Stem Cell-Derived Embryo Models: Potential Platforms for Investigating Early Development and Embryotoxicity

Abstract

Stem cell-derived embryo models (SCDEMs) create opportunities to investigate the morphological dynamics and underlying mechanisms of embryonic development, implantation, and post-implantation progression by recapitulating pre- and peri-implantation stages in vitro—an area that conventional in vivo approaches struggle to investigate. This review provides a comprehensive overview of SCDEMs, detailing the methodologies used to generate synthetic embryos and the diverse types of stem cells employed. Furthermore, we describe how closely these models recapitulate key developmental processes pre- and post-implantation, thereby establishing their value as a platform for studying early mammalian embryogenesis. In addition, we suggest that synthetic embryos are valuable tools for studying environmental toxicity, yet ethical and technical constraints limit systematic in vivo investigations. We evaluate the strengths and limitations of these models in embryotoxicity studies and highlight future research strategies. SCDEMs are expected to significantly advance the broader field of early mammalian developmental biology, with impacts extending well beyond their use in embryotoxicology.

Introduction

Stem cell-derived embryogenesis has emerged as a platform for studying early mammalian development that does not rely on oocytes or fertilized embryos (1). By combining embryonic pluripotent and extra-embryonic stem cells, researchers have developed synthetic embryo models that recapitulate key features of natural blastocysts and post-implantation embryos (2, 3). This not only circumvents the ethical and practical limitations associated with the use of human embryos but also enables precise manipulation and observation of early developmental processes in an in vitro environment. Using blastoids, stem cell-derived blastocyst-like structures, researchers can investigate peri-implantation events including trophoblast invasion, epiblast (EPI) maturation, and primitive endoderm (PrE) patterning. Recent advancements have enabled the formation of synthetic post-gastrulation embryo-like structures that model the post-implantation human conceptus, corresponding to approximately 14 days after fertilization (4). These in vitro embryogenesis models provide valuable insights into complex morphogenetic events and lineage decisions that occur during early embryonic patterning.

Initially, synthetic embryos were assembled by combining the in vitro counterparts of the major blastocyst lineages, with embryonic stem cells (ESCs) representing the EPI, trophoblast stem cells (TSCs) mimicking the trophectoderm (TE), and extra-embryonic endoderm cells (XENCs) modeling the PrE (2, 3). However, the adoption of alternative stem cell types and reprogramming strategies has enabled the generation of synthetic embryos that closely mimic natural embryonic development, with broad developmental potential. For example, the use of totipotent-like stem cells (TLSCs), induced XENCs (iXENCs), and induced TSCs (iTSCs) has broadened the developmental capacity of such models, enabling the efficient reconstruction of both embryonic and extra-embryonic compartments in vitro (5-10). In addition to their contribution to developmental biology, stem cell-derived embryo models (SCDEMs) are increasingly being used to investigate how environmental exposure can perturb defined developmental stages (11-15). In particular, they offer potential for studying the effects of endocrine-disrupting chemicals (EDCs) and other toxicants in early embryogenesis, an area that has been difficult to study systematically (16). In this review, we focus on the recent advances in the use of synthetic embryo models to investigate the fundamental aspects of early embryonic development and implantation. Furthermore, we highlight the emerging applications of synthetic embryo models in embryotoxicity testing, in which their ability to recapitulate critical developmental programs enables evaluation of the effects of environmental toxicants, pharmaceuticals, and other chemical agents on embryonic viability and molecular characteristics. We also explore options for integrating these models into future toxicological assessments.

Embryonic Development and Implantation In Vivo

Embryonic development begins with fertilization, forming a zygote that undergoes cleavage division to form a morula, and then a blastocyst, which is the final stage before implantation (Fig. 1A) (17). Fluid accumulation between blastomeres gives rise to the blastocoel, initiating lineage segregation into the TE and inner cell mass (ICM) (18). In mice, lineage segregation begins around embryonic day (E) 3.0 and is completed with blastocyst formation by E3.5. Human embryos exhibit a delayed developmental timeline, with equivalent events occurring approximately 1.5 days later; human blastocyst formation occurs around E5. Hippo signaling regulates this segregation through YAP/TAZ signaling pathway (19). In the outer cells, blastomeres containing an apical domain, YAP/TAZ are localized to the nucleus and induce Cdx2 expression through interactions with TEAD4 (Fig. 1B). In contrast, inner cells, which lack apical polarity, retain YAP/TAZ in the cytoplasm, leading to the suppression of Cdx2 expression while maintaining high Oct4 levels. As the blastocyst expands, the ICM further segregates into the EPI and PrE, regulated by Nanog and Gata6, respectively, thereby establishing three primary lineages (17). This segregation is completed by E4.5 in mice and E6.0 in humans.

Fig. 1.

Comparative developmental stages in early mouse and human embryos. (A) Preimplantation development of mouse (upper) and human (lower) embryos. (B) Hippo–YAP mediated segregation of the trophectoderm (TE) and inner cell mass in mouse blastocysts. (C) Peri-implantation development in mouse (upper) and human (lower) embryos. (D) Differential behaviors of mural and polar TE during implantation in mouse and human embryos. E: embryonic day.

After hatching from the zona pellucida, the blastocyst attaches to the uterine wall, where TE cells initiate invasion, leading to implantation by E4.5 in mice and E8∼9 in humans (20-22) (Fig. 1C). In mice, blastocyst implantation is typically initiated through the mural TE, whereas in humans, blastocysts are preferentially implanted via the polar TE (Fig. 1D). Gastrulation proceeds with primitive streak formation and the specification of the ectoderm, mesoderm, and endoderm (23). In mice, the germ layers are organized into a cylindrical egg-shaped structure, whereas in humans, they adopt a flat disc-shaped structure. From this stage onward, the embryo begins to exhibit the key features of early organogenesis. In human research, ethical guidelines typically prohibit studies beyond E14 because this point is widely regarded as the threshold for individual human life. Therefore, early-implantation stage synthetic embryos can serve as ideal platforms for studying early development and have applications in mechanistic toxicology research (24, 25).

Stem Cell Types and Stem Cell-Derived Model Systems

Studying mammalian development and the mechanisms underlying embryogenesis requires access to oocytes and fertilized embryos. However, research on these materials faces significant ethical concerns and practical limitations related to their availability. Recent advances in stem cell biology have enabled the generation of embryo-like structures that faithfully recapitulate early developmental processes in vitro. In vitro embryogenesis relies on stem cell types that either correspond to or differentiate into distinct embryonic and extra-embryonic lineages (26). ESCs derived from pre-implantation EPI contribute to all embryonic tissues, but lack the ability to form extra-embryonic lineages (27). Initially, to reconstruct embryonic compartments in vitro, TSCs, which can give rise to trophoblast lineages, and extra-embryonic endoderm (XEN) cells, which represent the PrE, were commonly incorporated into ESCs (28). By utilizing cells capable of differentiating into both embryonic and extra-embryonic lineages, blastocyst-like structures, known as blastoids, can be generated from a single cell type. For example, extended pluripotent stem cells (EPSCs) or stem cells with totipotent-like properties can differentiate into both embryonic and extra-embryonic lineages and self-organize into blastoids (5, 7). EPSCs, which are either derived from eight-cell stage embryos or converted from ESCs under specific culture conditions, can contribute to both embryonic and extra-embryonic lineages (6, 29, 30). Although blastoids are generated solely from EPSCs (5), the capacity for generating functional TE remains limited (31). Stem cells, which exhibit the transcriptional and developmental features of two-cell stage embryos, demonstrate enhanced totipotency with the ability to differentiate into both embryonic and extra-embryonic lineages (7). Depending on the induction protocol used, these stem cells are variably referred to as totipotent potential stem cells (TPSCs), TLSCs, or totipotent blastomere-like cells (TBLCs) (7, 8, 32-34). Hereafter, we refer to these cell types with totipotent-like properties as totipotent stem cells (TotiSCs). TotiSCs were initially used to develop mouse embryo models, with subsequent advances extending these approaches to human embryo-like models (35-37). TotiSCs can contribute to both embryonic and extra-embryonic lineages when introduced into blastocysts, forming chimeras. One of the most important features of TotiSCs is their ability to generate blastoids composed of EPI-, TE-, and PrE-like cells. These blastoids can be implanted into the uteri of pseudopregnant mice to initiate implantation-like responses. Therefore, TotiSC-based synthetic embryo platforms are valuable tools for studying early lineage segregation, self-organization, and morphogenetic events in vitro.

Synthetic embryos using ESCs, TSCs, and XENCs

Early efforts to generate synthetic embryos involved combining ESCs and TSCs to form ETS embryos that recapitulated blastocyst-like features but exhibited limited post-implantation development (3, 28). To address this limitation, these initial models were cultured on extracellular matrix (ECM) to compensate for the absence of visceral endoderm (VE)-secreted factors (28, 38). However, with the addition of XENCs, which represent PrE-like components, ETX embryos can develop to the E5.5∼7.0 gastrulation stage (2).

Although wild-type ESCs, TSCs, and XENCs can form synthetic embryos, iTSCs and iXENCs have been used to generate ETX embryos more efficiently. iTSCs were generated by overexpressing Cdx2 in ESCs (39), whereas iXENCs were derived by inducing Gata4 or Gata6 overexpression in ESCs (40, 41). The combination of ESCs and iXENCs gives rise to ETiX embryos, which show enhanced developmental progression to E8.5, forming neural tubes, somites, blood islands, and beating hearts (9). In addition, ESCs, iTSCs, and iXENCs can be combined to form EiTiX embryos (10, 39, 42). EiTiX models recapitulate anterior–posterior (A–P) axis formation and further develop to E8.0∼8.5, forming headfolds, hearts, chorions, and other early organ primordia (39, 42).

Synthetic embryos using EPSCs and TotiSCs

EPSC-derived blastoids generated from a single stem cell line show the partial recapitulation of pre-implantation embryo structures. The EPI lineage gives rise to a proper embryo; however, blastoids from the extra-embryonic lineages predominantly develop PrE-like structures with limited functional TE formation, which is essential for implantation and placental development (43). To address this limitation, Sozen et al. (6) incorporated TSCs into an EPSC-derived blastoid system that facilitated more robust and spatially organized TE lineage formation. By adding TSCs, they enabled the generation of blastoids with improved compartmentalization, resembling that of natural blastocysts, including the establishment of a polarized TE layer and distinct segregation between the embryonic (EPI) and extra-embryonic lineages, especially PrE lineages. Upon uterine transfer, these blastoids induced decidualization, forming decidua equivalent to those observed at E7.5 in natural embryos, although they failed to progress beyond this stage. More recently, efforts have focused on optimizing culture conditions and transcriptional cues to promote TE differentiation directly from EPSCs without the need for exogenous TSCs.

To further improve lineage fidelity and model early developmental potential, TotiSCs, which exhibit greater extra-embryonic differentiation capacity than EPSCs, were utilized. Notably, a single TotiSC can proliferate and differentiate into EPI-, TE-, and PrE-like cells, giving rise to blastoids that can be implanted into the uterus (33). These implanted blastoids induced decidualization in vivo; however, embryonic development was scarcely observed. In contrast, under in vitro culture conditions, cylindrical egg-shaped structures resembling early post-implantation embryos were formed. TPSCs, which resemble two-cell stage embryos, have been explored as an alternative source for blastoid formation, but significant limitations in advancing post-implantation development remain (8, 32, 33). TPSCs were generated by culturing ESCs, EPSCs, or two-cell embryos in CPEC medium (CD1530＋VPA＋EPZ-004777＋CHIR99021). Upon acquiring a totipotent-like state, these cells show elongated telomeres and a transcriptomic profile that mirrors that of two-cell embryos. When injected into eight-cell embryos to form chimeras, TPSCs contribute to both the embryonic and extra-embryonic lineages, including the fetal gonads (7).

TLSCs, generated by DOT1L inhibition with SGC0946 and boosted with IL-6＋AS8351, populated both the ICM and TE when injected into eight-cell embryos. They self-assemble into blastoids, which undergo efficient compaction and polarization, thus establishing discrete TE- and ICM-like compartments. After uterine transfer, these blastoid implants induce decidualization and subsequently give rise to EPI and extra-embryonic derivatives both in vivo and in vitro (32).

TBLCs were generated by treating mouse embryonic stem cells (mESCs) with the splicing inhibitor, Pladienolide B. In chimeric assays, TBLCs populate the ICM, TE, EPI, ectoplacental cone (EPC), extra-embryonic ectoderm (ExE), and germ line (44). Zhang et al. (33) reported that TBLC-derived blastoids exhibited an efficiency of approximately 80% and decreased the proportion of PrE cells, unlike EPS-blastoids which showed a reduced TE composition. Although these blastoids did not develop in vivo, they demonstrated implantation capability and induced decidualization. Luo et al. (34) used TBLC-derived blastoids, resembling natural blastocysts, to examine the effects of H₂O₂ on embryonic development and implantation. They revealed the detrimental effects of oxidative stress on embryogenesis and highlighted the feasibility of using TotiSC-derived blastoids in embryonic toxicology studies. In contrast to other TBLC-based studies, Zhang et al. (45) demonstrated that totipotency could be induced by short-term high-concentration Pladienolide B treatment. The resulting TBLC-derived blastoids exhibit blastocyst-like morphology and are capable of developing beyond the implantation stage. Chemically induced TotiSCs (ciTotiSCs) were generated in TAW medium (TTNPB, 1-azakenpaullone, and WS6). ciTotiSCs mirror two-cell embryos at the epigenomic and metabolomic levels and contribute to ICM, TE, EPC, and ExE in chimeras. However, their blastoid-forming capacity has not yet been investigated, remaining an important direction for future research (46).

Collectively, various approaches have developed to derive TotiSCs, which in turn have facilitated extensive studies using these cells to generate TotiSC-derived blastoids. The key features of these models are listed in Table 1.

Table 1.

Summary of mouse SCDEMs

Model type	Cell components	Developmental stage recapitulated	Key lineages/structures	Reference
ETS embryoid	ESC＋TSC	E6.5	-ESC–TSC self-assembly in ECM -Sequential cavitation and cavity fusion -Nodal–Wnt–BMP signaling axis -Asymmetric mesoderm/PGC induction -Symmetry breaking without AVE	(28)
ETX embryoid	ESC＋TSC＋XEN	∼E7.0	-Symmetry breaking at ES/TS boundary -Induction of mesoderm and PGC markers -EMT and migratory T＋cells -XEN layer regionalization -Partial gastrulation without mesodermal wings -Recapitulates early/mid gastrulation in correct spatiotemporal order -Gene expression resembles E6.5∼E7.5 embryos	(2)
ETX embryoid	ESC＋TSC＋XEN	E4.5∼E6.5	-Stepwise cavity formation and fusion -Patterned VE (AVE/posterior) -EPI, mesoderm, and PGC marker induction -Triggers implantation response, but lacks extra-embryonic barriers -Mimics early development; limited post-implantation progression	(81)
ETiX embryoid	ESC＋TSC＋Gata4- ESC (XEN-like)	E5.5∼E6.5	-Express EPI, ExE, VE, and AVE markers (Cer1, Lefty1, Dkk1) -Undergo symmetry breaking and A–P axis specification -Form embryonic & extra-embryonic mesoderm (Runx1＋) -Improved VE-like layer from Gata4 ESCs; no XEN needed -Limited culture beyond day 6	(40)
ETiX embryoid	ESC＋TSC＋Gata4- ESC (XEN-like)	E8.0	-Mimic E7.5∼E8.0 embryos with neural tube, tail bud, amnion, yolk sac -Undergo neurulation with three germ layers and cardiac-like beating region -Generate PGCs (Stella＋ Sox2＋ Nanog＋) near allantois -Form blood islands (Runx1＋) and extra-embryonic tissues -Incomplete somitogenesis (weak Meox1/2); gut (Foxa2＋) not detected -High transcriptomic similarity to natural embryos with minor lineage differences	(9)
ETiX embryoid	ESC＋TSC＋Gata6- ESC (XEN-like)	E7.5	-Show AVE formation, A–P axis, and gastrulation -Form mesoderm (T＋ Eomes＋) and PGCs (Blimp1＋) -Occasionally show headfold and cardiac region by day 7∼8	(41)
EiTiX embryoid	ESC＋ Cdx2- ESC (TE-like)＋Gata4- ESC (XEN-like)	E6.5∼E8.5	-Mimic A–P axis and EMT -Form PS, neuroepithelium, heart, and tail -Generate VE, EPI, ExE lineages entirely from ESCs -Develop amnion, yolk sac, and chorion-like membranes -Transcriptomes match natural embryos; PGCs not detected (↓BMP) TGCs absent	(39)
EiTiX embryoid	ESC＋ Cdx2- ESC (TE-like)＋Gata4- ESC (XEN-like)	E5.5∼E8.5	-Mimic neural tube, heart, somites, and foregut -Show A–P axis, PS, and gastrulation -Form VE, EPI, ExE, amnion, yolk sac, EPC, and blood islands -Generate PGCs and RUNX1＋ hematopoietic progenitors -Transcriptomes and morphogenesis closely match natural E8.5 embryos -Achieve organogenesis-stage structures ex utero	(42)
EPS- blastoids	EPSC＋TSC	E4.5	-Three lineages (EPI, PrE, TE); embryonic–abembryonic axis (Cdx2, Tfap2c, Krt8) -PrE-like cells match E4.5 PrE (not DE) -Develop into ETX-like cylinders (EPI, ExE, VE)	(6)
EPS- blastoids	EPSC	E4.5	-Form all three lineages: EPI (Nanog＋), PrE (Gata4＋), TE (Cdx2＋) -Exhibit rosette-like EPI polarity, lumenogenesis, and VE-derived basement membrane -Develop into egg-cylinder-like structures in vitro -Post-implantation (IVC/uterine transfer): mimic E6.5∼E7.5, but with limited growth	(5)
TPS blastoid	TPSC	E4.5	-Similar to EPI, TE, PrE in blastocyst -Implantation-competent; induce decidualization	(7)
TLSC blastoid	TLSC	E3.5∼E4.5	-Similar to EPI, TE, PrE in blastocyst -Recapitulate compaction and polarity -Implantation-competent; induce decidualization -Limited post-implantation development	(32)
TBLC blastoid	TBLC	E4.0∼E5.0	-Similar to EPI, TE, and PrE in blastocyst -Form blastoids from TBLCs or single TBLCs -Minor mislocalized Cdx2＋ cells in inner regions -PrE population lower than natural blastocysts -limited post-implantation (E6.5∼E7.5)	(33)
TBL blastoid	TBLC	E3.5∼E4.5	-Closer transcriptomic match to natural blastocysts than EPS-blastoids -Induce decidualization upon uterine transfer -Mostly proper lineage segregation	(34)
mTBLC blastoid	TBLC	E3.5∼E4.5	-Similar to EPI, TE, and PrE in blastocyst -Show polarity (E-cadherin, PAR6), active YAP -Develop egg-cylinder-like structures in vitro and in vivo (E6.5-like) -Implantation success; limited post-implantation development	(8)

A–P axis: anterior–posterior axis, DE: definitive endoderm, E: embryonic day, ECM: extracellular matrix, EMT: epithelial–mesenchymal transition, EPI: epiblast, ESC: embryonic stem cell, ExE: extra-embryonic ectoderm, IVC: in vitro culture, PGC: primordial germ cell, PrE: primitive endoderm, PS: primitive streak, SCDEMs: stem cell-derived embryo models, TBLC: totipotent blastomere-like cells, TE: trophectoderm, TSC: trophoblast stem cell, VE: visceral endoderm, XEN: extra-embryonic endoderm.

Synthetic embryos using naïve pluripotent stem cells

Pluripotent stem cells (PSCs) are classified into two distinct states: naïve and primed. Primed PSCs resemble the post-implantation EPI and exhibit limited developmental potential, being largely limited to germ-cell formation and rarely contributing to the chimera following blastocyst injection. In contrast, naïve PSCs resemble pre-implantation ICM, exhibit broader developmental potential, and are commonly used for blastoid generation. Notably, naïve human pluripotent ESCs can differentiate into both embryonic and extra-embryonic lineages, thereby expanding their applicability in human synthetic embryo models (Table 2) (47-49). Recently, a third type of PSCs, formative PSCs, has been proposed. Their differentiation potential lies between that of naïve and primed PSCs (50).

Table 2.

Summary of human SCDEMs

Model type	Cell components	Developmental stage recapitulated	Key lineages/structures	Reference
Human blastoid	Human naive PSCs	E6	-ICM (SOX2＋ATA6＋), TE (weak GATA6＋) -Similar to EPI, TE, and PrE in blastocyst -Support derivation of all blastocyst lineage stem cells -Peri-implantation-like structures (low frequency)	(35)
Human blastoid	Human naive PSCs	E6∼7	-ICM (OCT4＋), ICM/TE (TFAP2C＋) -Resemble blastocyst-stage EPI/TE/PrE -Day 4: GATA2/TEAD3 ↑CDX2 and SLC12A3 ↓in TE -EPI: KLF17＋ PrE: PDGFRα＋	(36)
Human blastoid	Human naive PSCs	E5∼7	-TE-like cells (GATA2＋ATA3＋DX2＋TROP2＋) with polarity and junctions (aPKC, CDH1) -Inner cell cluster: EPI (OCT4＋) and PrE (GATA4＋OX17＋DGFRα＋) -Three lineage states confirmed by scRNA-seq (TE, EPI, and PrE) -Interact with endometrium	(37)
Human blastoid	Human naive PSCs	E6∼7	-OCT4＋ (EPI), GATA3＋ (TE), SOX17＋ (PrE) -Post-implantation; Match E14 embryo (EPI/PrE/EVT/STB)	(57)
Human blastoid	Human naive PSCs	E5∼6	-EPI-like (SOX2＋GATA6−GATA3−), TE-like (SOX2−GATA6＋GATA3＋), PrE-like (SOX2−ATA6＋ATA3−) -Post implantation: mesoderm, definitive endoderm, hemogenic endothelium (E16∼19 stage) -Accelerated developmental transition from PS	(52)
Human iBlastoid	Human naive PSCs (using reprogrammed fibroblasts)	E5∼7	-EPI (NANOG＋, OCT4＋), TE (CDX2＋, GATA2/3＋, KRT8＋), PrE (SOX17＋, GATA6＋) -Polar/mural TE-like subtypes observed -Blastocoel-like cavity, pro-amniotic-like cavity after IVC, TE polarization (CCR7) -Polarization and partial peri-implantation morphogenesis in vitro	(51)
Human blastoid	Human naive ESCs＋TE-like cells	E6∼7	-Apical polarization, EPI maintenance (NANOG＋), TE maturation enhancement -Upregulation of polar TE markers (NR2F2, CCKBR, PTN); downregulation of mural TE markers (ALPG, CITED4, TUBB4A) -Increased EVT-like cell formation and extra-embryonic lineage development -Involves TCA–TE feedback loop via dm-αKG	(53)
Human EPS- blastoid	Human EPSCs ＋ TE-like cells	E5∼7	-EPI (OCT4＋), TE (GATA2/3＋, KRT8＋), PrE (GATA6＋) -Match to E6∼14 IVC embryos by day 8∼10	(48)
	Human EPSCs	E5∼6	-Partial induction of TE (CDX2, KRT8/18; marginal GATA3), EPI (KLF4), PrE (PDGFRA, GATA6) -Struggled to form a cohesive TE-like epithelium	(49)
Human 4CL blastoid	Human naive ESCs ＋TE-like cells ＋PrE-like cells	E6∼7	-Methylation/Imprinting status similar to blastocyst (CpG islands, promoters) -Post-implantation–like structures by day 14 (e.g., trilaminar disc, PS, mesoderm, PGC-like cells); transcriptome matched E14 human embryos	(55)
Human post-implantation embryoid	hESCs ＋PrE-like cells ＋TB-like cells	E13∼14	-EPI (SOX2, NANOG), amnion (CDX2/ISL1/VTCN1), extra-embryonic mesenchyme (GATA6/TBX20/HAND1), PrE (SOX17/GATA6), PGC-like (TFAP2C/SOX17/NANOG/BLIMP1) -Amnion cavity by day 6∼8; occasional AP symmetry; no mature PS, rare local T -GATA3-TFAP2C cells provide essential BMP	(56)
Human post-implantation embryoid	hESCs ＋PrE-like cells ＋extra-embryonic mesoderm-like cells ＋TE-like cells	E13∼14	-EPI (OCT4/SOX2), yolk sac (SOX17/GATA6), amion (ISL1/VTCN1), extra-embryonic mesoderm (FOXF1/VIM), STB (GATA3/CPM/SDC1), streak (T), PGC-like (BLIMP1) -Bilaminar disc with amnion & yolk-sac cavities; A–P axis by day 6 -TB secretes hCG	(4)

E: embryonic day, EPI: epiblast, ESC: embryonic stem cell, EPSC: extended pluripotent stem cell, EVT: extravillous trophoblast, hCG: human chorionic gonadotropin, hESCs: human embryonic stem cells, ICM: inner cell mass, IVC: in vitro culture, PGC: primordial germ cell, PrE: primitive endoderm, PS: primitive streak, PSC: pluripotent stem cell, SCDEMs: stem cell-derived embryo models, STB: syncytiotrophoblast, TB: trophoblast, TE: trophectoderm.

Human blastoids have been generated using naïve PSCs (35-37, 51-53), which can be derived by culturing cells in specific naïve-inducing media, such as 5iLA (five kinase inhibitors＋hLIF＋Activin A) (54), PXGL (PD0325901＋XAV-939＋Gö 6983＋hLIF) (36, 37), and 4CL (four chemicals＋hLIF) (55). These models recapitulate the lineage segregation observed in fertilized embryos, forming blastoids composed of the EPI, TE, and PrE lineages, with some progressing to stages resembling early gastrulation in in vitro culture systems that support peri-implantation development (4, 56). Human naïve PSCs cultured in PXGL medium were seeded onto non-adherent U-bottom plates, where they formed aggregates with expanded cystic structures by day 3 and expressed TE, PrE, and EPI markers. When cultured in vitro on Geltrex-coated dishes, these blastoids formed blastocyst outgrowths and exhibited yolk sac-like and amniotic cavity-like structure (36). Karvas et al. (57) used 5iLA naïve human PSCs to derive blastoids for post-implantation development. After 4∼5 days of culture, blastoids were formed from the naïve PSCs, and single-cell RNA sequencing (scRNA-seq) on day 7 revealed that they efficiently generated EPI, PrE, and both mural and polar TE lineages. The gene expression profiles of the day-14 blastoids clustered with those of E8∼E14 post-implantation natural embryos, although they lacked a clearly defined primitive streak.

Inadequate formation of TE-like cell components is a common challenge in human blastoid models. Therefore, many researchers have attempted to improve TE specification during blastoid formation (53, 55). For instance, Van Nerum et al. (53) suggested that metabolic rewiring plays a critical role in TE specification during blastoid formation. In their study, PXGL-induced naïve PSCs were cultured in AggreWells in the presence of dimethyl α-ketoglutarate (dm-αKG), a metabolite known to modulate chromatin-modifying enzymes. This treatment enhances TE differentiation and maturation during blastoid formation (53). Notably, dm-αKG-treated aggregates showed nuclear YAP/TAZ localization in their outer cells, corresponding to the TE, but not in their inner cells, which represent the ICM. These findings highlight the potential of dm-αKG-mediated metabolic regulation to improve TE specification, overcoming a key limitation in human blastoid generation.

Xie et al. (55) successfully generated TE-like and PrE-like cells using 4CL naïve human PSCs. The 4CL blastoids preserved genomic imprinting and closely recapitulated the DNA methylation landscape of natural human blastocysts (55). The in vitro culture of these 4CL blastoids resulted in the morphological formation of embryonic disc-like structures, trophoblastic shells, amniotic cavities, and yolk sacs. By day 14, scRNA-seq revealed the emergence of the primitive streak, extra-embryonic mesoderm, hematopoietic, and endothelial precursor cell populations, modeling post-implantation development. This study narrowed existing gaps in the understanding of human embryogenesis and enhanced the scope of future studies. Recently, Balubaid et al. (58) compared the cell-type composition of a previously reported blastoid model derived from EPSCs or naïve PSCs using single-cell transcriptomic analysis. They suggested that the lineage composition of blastoids varies depending on the type of stem cells used; EPSC-derived blastoids tend to be enriched in PrE cells, whereas naïve PSC-derived blastoids exhibit a higher proportion of EPI lineages. In particular, they identified alternative PrE subtypes that are not present in natural blastocysts, indicating that current models may include cell populations that do not fully recapitulate in vivo developmental profiles. Zhao et al. (59) also demonstrated that, although human blastoids contain the expected cell types, they exhibit notable transcriptomic differences compared to natural blastocysts. These findings highlight that the cellular composition of blastoids and the corresponding post-implantation model can vary depending on the cell type and cell line used. This suggests that the careful selection of specific stem cell types or lines may be a key strategy for more accurately recapitulating the developmental features of natural blastocysts or gastrula-stage embryos.

Stem cell-derived models of post-implantation development

In addition to embryo-like models such as blastoids and embryoids, other stem cell-based systems have been developed to mimic specific aspects of post-implantation development (39, 42). Although these systems do not fully recapitulate the entire embryo, several stem cell-derived models have been developed to mimic specific post-implantation structures (60-65). A representative example is gastruloids, which have been studied in both mouse and human systems (Table 3). Gastruloids recapitulate key post-implantation processes, including symmetry breaking, primitive streak formation, and germ layer specification. Gastruloids are typically generated by forming aggregates of PSCs in low-attachment U-bottom plates, followed by embedding in Matrigel to support three-dimensional (3D) morphogenesis that mimics gastrulation (66). The culture process generally involves the activation of the Wnt and FGF signaling pathways while concurrently inhibiting TGFβ and BMP signaling. Under these conditions, gastruloids undergo symmetry breaking, primitive streak formation, and A–P axis establishment, ultimately leading to somite formation. Gastruloids that successfully generate somites are referred to as somitoids (67). Owing to their ability to model A–P axis patterning and segmentation, these structures are also termed axioloids (68) or segmentoids (69). Consequently, they provide an in vitro platform for studying gastrulation mechanisms (61-64, 70).

Table 3.

Summary of human stem cell-derived post-implantation models

Model type	Species	Cell components	Key structures/features	Reference
Gastruloid	Mouse	mESCs	-Sequential activation of gastrulation genes (e.g., Mixl1, Eomes, Gsc, Chrd) and Hox clusters -Emergence of mesendoderm, neuroectoderm, and endoderm -Formation of node-like (Nodal＋) and gut-tube–like structures	(61)
Gastruloid	Mouse	mESCs	-A–P axis formation -Neuromesodermal progenitors; neural, mesodermal, cardiac, and head mesenchyme anterior -Functional segmentation clock	(70)
Gastruloid	Mouse	mESCs	-mESC-derived embryoids exposed to cardiogenic factors form Mesp1＋progenitors -Give rise to Flk1＋ardiovascular lineages -Establish cardiac crescent-like structures with beating regions and Ca2＋ dynamics -Display spatial proximity to gut-tube–like structures separated by endocardial-like layers.	(63)
Gastruloid	Mouse	mESCs	-Gastruloids in cardiovascular-inducing conditions express early hematopoietic markers (CD34, c-Kit, CD41) -Generate Ter-119＋erythroid-like cells -Hematopoietic cells localize anteriorly near vascular-like plexus -Model early blood development stages in vitro.	(64)
Gastruloid	Human	hESCs	-Human gastruloids derived from hESCs exhibit embryo-like A–P patterning -Posterior Wnt and anterior BMP signaling domains -Axial organization approximates E17∼21 with central somitic and posterior presomitic regions	(62)
Somitoid	Human	hPSCs	-PSM/somite markers (TBX6, HES7), neural tube markers (SOX2, SOX1, PAX6) -High Wnt: NMP, PSM, somite, late somite -Low Wnt: neural differentiation from NMPs -1 somite per HES7 cycle (∼5 hr) disrupted by NOTCH inhibition (DAPT)	(67)
Axioloid	Human	hiPSCs	-NMPs markers (TBXT, SOX2), PSM/somite markers (HES7, TBX6, MESP2), somitic mesoderm (MEOX1, UNCX, TBX18) -Overlap with human CS12 embryo cell types (NMP, PSM, somites, endothelium) -Robust epithelial somites only with RA＋Matrigel -Segmentation coupled to clock oscillations (∼5 hr per HES7 cycle, 1 somite formed per cycle)	(68)
Segmentoid	Human	hiPSCs	-NMPs markers (TBXT, SOX2), posterior PSM markers (MSGN1, TBX6, HES7), anterior PSM markers (TCF15, MESP2), somites markers (PAX3, UNCX, TBX18) -Overlap with E9.5 mouse paraxial mesoderm -HOX activation up to HOX9 (thoracic identity) -HES7 oscillations (∼4 to 5 hr) drive segmentation	(69)
Amnioid	Human	hPSCs	-Amniotic sac–like cyst with pro-amniotic cavity -Polarized amniotic ectoderm–like and epiblast–like cells -Asymmetric epithelial organization -Posterior primitive streak–like domain with EMT (SNAI1-dependent) -Asymmetric BMP–SMAD signaling modulates Amnioid stability	(60)
	Human	hESCs	-Fluid-filled amniotic sac–like structure -Inner amniotic ectoderm (ISL1＋, TFAP2C＋, GATA3＋), outer extra-embryonic mesoderm (GATA6＋, HAND1＋) -Self-organization from hESCs via BMP4＋Wnt signaling -Additional yolk sac–like cells (endoderm: SOX17＋, HNF4A＋, endothelial cells: CD34＋, PLVAP＋) -GATA3 sufficient to induce amnion -Amniotic-like fluid mirrors native amniotic composition	(65)

A–P axis: anterior–posterior axis, E: embryonic day, EMT: epithelial–mesenchymal transition, ESC: embryonic stem cell, hESCs: human embryonic stem cells, hiPSCs: human induced pluripotent stem cells, LPM: lateral plate mesoderm, mESCs: mouse embryonic stem cells, NMP: neuromesodermal progenitor, PSM: presomitic mesoderm.

Amnioids have also been established in humans (60, 65). The amnion is an extra-embryonic tissue that plays a key role in embryonic development (71). Amnioids represent an advanced model that recapitulates aspects of extra-embryonic tissue formation beyond the stages captured by gastruloids and embryoids (65). Shao et al. (60) cultured human embryonic stem cells (hESCs) in a 3D environment to generate a post-implantation amniotic sac embryoid (PASE) model. This structure recapitulated the architecture of the human amniotic sac and exhibited SNAI1-dependent epithelial-to-mesenchymal transition phenotypes. The self-organization of PASEs is driven by an intrinsic program of asymmetrical BMP–SMAD signaling pathways. This model provides a complementary in vitro platform to bridge the gap between in vivo and ex vivo studies (60).

Additionally, Gharibi et al. (65) developed post-gastrulation amnioids (PGAs) using primed hESCs to model human amniotic tissue beyond the gastrulation stage. PGAs comprise multiple human extra-embryonic cell types, including amniotic ectoderm, yolk sac endoderm, and extra-embryonic mesoderm. Unlike other SCDEMs, which are largely restricted to modeling the early gastrula, PGAs successfully recapitulate developmental progression beyond the onset of gastrulation with over 90% formation efficiency. These models were formed through self-organization without the need for exogenous differentiation stimuli, and exhibited detectable expression of proteins involved in folate biosynthesis and related metabolic pathways. These findings suggest that the amniotic sac participates in folate metabolism during early pregnancy, prior to placentation. They also demonstrated that GATA3 activity was both necessary and sufficient for amnion induction, and that BMP4 and WNT5A signals from extra-embryonic cells could promote embryonic cell differentiation, thereby providing a platform to investigate amnion–embryo interactions and conduct developmental toxicity screening (65).

Applications in Developmental Biology

Modeling early lineage segregation and signaling

Early lineage segregation is regulated by spatial signaling cues. The Hippo pathway modulates YAP localization, induces Cdx2 in the outer cells that form the TE, and contributes to extra-embryonic tissues. The inner cells retain Oct4 and give rise to ICM (72). Within the ICM, the FGF4–ERK axis directs EPI versus PrE specification; FGF4 secreted by EPI cells suppresses Nanog and induces Gata6 in adjacent cells that express FGF receptors, thereby promoting PrE gene expression (73). Wnt/β-catenin signaling also contributes; its inhibition reduces EPI specification while promoting PrE fate (74). PrE cells align adjacent to the blastocoel, forming an epithelial layer, while low-FGF cells maintain their EPI identity (Nanog^＋). FGF signaling regulates GATA6 expression through both active transcription and protein stability. This indicates its multifaceted role in PrE specification (75). During gastrulation, BMP4 and Wnt3 signaling promote mesoderm and axis formation through factors such as brachyury (T) (76, 77). Although lineage segregation is well characterized in mice, notable differences exist between species, with humans showing particularly prominent differences. For example, CDX2 expression occurs later in humans. Additionally, in human blastocysts, the expression of NANOG and GATA4 in the ICM is not affected by the inhibition of FGF, MEK, or Erk signaling. These differences suggest the presence of distinct regulatory networks during early development in primates (78, 79).

As previously noted, stem cell-derived models exhibit distinct TE, ICM, and EPI lineage specifications, effectively recapitulating the structural organization of natural embryos. Lineage assessment in stem cell-derived systems has demonstrated that TE-, PrE-, and EPI-like cells are spatially organized and exhibit lineage-appropriate functions, such as implantation-related activity or multilineage differentiation capacity (6, 35). These synthetic embryo models recapitulate the key morphogenetic events of early embryogenesis, including lumen formation and TE/EPI compartmentalization (3, 57). Therefore, SCDEMs can serve as powerful platforms for studying early lineage specification, morphogenesis, and signaling pathways involved in embryonic development (80). For instance, Kagawa et al. (37) showed that the inhibition of the Hippo pathway leads to YAP1 nuclear localization, which in turn promotes TE specification and cavitation in human blastoids. Moreover, when the YAP1–TEAD interaction or aPKC activity was disrupted, morphogenesis was impaired. These results confirm that Hippo–YAP signaling plays a critical role in early development (37). Li et al. (5) demonstrated that EPS-blastoid formation requires intact Hippo–YAP signaling. During EPS-blastoid development, YAP progressively localizes to the nuclei of outer cells. Disrupting the inhibition of the YAP–TEAD4 interaction impaired cavity formation, highlighting the essential role of Hippo–YAP signaling in TE specification (5). Additionally, Rivron et al. (3) demonstrated that embryonic cells support trophoblast proliferation and morphogenesis via BMP4/Nodal–KLF6 signaling. FGF4 and IL-11 promote trophoblast self-renewal and CDX2 expression, whereas WNT6/7B contribute to epithelial maturation, establishing an implantation-competent TE (3). Harrison et al. (28) showed that nodal/activin signaling from the ESC compartment is essential for ExE cavitation, as its inhibition impairs cavity formation in both ETS- and natural embryos. They further demonstrated that canonical Wnt signaling is required for mesoderm specification during ETS embryogenesis, as evidenced by the loss of T (Brachyury) expression upon Wnt inhibition (28). Moreover, the ETX model confirmed that nodal signaling affects cavity formation in the TSC component (81). ETX embryoids generated from Nodal^−/− ESCs exhibited developmental outcomes comparable to those treated with the TGF-β receptor inhibitor SB431542. These findings indicate that ESC-derived nodal signaling plays an important role in the development of TSC components (81). Furthermore, Karvas et al. (57) demonstrated that the inhibition of Wnt signaling induces EPI component expansion, indicating that Wnt regulates lineage specification during the post-implantation culture of human blastoid models. Guo et al. (52) observed that the number of PrE cells increased when a BRAF/MEK signaling inhibitor was removed, confirming that BRAF/MEK signaling is essential for PrE specification in human blastoids (52). Although SCDEMs cannot fully recapitulate the in vivo embryonic environment, these findings collectively suggest that key signaling pathways are functionally conserved in vitro. Such models provide a valuable platform for dissecting the molecular mechanisms underlying early development, particularly when their signaling pathways are perturbed by inhibitors or genetic modifications. They offer a promising approach for investigating how toxicants or drugs influence specific developmental stages—an area that remains challenging to study in vivo.

In vitro modeling of peri-implantation events

The peri-implantation period marks a pivotal window during which the embryo establishes physical and molecular interactions with the maternal environment (82). In mice, this stage begins at approximately E4.5, as the blastocyst hatches from the zona pellucida and expands within the uterine lumen (81, 83). The TE layer subsequently adheres to the uterine epithelium through interactions with ECM components such as β1 integrin and laminin, initiating apicobasal polarity and facilitating implantation (84). Following attachment, EPI cells undergo morphological reorganization and form pro-amniotic cavities (85). By E6.5, the embryo enters gastrulation, marked by the formation of a primitive streak (86).

In humans, comparable events occur later—around E6∼7—and are characterized by a more invasive implantation strategy accompanied by maternal decidualization (87). Shortly after implantation, the blastocyst establishes structural asymmetry as the TE differentiates into the polar and mural subtypes (88). These subpopulations exhibit species-specific functional roles; in mice, the mural TE facilitates uterine attachment (88), whereas in humans, syncytiotrophoblasts (STB) derived from the polar TE are primarily responsible for invading the endometrial lining (87). These interspecies differences emphasize the importance of developing in vitro models that accurately recapitulate the distinct peri-implantation developmental features of each species. Kagawa et al. (37) cultured human blastoids on an open-faced endometrial layer (OFEL) treated with estradiol (E2), progesterone, cAMP, and the Wnt inhibitor XAV939. The human blastoids attached to the OFEL via their polar TE, displaced the epithelial cells, and immediately secreted chorionic gonadotropin-β, demonstrating an incipient invasive response. During extended culture, in vitro cultured blastoids generated STB, extravillous trophoblasts, a pro-amniotic-like cavity, and XEN-like tissue. Despite not fully recapitulating the gross anatomy of a day-13 post-implantation human embryo, these in vitro models captured embryo–uterine interactions and the early lineage dynamics characteristic of E13 (37). However, pharmacological blockade of the IL-6/STAT3 or Hippo (MST1/2) pathways resulted in EPI-deficient blastoids that were incapable of attachment, indicating that EPI-derived signals are essential for polar TE maturation and implantation competence. Similarly, Yu et al. (89) explored how endometrial stromal cell morphology influences growth and lineage maturation. Human blastoids plated on hormone-primed immortalized endometrial stromal cells (IESCs) showed enhanced proliferation of the EPI and trophoblast lineages and reduced apoptosis. Notably, IESC-conditioned extended blastoid culture medium alone was sufficient to boost EPI and trophoblast proliferation, indicating its paracrine effects. In contrast, the formation of multinucleated STB-like cells requires direct cell–cell contact with the stromal monolayer. Similar responses were observed in natural blastocysts, underscoring the physiological relevance of this co-culture system. Collectively, these findings suggest that blastoid–IESC co-culture provides a powerful in vitro platform for dissecting peri-implantation embryo–uterine stromal interactions and the mechanisms underlying trophoblast invasion (89).

Furthermore, Karvas et al. (57) cultured human blastoids on a thick 3D ECM in a serum-free medium supplemented with E2 to mimic a receptive endometrium. Under these conditions, blastoids progressed beyond the peri-implantation E14 stage to developmental stages equivalent to E15∼E17. The study identified the polar TE as the source of the earliest invasive syncytium and demonstrated the formation of PS-like structures as well as a branched placental villous tree. Together, these findings established long-term 3D-cultured blastoids as a robust in vitro platform for investigating early human embryogenesis and placental development (57).

Recently, Shibata et al. (90) generated an apical-out endometrial organoid (AO-EMO) that integrates human endometrial stromal cells (eSC) and HUVECs in a single 3D sphere. Human blastoids attach to the AO-EMO via their polar TE, after which syncytial trophoblasts breach the epithelial barrier, invade the stroma, and directly fuse with the eSCs. Human blastocysts exhibit comparable attachment and fusion behaviors. This feto–maternal assemblage is a versatile in vitro platform for studying early embryo–uterine interactions and implantation mechanisms (90).

Environmental Toxicology as a Future Application

Environmental toxicants pose a significant threat not only to adult health but also to early embryonic development (91); however, the underlying mechanisms remain poorly understood (92, 93). A major limitation is the lack of accessible and biologically relevant model systems that can faithfully recapitulate early developmental stages, including peri-implantation events and lineage specifications (94, 95). Traditional approaches, such as mammalian whole-embryo culture (96) and aquatic embryo toxicity tests (97, 98), have offered valuable insights, but are limited by species differences (99) and ethical concerns (100). Therefore, the application of in vitro SCDEMs that closely mimic in vivo embryogenesis for embryotoxicity testing is essential for elucidating the mechanisms of developmental toxicity and providing more predictive and ethically acceptable assessments.

Early embryos as sensitive indicators of toxicant exposure

During early embryonic development, small cell populations undergoing rapid proliferation are highly vulnerable to developmental perturbations (101). Critical processes, such as cavity formation and cell polarity, can be compromised by minimal damage (102). In addition, epigenetic programming and reprogramming, which are essential for development, occur during early embryogenesis (103). Therefore, exposure to toxicants during the early embryonic stages can cause persistent developmental defects (104). Key developmental stages, including lineage segregation and body-axis patterning, are tightly regulated by signaling pathways such as the Wnt, FGF, and BMP pathways, which are susceptible to disruption by EDCs (105). Implantation relies on hormonally regulated interactions between trophoblasts and the maternal endometrium (106), and disturbances can lead to implantation failure and miscarriage (107). Therefore, assessing early embryogenesis is vital, not only as a sensitive indicator of developmental toxicity but also as a critical window for understanding the mechanisms underlying developmental failure. Given the technical and ethical challenges associated with studying human embryos in vivo, egg-free SCDEMs provide reproducible platforms for assessing toxicant effects (108).

Advantages and challenges of stem cell-derived models for toxicology

One of the major advantages of SCDEMs is their ability to recapitulate specific stages of embryonic development, allowing their flexible use in toxicological contexts. For example, EPSC- and TPSC-derived blastoids mimic the cellular architecture of the blastocyst stage and are thus suitable for assessing toxicant effects during pre-implantation development (5, 32). In contrast, ETX, ETiX, and EiTiX embryoids enable the modeling of post-implantation structures, encompassing the interactions between EPI, ExE, and VE (9, 39, 77). Therefore, these systems allow real-time monitoring of embryonic and extra-embryonic differentiation in an in vitro implantation-like environment. Accordingly, by selecting synthetic embryo models that correspond to the desired embryonic stage, targeted toxicity assessments can be performed for specific developmental windows, including post-implantation (Fig. 2).

Fig. 2.

Current SCDEMs for stage-resolved toxicant testing. The schematic highlights three experimental windows—(i) pre‑implantation, (ii) peri‑implantation, and (iii) gastrulation—and the corresponding in‑vitro models used to probe toxicant effects (blastoids; blastoid–endometrial‑organoid co‑cultures; embryoids; gastruloids; and amnioids). Red circles denote points of exposure to environmental toxicants (e.g., bisphenol A and nanoplastics). SCDEMs: stem cell-derived embryo models.

Various environmental toxicants—such as bisphenol A (BPA), perfluoroalkyl substances, and micro- and nano-plastics—have been shown to negatively affect germ cells and early embryonic development (Table 4). These compounds impair gamete function, reduce fertilization rates, and disrupt subsequent embryonic development processes, including cleavage, blastocyst formation, and implantation (109-114). EDCs are particularly concerning because they can interfere with hormone signaling pathways essential for early development, even at low concentrations, and because they are widespread in the environment (115). Although in vivo studies have provided valuable models for testing environmental toxicants, their application is limited by ethical issues and experimental complexity. Recent studies have employed stem cell-derived models to investigate the effects of environmental toxicants on early development, but toxicological testing using blastoids or ETX embryoids remains in its infancy. Zhang et al. (13) used mouse blastoids to investigate the effects of 2,4,6-triiodophenol, a halogenated disinfection byproduct, on early embryonic development; exposure induced apoptosis during blastoid formation and increased the proportion of ICM-like cells compared to that in in vitro-fertilized embryos.

Table 4.

Representative environmental toxicants and their effects on germ cells and early embryonic development

Toxicant	Model system	Reported negative effects	Reference
BPA	Mouse (in vivo, oral exposure)	-Disrupted estrous cyclicity -Dose-dependent reduction in implantation sites -Impaired decidualization; smaller implantation chambers -Downregulation of ERα and PGR target genes (e.g., Lif, Hand2)	(109)
	Early mouse embryo (in vitro)	-Zygote-stage exposure impaired first cleavage and disrupted actomyosin polarity -Early exposure reduced development to the blastocyst stage -Morula-stage exposure increased YAP1＋cells and disrupted polarity -Blastocyst-stage exposure impaired hatching	(112)
Benzo(a)pyrene	Mouse (in vivo, oral exposure)	-Decreased protein levels of endometrial receptivity and decidualization markers (Hoxa10, BMP4) -Impaired uterine function and implantation due to activation of Wnt/β-catenin signaling and reduced endometrial cell apoptosis	(110)
Nanoplastics	Mouse (in vivo, oral exposure)	-Reduced male fertility -Decreased sperm count, viability, and motility -Oxidative stress in testis -Increased apoptosis and inflammation	(113)
	Early mouse embryo (in vitro)	-Reduced blastocyst development and hatching rate -Increased cell death and developmental arrest at cleavage stages -Decreased blastocoel cavity size and overall IVC area -Elevated ROS levels and downregulation of antioxidant genes (e.g., Sod1, Gpx1, Cat)	(114)
Perfluoro-octanoic acid (a PFAS)	Human spermatozoa (in vitro)	-Reduced progressive motility -Decreased mitochondrial respiratory activity -Reduced membrane fluidity	(111)

BPA: bisphenol A, IVC: in vitro culture, PFAS: per- and polyfluoroalkyl substances, ROS: reactive oxygen species.

Post-implantation-like models, such as gastruloids, have been used to examine toxicant-induced perturbations in key developmental signaling pathways, including the Wnt/β catenin pathway (116). Compared to blastoids, gastruloid models have been extensively explored for their toxicological applications. Rebuzzini et al. (14) demonstrated that BPA exposure disrupted gastruloid elongation, impaired germ layer polarization, and interfered with Wnt/β-catenin signaling during CHIR activation. Mantziou et al. (12) established human and mouse gastruloid models to assess the developmental potential of seven teratogenic toxicants, namely retinoic acid, valproic acid, bosentan, thalidomide, phenytoin, ibuprofen, and penicillin G, which induce defective gastruloid formation, including reduced elongation and defects in axial patterning. Although human and mouse models showed some differences in their responses to toxicants, both demonstrated clear developmental impairments. Huntsman et al. (15) generated gastruloids using P19C5 mouse embryonic carcinoma cells and assessed the morphological effects of various reference chemicals. The results of these gastruloid-based assays for a broad range of drugs closely matched the previously reported in vivo results from rodent studies, supporting the potential of gastruloids as an efficient alternative to conventional rodent-based developmental toxicity testing. However, because this study employed embryonic carcinoma cells, which have inherently limited differentiation potential, the ability of the model to fully recapitulate embryonic development may be constrained. Marikawa et al. (11) developed a human ESC-derived gastruloid platform to assess compound-induced developmental toxicity and highlighted its potential as a foundational tool for future embryo toxicity screening. Recently, we (117) reported mouse ESC-derived embryo models to evaluate the embryotoxicity of BPA. We showed that BPA treatment impaired both blastoid formation and the in vitro implantation process. Notably, the BPA-induced impairment of blastoid formation was attributable to the accumulation of reactive oxygen species (ROS), which could be rescued by the antioxidant glutathione. Taken together, these results suggest that in vitro SCDEMs, including blastoids and gastruloids, represent valuable tool for the early assessment of developmental toxicity.

Broader Applications of SCDEMs

Beyond their established value in early embryogenesis and environmental toxicology studies, SCDEMs have a broad potential for diverse research applications (118). In addition to facilitating basic research on pre- and post-implantation development and organogenesis, these models can also be generated from patient-derived PSCs for disease modeling. Furthermore, their compatibility with precise genetic and epigenetic editing enables detailed mechanistic studies on disease pathogenesis and tissue development (119). Lodewijk et al. (119) showed that the CRISPRa-mediated activation of endogenous lineage regulators could drive mESCs to self-assemble into blastoids. The generation of trophoblasts via the activation of Cdx2 and PrE cells and the activation of Gata6 from pluripotent ESCs, together with ESCs, leads to the formation of embryoids resembling E5.5-stage natural embryos. The resulting structures contain the major extra-embryonic lineages: trophoblasts, PrE, and extra-embryonic mesoderm. Notably, these embryoids were obtained without exogenous transcription factors or lineage-specific signaling cues.

SCDEMs can be combined with bioengineered implantation assays and adapted for the investigation of host–pathogen interactions (120). Wu et al. (120) generated placental trophoblast organoids using human TSCs to elucidate the permissiveness of human trophoblasts and placenta to Zika virus infection. They demonstrated that human TSCs were permissive to Zika virus infection, which disrupted TSC stemness and impaired the proliferation of cytotrophoblast cells, a phenotype reminiscent of preeclampsia. In addition, trophoblast-like cells acquired increased resistance to Zika virus infection during the differentiation when exposed to Zika virus, providing new insights into how early ZIKV infection alters placental trophoblast development. Applying this approach to these models offers a practical platform to examine the impact of viral exposure on implantation in a controlled and systematic manner.

Furthermore, synthetic embryo models can be integrated with AI-driven drug screening platforms, including deep-learning approaches that surpass current high-throughput methods. Pan et al. (121) applied deep-learning approaches to identify potential therapeutic drugs for vascular malformations. They employed an in vivo system by injecting mutated vascular endothelial cells into the kidney capsule and subsequently screened for effective compounds. Although this study did not utilize 3D synthetic embryo models, it highlights the potential for integrating deep-learning-based drug discovery with advanced in vitro developmental platforms.

Although this review focuses on mouse and human models, analogous stem cell-derived embryo systems are emerging in non-human primates, pigs, and other mammals (122, 123). The ISSCR defines embryo modeling broadly as ‘assembling, differentiating, aggregating, or recombining cell populations in a manner that models or recapitulates key stages of embryonic development’ (124). This definition emphasizes that valuable insights can be gained without reconstructing a complete embryo. When conducted under rigorous ethical guidelines, continued progress in these diverse species promises to open new avenues for life science and stem cell research.

Conclusion and Perspectives

SCDEMs, or synthetic embryos, enable spatiotemporal investigation of key signaling pathways and, through the use of gene-edited reporters or knockout lines, allow direct functional validation. They can also be combined with live fluorescence imaging and high-throughput multiplex screening. This combination enables efficient toxicological testing while easing the ethical and economic burdens of in vivo studies, and could therefore reduce reliance on animal experimentation. Future challenges include the establishment of standardized protocols for generating synthetic embryos as current approaches vary across laboratories. In addition, further research is required to elucidate the interactions between the embryo and maternal uterus during implantation, which are critical for sustaining pregnancy and supporting embryonic development. Continued research is expected to elucidate stage-specific toxicant responses and their underlying molecular mechanisms during embryonic development.