577, 537–542 20. Diaz-Cuadros, M. et al. (2020) In vitro characterization of the
7. Ma, H. et al. (2019) In vitro culture of cynomolgus monkey em- human segmentation clock. Nature 580, 113–118
bryos beyond early gastrulation. Science 366, eaax7890 21. Koike, H. et al. (2019) Modelling human hepato-biliary-
8. Niu, Y. et al. (2019) Dissecting primate early post-implantation pancreatic organogenesis from the foregut-midgut boundary.
development using long-term in vitro embryo culture. Science Nature 574, 112–116
366, eaaw5754 22. Veenvliet, J.V. et al. (2020) Mouse embryonic stem cells self-or-
9. Tan, T. et al. (2021) Chimeric contribution of human extended ganize into trunk-like structures with neural tube and somites.
pluripotent stem cells to monkey embryos ex vivo. Cell 184, Science 370, eaba4937
2020–2032.e14 23. van den Brink, S.C. et al. (2020) Single-cell and spatial tran-
10. Kyprianou, C. et al. (2020) Basement membrane remodelling scriptomics reveal somitogenesis in gastruloids. Nature 582,
regulates mouse embryogenesis. Nature 582, 253–258 405–409
11. Aguilera-Castrejon, A. et al. (2021) Ex utero mouse embryogenesis 24. Martyn, I. et al. (2018) Self-organization of a human organizer by
from pre-gastrulation to late organogenesis. Nature 593, 119–124 combined Wnt and Nodal signalling. Nature 558, 132–135
12. Partridge, E.A. et al. (2017) An extra-uterine system to physiologically 25. Rossi, G. et al. (2021) Capturing cardiogenesis in gastruloids.
support the extreme premature lamb. Nat. Commun. 8, 15112 Cell Stem Cell 28, 230–240.e6
13. Rossant, J. and Tam, P.P.L. (2021) Opportunities and chal- 26. Zheng, Y. et al. (2019) Dorsal-ventral patterned neural cyst from
lenges with stem cell-based embryo models. Stem Cell Rep. human pluripotent stem cells in a neurogenic niche. Sci. Adv. 5,
16, 1031–1038 eaax5933

28 Trends in Cell Biology, January 2022, Vol. 32, No. 1

Trends in Cell Biology
OPEN ACCESS

27. Rivron, N.C. et al. (2018) Blastocyst-like structures generated 46. de Bree, K. et al. (2018) The development of the human notochord.
solely from stem cells. Nature 557, 106–111 PLoS One 13, e0205752
28. Liu, X. et al. (2021) Modelling human blastocysts by 47. Christodoulou, N. et al. (2018) Sequential formation and resolution
reprogramming fibroblasts into iBlastoids. Nature 591, 627–632 of multiple rosettes drive embryo remodelling after implantation.
29. Yanagida, A. et al. (2021) Naive stem cell blastocyst model cap- Nat. Cell Biol. 20, 1278–1289
tures human embryo lineage segregation. Cell Stem Cell 28, 48. Shahbazi, M.N. et al. (2017) Pluripotent state transitions coordi-
1016–1022.e4 nate morphogenesis in mouse and human embryos. Nature
30. Fan, Y. et al. (2021) Generation of human blastocyst-like struc- 552, 239–243
tures from pluripotent stem cells. bioRxiv Published online 49. Vento-Tormo, R. et al. (2018) Single-cell reconstruction of
March 9, 2021. https://doi.org/10.1101/2021.03.09.434313 the early maternal-fetal interface in humans. Nature 563,
31. Sozen, B. et al. (2021) Reconstructing human early embryogenesis 347–353
in vitro with pluripotent stem cells. bioRxiv Published online March 50. Liu, Y. et al. (2018) Single-cell RNA-seq reveals the diversity of
12, 2021. https://doi.org/10.1101/2021.03.12.435175 trophoblast subtypes and patterns of differentiation in the
32. Yu, L. et al. (2021) Blastocyst-like structures generated from human placenta. Cell Res. 28, 819–832
human pluripotent stem cells. Nature 591, 620–626 51. Turco, M.Y. et al. (2018) Trophoblast organoids as a model for
33. Sozen, B. et al. (2018) Self-assembly of embryonic and two maternal-fetal interactions during human placentation. Nature
extra-embryonic stem cell types into gastrulating embryo-like 564, 263–267
structures. Nat. Cell Biol. 20, 979–989 52. Chang, W.L. et al. (2018) PLAC8, a new marker for human inter-
34. Zhang, S. et al. (2019) Implantation initiation of self-assembled stitial extravillous trophoblast cells, promotes their invasion and
embryo-like structures generated using three types of mouse migration. Development 145, dev148932
blastocyst-derived stem cells. Nat. Commun. 10, 496 53. Turco, M.Y. and Moffett, A. (2019) Development of the human
35. Amadei, G. et al. (2021) Inducible stem-cell-derived embryos placenta. Development 146, dev163428
capture mouse morphogenetic events in vitro. Dev. Cell 56, 54. Hemberger, M. et al. (2020) Mechanisms of early placental
366–382.e9 development in mouse and humans. Nat. Rev. Genet. 21, 27–43
36. Li, R. et al. (2019) Generation of blastocyst-like structures from 55. Posfai, E. et al. (2021) Evaluating totipotency using criteria of in-
mouse embryonic and adult cell cultures. Cell 179, 687–702.e18 creasing stringency. Nat. Cell Biol. 23, 49–60
37. Zhao, C. et al. (2021) Reprogrammed iBlastoids contain amnion- 56. Shen, H. et al. (2021) Mouse totipotent stem cells captured and
like cells but not trophectoderm. bioRxiv Published online May 7, maintained through spliceosomal repression. Cell 184,
2021. https://doi.org/10.1101/2021.05.07.442980 2843–2859.e20
38. Okae, H. et al. (2018) Derivation of human trophoblast stem 57. Gao, X. et al. (2019) Establishment of porcine and human ex-
cells. Cell Stem Cell 22, 50–63.e6 panded potential stem cells. Nat. Cell Biol. 21, 687–699
39. Pijuan-Sala, B. et al. (2018) Single-cell transcriptional profiling: a 58. Yang, J. et al. (2017) Establishment of mouse expanded poten-
window into embryonic cell-type specification. Nat. Rev. Mol. tial stem cells. Nature 550, 393–397
Cell Biol. 19, 399–412 59. Kinoshita, M. et al. (2021) Capture of mouse and human stem
40. McDole, K. et al. (2018) In toto imaging and reconstruction of cells with features of formative pluripotency. Cell Stem Cell 28,
post-implantation mouse development at the single-cell level. 453–471.e8
Cell 175, 859–876.e33 60. Wang, X. et al. (2021) Formative pluripotent stem cells show fea-
41. Huang, Q. et al. (2020) Intravital imaging of mouse embryos. tures of epiblast cells poised for gastrulation. Cell Res. 31,
Science 368, 181–186 526–541
42. Theiler, K. (1989) The House Mouse: Atlas of Embryonic 61. Yu, L. et al. (2021) Derivation of intermediate pluripotent stem
Development, Springer-Verlag cells amenable to primordial germ cell specification. Cell Stem
43. Deglincerti, A. et al. (2016) Self-organization of the in vitro at- Cell 28, 550–567.e12
tached human embryo. Nature 533, 251–254 62. Io, S. et al. (2021) Capturing human trophoblast development
44. Shahbazi, M.N. et al. (2016) Self-organization of the human embryo in with naive pluripotent stem cells in vitro. Cell Stem Cell 28,
the absence of maternal tissues. Nat. Cell Biol. 18, 700–708 1023–1039.e13
45. Ghimire, S. et al. (2021) Human gastrulation: The embryo and its 63. Guo, G. et al. (2021) Human naive epiblast cells possess unre-
models. Dev. Biol. 474, 100–108 stricted lineage potential. Cell Stem Cell 28, 1040–1056.e6

Trends in Cell Biology, January 2022, Vol. 32, No. 1 29