Investigation of embryogenesis in a dish: the development and application of gastruloid models from pluripotent stem cells

Abstract

Gastrulation constitutes a critical phase of mammalian embryogenesis, during which the embryo undergoes morphological changes, germ layer determination, and body axes establishment, laying the foundation for subsequent embryonic development. However, due to the technical and ethical restrictions, embryo-based studies during the gastrulation period are limited in both the scope of scientific questions worth exploring and experimental tractability. Gastruloids, three-dimensional aggregates recapitulating embryos at the gastrulation stage, emerge as an in vitro platform to dissect the dynamics of this process. Owing to their high-throughput and tractable advantages, gastruloids hold broad application potential. In this review, we summarize current knowledge on mammalian embryogenesis and the insights from both mouse and human gastruloid models. We also discuss the current applications of gastruloids in embryogenesis, disease modeling, drug development, and regenerative medicine.

Introduction

Gastrulation, the cornerstone of mammalian embryogenesis, converts a simple blastula into a trilaminar gastrula through the establishment of the three primary germ layers (ectoderm, mesoderm, and endoderm) and the emergence of three orthogonal body axes (anterior–posterior, dorsal–ventral, left–right). Historically, insights into human gastrulation were constrained to static analyses of the Carnegie Collection, a repository of preserved human embryos analyzed through anatomical and histological methods [1]. While murine models have dominated gastrulation research since the late 20th century, human gastrulation remains poorly understood due to ethical barriers (the “14-day rule”) and technical limitations in embryo culture [2–4]. Crucially, foundational principles of axis determination, primitive streak formation, and morphogenetic movements derive predominantly from murine embryology, despite mounting evidence of human-specific gastrulation patterns. Species divergences at molecular, cellular, and morphological scales necessitate the use of ethically compliant, human-specific experimental paradigms to study gastrulation.

Murine and human gastrulation exhibit key differences in both structures and gene activities due to their evolutionary divergences. In mice, gastrulation occurs in a cup-shaped embryo, while humans adopt a bilaminar disc. This structural difference affects cell movements and signaling patterns [5–9], highlighting the limitations of using mouse models to reflect human development and stressing the need for human-specific models. Currently, many stem cell-based models have emerged to compare human and mouse developmental dynamics in vitro, including blastoids, peri-implantation models, and gastruloids, representing embryonic stages before, around, and after implantation [10–14]. Recent advances in in vitro models, particularly three-dimensional (3D) gastruloids, have emerged as essential tools for bridging the knowledge gaps between mouse and human systems [10, 15]. The 3D gastruloids are self-organizing models derived from pluripotent stem cells (PSCs) that recapitulate key aspects of gastrulation, including spatial patterning, cell fate specification, and morphogenesis. This model provides an ethically permissible and experimentally tractable platform to study embryogenesis and developmental diseases in both mice and humans [4, 16–18].

Research on these gastruloids has progressed in tandem with advancements in both mouse and human systems. In mice, 3D gastruloids were first developed in 2014, demonstrating the self-organization of embryonic stem cells (ESCs) into 3D structures that recapitulate the germ layer patterning and body axis formation in early embryos [16]. Later studies advanced the field by capturing live imaging of gene product gradients and dynamics during development [19]. In 2020, similar 3D gastruloids were successfully created using human ESCs (hESCs) without extra-embryonic tissues [17]. These human models replicate key features of early human embryos, such as the formation of the primitive streak and germ layers. Additionally, research has shown that WNT pathway activation can induce axial elongation in these models, similar to “trunk” formation in human embryos [20]. Therefore, these 3D gastruloids bridge the gap between in vitro and in vivo gastrulation studies, providing a valuable platform for investigating this critical phase of embryogenesis.

In this review, we summarize the importance of gastruloids in advancing the research of human gastrulation by comparing mouse and human development and tracing the development from mouse to human gastruloid models. We also emphasize their current applications in disease modeling, drug screening, and forward-looking potential in personalized medicine, while reinforcing the advantages of gastruloids over traditional interspecies models.

A brief introduction of mouse gastrulation

Murine embryogenesis begins with fertilization, initiating a precisely timed developmental cascade. Maternally inherited transcripts and proteins orchestrate initial cleavage divisions, progressing from the zygote to the morula (16–32 cells) by embryonic day 2.5 (E2.5) [21]. Compaction occurs between E2.5 and E3.0, during which cells polarize and form tight junctions [22]. By E3.5, blastocyst formation demarcates the pluripotent inner cell mass (ICM) from trophoblast progenitors, with the blastocoel cavity enabling uterine implantation by E4.5 [23]. During the post-implantation stage (E5.5–E6.0), ICM segregation yields the primitive endoderm (hypoblasts in humans) and the embryonic epiblast. Radial symmetry breaking transforms the epiblast into a cylindrical embryonic sac (egg cylinder), a murine-specific architecture critical for subsequent gastrulation (Fig. 1A) [24, 25]. Concurrently, the anterior visceral endoderm (AVE) migrates to the prospective anterior pole, secreting Lefty1 and Cer1 to antagonize the posteriorizing signals including Wnt and Nodal (Fig. 1D) [26, 27]. This AVE-mediated signaling gradient polarizes the anterior-posterior (A-P) axis, while Bmp/Wnt/Nodal gradients subsequently refine the dorsal-ventral (D-V) and left-right (L-R) axes (Fig. 1B), establishing the coordinate system for gastrulation.

Fig. 1.

The early embryonic development of mice and humans. A The morphological dynamics from the zygote to the early gastrulation stage. The period in mice is from fertilization to E8. The period in humans is from fertilization to 20 dpf. AVE, anterior visceral endoderm. ExE, extraembryonic ectoderm. PGC, primordial germ cell. B The sections of mouse and human embryos from different perspectives. C The development of the paraxial mesoderm, intermediate mesoderm, and lateral plate mesoderm in mouse and human embryos. D The signaling pathways involved in the regulation of gastrulation

Gastrulation occurs between E6.5 and E8.5, during which pluripotent epiblast cells migrate and give rise to the three germ layers: ectoderm, mesoderm, and endoderm. This process begins with the formation of the primitive streak at the posterior epiblast, driven by Nodal and Wnt/β-catenin signaling (Fig. 1D). Epiblast cells undergo epithelial-mesenchymal transition (EMT), losing their adhesion properties and migrating inward through the streak [28]. These migrating cells form the definitive endoderm (DE), which coordinately interact and integrate into the primitive endoderm by intercalation and mesenchymal-epithelial transition (MET). The integrated endoderm undergoes internalization and forms the primitive gut tube, which gives rise to future internal organs [29]. Simultaneously, Nodal signaling plays a crucial role in determining cell fates through its gradient distribution. High levels of Nodal signaling induce cells to differentiate into mesoderm and endoderm, while low levels maintain ectoderm characteristics (Fig. 1D) [30]. Fgf8 promotes the proliferation and migration of epiblast cells, supporting mesoderm formation, and collaborates with Wnt signaling to ensure proper cell migration and positioning [31]. Bmp signaling primarily influences ectoderm differentiation and A-P axis formation [28].

Mesoderm cells subsequently differentiate into several subtypes, including the axial mesoderm, paraxial mesoderm, intermediate mesoderm, and lateral plate mesoderm (Fig. 1C) [32]. The paraxial mesoderm undergoes segmentation along the A-P axis to form somites, a process regulated by the segmentation clock and Fgf/Wnt/Notch signaling pathways. These somites later differentiate into key structures such as the vertebrae, muscles, and dermis. At E7.5, the primitive streak’s anterior end forms the node, which gives rise to the notochord, an essential axial structure (Fig. 1A). Cells from the node extend anteriorly to form the notochordal process, which eventually develops into the notochordal canal before solidifying into the notochord. The notochord secretes key signaling molecules, including Chordin and Noggin, which antagonize Bmp signaling. This process not only patterns the neural plate but also establishes the midline [33]. Neural tube formation begins around E8.5, as the neural plate folds and fuses at the midline. While the molecular and cellular events of gastrulation have been extensively characterized in murine models, human embryogenesis exhibits distinct spatiotemporal dynamics and regulatory networks. We next provide a brief introduction to human gastrulation, highlighting key milestones and unique features of early human embryogenesis.

A brief introduction of human gastrulation

Human embryogenesis initiates at fertilization and undergoes a cleavage process mechanistically similar to that observed in mice, albeit with species-specific temporal and molecular adaptations [34]. By 4 days post-fertilization (dpf), the morula stage is attained, followed by compaction and subsequent differentiation into a blastocyst characterized by a fluid-filled blastocoel cavity and the segregation of trophectoderm (TE) and ICM lineages (Fig. 1A) [35]. Within the ICM, the epiblast lineage establishes apical-basal polarity, proliferates to form a rosette-like structure, which coalesces into the amniotic cavity, a key milestone in human amniogenesis [36, 37]. Concurrent with amniotic cavity expansion, epiblast-derived cells undergo reorganization to generate the amniotic ectoderm, which envelops the dorsal embryonic surface and interfaces with the extraembryonic mesoderm. In parallel, hypoblast progenitors differentiate into the primitive endoderm, giving rise to cuboidal cells that organize into the exocoelomic membrane (also known as Heuser’s membrane), thereby delineating the primitive yolk sac [38]. By week 2, these coordinated morphogenetic events culminate in the establishment of the bilaminar embryonic disc (Fig. 1A).

Human gastrulation commences at approximately 14 dpf with the emergence of the primitive streak, driven by conserved NODAL and WNT signaling pathways analogous to those in mouse models [39]. Symmetry breaking and A-P axis establishment depend on signaling from an AVE-like signaling center, a structure functionally equivalent to the murine AVE. Unlike the cylindrical mouse embryo, human gastrulation occurs within a bilaminar disc, with the primitive streak extending radially across the epiblast surface. At the anterior terminus of the primitive streak, the human node (an Hensen’s node homolog) emerges as a transient organizer, orchestrating notochord progenitor specification through coordinated signaling.

WNT signaling stabilizes cytoplasmic β-Catenin accumulation, thereby reinforcing epithelial polarity and directing collective cell migration along the primitive streak. This process is dynamically antagonized by Dickkopf-1 (DKK-1)-mediated WNT inhibition (Fig. 1D) [40, 41]. Between 10 and 12 dpf, epiblast cells undergo WNT/FGF2-dependent EMT, characterized by E-CADHERIN downregulation and SNAIL1 activation, enabling delamination and mesodermal progenitor migration [42, 43]. By 12–14 dpf, lineage specification becomes molecularly distinct: FOXA2⁺ DE ingress ventrally; FGF2/ERK signaling promotes T (also named Brachyury)⁺ mesoderm commitment; while BMP2-driven chemotaxis guides mesodermal sheet migration [43–45].

A dorsal-ventral BMP4 morphogen gradient originating from the amniotic ectoderm instructs cell fate decisions: high levels of BMP4 stabilize ectodermal identity via SMAD1/5/8 phosphorylation, whereas low levels of BMP4 permit mesendodermal differentiation [45, 46]. Following germ layer segregation, the trilaminar disc undergoes axial organization. The notochord secretes Sonic hedgehog (SHH) to pattern the ventral neural tube [47], while the paraxial mesoderm segments into somites, recapitulating the process in mouse (Fig. 1C) [48].

Axis specification in humans exhibits conserved principles with mice but diverges in spatiotemporal regulation: Wnt/β-catenin and BMP antagonism establish the dorsal-ventral axis after embryo implantation, while left-right asymmetry is initiated later via cilia-driven nodal flow in the embryonic disc [49, 50].

In summary, human gastrulation constitutes a pivotal phase in embryogenesis, coordinating spatiotemporal programs of germ layer specification, large-scale cell rearrangements, and morphogen gradient establishment to generate the embryonic body plan. Deciphering human gastrulation dynamics provides fundamental insights into the epigenetic-signaling interplay governing cell fate bifurcations and self-organizing principles of tissue morphogenesis.

The development of human and mouse gastruloids

A brief history of gastruloid technique

Gastrulation is a vital stage in mammalian embryo development, however, the signaling pathways and molecular mechanisms involved remain largely unexplored. Before the successful derivation of human gastruloids, research on early human embryogenesis faced three principal challenges: (1) limited access to developmentally staged human embryos due to ethical and technical barriers. The internationally ratified 14-day rule, which prohibits the in vitro culture of human embryos beyond primitive streak formation, further restricted systematic experimental investigation of post-implantation development; (2) technical limitations in maintaining ex vivo embryo viability beyond implantation stages (approximately 7 dpf); and (3) critical biological divergences between species that reduced translational relevance of animal models.

Although murine models provided valuable insights, their developmental processes diverged significantly from human-specific events, such as trophoblast specification and axial elongation kinetics, creating critical knowledge gaps in understanding human morphogenesis. Early in vitro attempts to model embryonic development relied on two-dimensional (2D) systems, which faced significant limitations. In adherent cultures, cell populations were heterogeneous and skewed toward specific cell types, failing to recapitulate the spatiotemporal coordination observed in vivo, where diverse cell populations collaborate to form tissues and organs. An alternative strategy adopted the approach of seeding hESCs on micropatterned surfaces, which generated coordinated gene expression patterns but fell short of reproducing axial organization, a hallmark of embryonic development [51]. These challenges in recreating spatial organization and multi-lineage coordination underscore the need for more physiologically relevant 3D models. Prior to gastruloids, embryoid bodies (EBs), simpler 3D aggregates concentrated on the potential of pluripotent stem cells to differentiate into derivatives of multiple germ layers in both mouse and human, serving as models of early developmental events. However, they fail to mimic later events such as morphogenesis and spatiotemporal patterning [52]. In contrast, gastruloids are designed to specifically model typical events during gastrulation. Although mouse gastrulation exhibits similarities with human gastrulation, there are notable morphological and genetic differences. The establishment of mouse gastruloids provides a conceptual and technical framework specifically for studying gastrulation in vitro, and this framework is instrumental for subsequent research to model human gastruloids [53]. The development of both mouse and human gastruloids is intermingled, and comparing the interspecies differences will promote our understanding of the complex human gastrulation process (Fig. 2).

Fig. 2.

The timeline of the development of mouse and human gastruloids. The establishment history of mouse (upper) [13, 16, 54, 55, 62, 63, 67, 68, 113] and human (lower) [17, 70, 73–76] gastruloids, from the generation of the first protocol to the promotion of more advanced gastruloid protocols

Current murine gastruloids are capable of simulating key developmental events spanning gastrulation and post-gastrulation stages, offering both practical and ethical alternatives for investigating early embryogenesis [16, 54–56]. Their development built upon foundational studies delineating molecular prerequisites for in vitro stem cell self-organization, including pioneering work on embryoid body formation and 3D culture optimization [57–60]. The field reached a milestone in 2014 when Baillie-Johnson et al. introduced the first standardized protocol generating self-organizing aggregates that spontaneously undergo symmetry breaking, axial elongation, and embryonic patterning through conserved signaling pathways [16, 19, 53, 61]. Leveraging these murine insights, pioneering research in 2020 engineered human gastruloids, 3D human PSC (hPSC)-derived structures mimicking human embryonic patterning, marking a paradigm shift in developmental modeling [17]. These systems employ rigorously validated hPSC lines such as MasterShef7 and S4-GATA6-GFP, selected for their differentiation competence and genomic stability. The resulting models now permit mechanistic investigation of previously inaccessible human-specific processes, including primitive streak specification and germ layer segregation.

Subsequent refinements to the murine platform enable the recapitulation of advanced morphogenic processes, including trunk-like structures [55], anterior neural tissues [58, 62–64], neural tube [55, 56, 65], somite [54, 55, 65, 66], gut tube [55, 67], and beating heart-like domain [67]. Hybrid models integrating mouse ESCs (mESCs) with trophoblast stem cells (TSCs) [68], extra-embryonic endoderm (XEN) cells [69], or Gata4-expressing ESCs [13] demonstrate enhanced spatiotemporal coordination. Human gastruloid development mirrors this trajectory, with recent innovations generating models containing neural structures [70, 71], hematopoietic structures [72], primordial germ cell-like cells (PGCLCs) [73, 74] and neural crest structure flanked with segmented somites [75]. A significant advance in structural integrity emerges with Peri-gastruloid, a model deriving from human extended pluripotent stem cells (hEPSCs) and incorporating both embryonic (epiblast) and extraembryonic (hypoblast) lineages. This model now recapitulates the formation of amniotic and yolk sac cavities, achieving unprecedented structural completeness [76] (Fig. 3).

Fig. 3.

The morphology of human gastruloids from selected protocols. The upper panel shows the morphological dynamics of human gastruloids in the initial protocol with culturing time. The lower panel shows the morphology of four different advanced human gastruloid protocols (early cardiac structures with affiliated neural tissue, primordial germ cell-like cells, peri-gastruloid, and RA-induced gastruloid)

The breakthroughs synergize with cutting-edge analytical technologies. Live imaging captures dynamic morphogenic events during axial formation [13], while high-resolution lineage tracing has identified murine primitive endoderm cells [13, 54, 77], neuroepithelial progenitors [58], trunk precursors [55], which also underscore the models’ utility in decoding developmental mechanisms. Additionally, single-cell transcriptomics reveals conserved spatiotemporal gene expression hierarchies between gastruloids and embryos [77]. In the future, integration of Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) screening and organoid fusion approaches promises to further bridge the gap between in vitro models and in vivo embryogenesis. Illustrated by key gastruloid models, we will discuss the properties of murine and human gastruloid models, review key developmental events, including symmetry breaking, axis patterning, and germ layer segregation, in accordance with the developmental order in vivo, and cover both evolutionary conservation and human-specific adaptations, necessitating systematic cross-species comparisons.

Symmetry-breaking morphogenesis: from WNT activation to axial primordium

Symmetry breaking, driven by Wnt/β-catenin signaling, demonstrates cross-species conservation yet distinct implementation strategies. In mESC-derived embryonic bodies, symmetry breaking manifests as elongation, cell movements, and the establishment of A-P polarity within 48 h of 3 µM Wnt-agonist CHIR99021 (CHIR) exposure [16, 19, 53]. In contrast, the human system requires a more complex modulation. Successful symmetry breaking involves both CHIR exposure in pre-treatment and early aggregation, as well as the balance of additional signaling pathways, such as BMP and Nodal [17]. Moreover, teratogens, such as retinoic acid (RA), can impair elongation and axial organization in human gastruloids [17]. The elongation and polarized gene expression observed during this process parallel the establishment of the A-P axis.

Symmetry breaking within aggregates is considered the first step in axial formation. The emergence of the three body axes (anterior-posterior, left-right, and dorsal-ventral) establishes the vertebrate body plan (Fig. 1B). Primary axial formation and associated gene networks are highly conserved across metazoans, including mice and humans. The Wnt/β-catenin signaling pathway drives symmetry breaking and plays a key role in cell fate decisions and pattern formation [16, 68, 69, 78]; T is also essential for primary axis formation and germ layer specification [68, 69, 78, 79]. The Cripto gene is required for A-P axis formation in mouse gastruloids, as evidenced by the failure of elongation in Cripto knockout aggregates [80]. However, the role of Cripto in human gastruloids remains uncharacterized. Nevertheless, the progress of symmetry breaking initiates the formation of the three axes for the body plan.

The establishment of left-right (L-R) asymmetry is indicated by the L-R asymmetric Nodal and Cer1 expression in the initial mouse gastruloid model. However, the presence of cilia and other L-R axis formation associated processes is absent [19]. In contrast to A-P and L-R axes, the first stem-cell-based model exhibiting dorsal-ventral (D-V) polarity was reported later. Veenvliet and colleagues demonstrated somitogenesis along the D-V axis after embedding 96-hour-old gastruloids in 5% Matrigel for 120 h [55]. In neural tube-like structures derived from assemblies of mESCs and XEN cells, neural cells exhibit dorsal characteristics, which implies that XEN cell-derived compartment can serve as a signal center for the induction of the D-V axis [56].

In summary, the elongation and axial establishment are crucial developmental events in embryonic development, serving as benchmarks for measuring the gastruloid formation efficiency, assessing the stem cell competency to aggregation and gastrulation, and evaluating cell pluripotency [80].

Germ layer specification in gastruloids: divergent pathways to primitive streak formation

After axial establishment, both mouse and human embryos undergo primitive streak formation, a crucial event during which cells differentiate into mesodermal and the endodermal lineages, forming epithelial structures. In gastruloids, the dynamics of this process exhibit similarities across species. For instance, the upregulation of T and SOX17 in human gastruloids mirrors the mouse models, marking mesoderm and DE specification, respectively. However, the relative signaling processes involved may be divergent. In mouse gastruloids, although the primitive streak structure has not been observed, the primitive streak process is reported to rely on the activation of Activin and Wnt/β-catenin signaling which upregulates T and β-Catenin expression [16, 81].The primitive streak marker T is reported to closely follow the Wnt agonist’s concentration and downregulate the pluripotency marker Sox2 [82]. In human gastruloids, while WNT/β-catenin signaling remains critical, additional pathways such as BMP and NODAL are also required for proper primitive streak formation and germ layer specification. Moreover, human gastruloids display unique spatiotemporal patterns of gene expression and cell sorting that differ from those observed in the mouse systems. It is reported that SOX17⁺ cells emerge in a stochastic yet deterministic manner, suggesting an interplay between intrinsic molecular states and earlier developmental cues [17]. These cells subsequently self-sort, with E-CADHERIN⁺ islands migrating to the periphery, a process reminiscent of the in vivo sorting observed in mouse gastruloids [83]. This stochastic emergence of Sox17⁺ cells marks the initiation of endoderm morphogenesis. The pattern in which these cells appear, both stochastic and deterministic, implies that the formation of DE is influenced by early developmental events and intrinsic molecular states [83]. Following the Sox17 expression, cells in the primitive endoderm undergo self-sorting, with E-cadherin⁺ islands flowing away from the tip of the gastruloid, aligning with the peripheral regions. This sorting process mirrors the in vivo sorting of E-cad⁺ and E-cad^- populations [83].

However, in contrast to the dynamic mesodermal and endodermal differentiation processes observed in vivo, mouse gastruloids do not undergo the classical EMT. Cells in the mouse gastruloids maintain an epithelial state and do not transition into a mesenchymal phenotype as they differentiate into mesodermal and endodermal lineages [83–85]. In contrast, unlike the classical EMT in vitro, DE cells in mouse gastruloid intercalate into the extra-embryonic tissue-derived visceral endoderm layer [85, 86], contributing to the formation of gut endoderm-like cell populations. In subsequent developmental stages, these cells form gut tube-like structures [55, 86].

In contrast, human gastruloids exhibit more complex morphogenetic behaviors that include partial EMT, characterized by the downregulation of E-CADHERIN and activation of SNAIL1, and extensive tissue reorganization as well as enhanced cell migration [18, 37]. DE cells in human gastruloids efficiently integrate into VE layers, forming lumen-like structures that resemble early gut tube development. These processes are accompanied by enhanced self-organization capabilities, enabling human gastruloids to recapitulate later developmental events, such as neural plate folding (SOX2⁺ polarized epithelium) and somite segmentation (periodic TBX6⁺ clusters), which are either absent or less organized in mouse gastruloids [17, 55]. By integrating human extraembryonic tissues with gastruloids and proper stimulation of extraembryonic signaling, a recent study observed the alignment and internalization of T⁺ cells at the center of the embryonic-disk-like structure [76]. Meanwhile, some T⁺ cells expressing N-CAD indicate the similar process with EMT, strongly indicating the existence of relevant mechanisms in the body. The emergence of this structure makes it easier to locate and track the subsequent biological events, making it possible to explore more precise mechanisms in the next stage. The primitive streak formation and EMT migration give rise to endoderm and mesoderm, allowing for the following tissue reorganization on a large scale. The node, forming in the anterior end of primitive streak, sends signals to induce migrating cells to a notochord fate, which promotes paraxial mesoderm to differentiate into somite structures.

The formation of node, notochord and somites in gastruloids follows an ordered process

The structures including node, notochord, and somites are pivotal in establishing the midline of vertebrate body, ensuring the correct localization of skeletal, muscle and other tissue types. While these processes in mouse embryogenesis are well-characterized in vivo, the morphogenesis and molecular mechanisms have not been elucidated in human embryogenesis. The advances in mouse gastruloid studies have enabled credible in vitro confirmation of the proposed mechanisms, while studies in human gastruloids reveal both conserved and divergent features compared with mouse, offering valuable insights into deciphering unknown developmental mechanisms.

In embryos, the ventral node is a midline structure that serves as an organizer for the establishment of L-R axis, mesodermal formation and notochord development. The leftward fluid flow is generated and detected by motile cilia in the node transduces signals, resulting in the precise L-R polarized tissue development [87]. In 2018, the first report of a node-like structure in gastruloids emerged, but it was noted that recognizable notochord derivatives were absent [19]. To date, gastruloids with fully developed notochord-like structures have not been successfully generated. However, recent advances in human gastruloid models suggest partial recapitulation of L-R patterning. For instance, asymmetric Nodal expression has been observed in human gastruloids [17], although motile cilia and nodal flow, which are critical for L-R symmetry breaking in vivo, remain unreported in these models. This gap could be attributed to the absence of extra-embryonic tissues or specific extracellular matrix (ECM) cues in vitro, which are essential for proper L-R signaling.

The formation of the node and notochord precedes somitogenesis during the late gastrulation stage. In somitogenesis, a string of somites emerges sequentially along the A-P axis from presomitic mesoderm (PSM) under the regulation of the node and notochord. In mouse gastruloids generated with the initial protocol, the emergence of somites in a randomized L-R manner is probably due to the lack of extra-embryonic tissues [15]. However, more recent studies have confirmed two key molecular activities involved in mouse somitogenesis, which are also observed in gastruloid models. The first is the ‘clock’ mechanism (or segmentation clock), which refers to an oscillating expression pattern of Notch-signaling genes [55]. The second is the ‘wavefront’, a signal gradient that decreases from the rostral to the caudal end of the embryo. Signaling molecules such as Fgf, Wnt and Bmp play key roles in the coordinated regulation of somite segmentation in mice [54, 55]. Among these, Fgf signaling is identified as the primary mediator of wavefront activity [88]. It has also been reported that embedding gastruloids on low-percentage (5–10%) Matrigel induces bilateral somite formation at 96 h post-aggregation [54, 55]. The specific stiffness of Matrigel may help recapitulate the ECM that envelops embryos in vivo, contributing to the formation of more structured somites.

In human models, the development of somites has been explored using organoid models, such as the somatoid derived from human induced PSCs (iPSCs). These models express somite-stage-specific marker genes but, unlike gastruloids, do not fully replicate embryo-like structures at the gastrulation stage. Notably, inhibition of FGF and WNT pathways has been shown to promote the efficient formation of somite structures in human models [89]. Pre-somitic mesoderm markers, including HES7, a transcription factor reported to regulate the pace of somitogenesis, are expressed in human peri-gastruloids [76]. In 2024, a novel protocol was developed to induce human gastruloids with neural tube and segmented somite structures by intermittently applying RA and Matrigel. Specifically, an initial pulse of 500 nM RA for the first 24 h, followed by a lower concentration of 100 nM RA and 5% Matrigel from 48 to 120 h, robustly generates trunk-like structures. Inhibition of BMP signaling leads to a reduction in somite marker gene expression, suggesting the importance of BMP signaling in somitogenesis [75]. This study reports segmented somite-like structures accompanied by periodic TBX6 expression. This pattern indicates the human somite mechanism may parallel the segmentation clock in mice, yet it is important to note that the complete oscillatory gene network observed in mouse somitogenesis has not been demonstrated in human models. The segmented structures are vital to interact with subsequent developmental events such as neurulation. In mouse embryo, the formation of somites functions as a signaling center to guide neural tube patterning [90], while experimental evidence in humans is still lacking.

Neurulation and the specification of neural precursors

During gastrulation in the embryo, the anterior ectoderm, located along the midline, thickens and folds to form the neural plate, marking the initiation of neuralization. This process is fundamental to neural morphogenesis in both in vivo and in vitro models. In mice, neural development is orchestrated by signaling and transcriptional activities from the AVE and DVE [13, 68, 69, 78]. Antagonists and transcription factors released by these regions are essential for anterior patterning in the embryo [64], a concept supported by recent findings [62]. Specifically, inhibition of Wnt signaling using antagonist XAV939 during early developmental stages can guide mouse gastruloids to form anterior neural tissues, including forebrain, midbrain and hindbrain-like regions. This observation underscores the critical role of Wnt signaling modulation in early neural differentiation [63]. Moreover, in models derived from mESCs and XEN cells (referred to as “neuruloids”), neural morphogenesis is initiated when XEN cells locally release Wnt inhibitors to suppress primitive streak formation, promoting the establishment of a tubular neural structure [56]. Therefore, the localized inhibition of Wnt signaling might be a key mechanism for robust neural tube-like structure assembly in vitro.

Recent studies have also highlighted that some posterior neural tissues arise independently of the anterior neural plate. In particular, bipotent progenitor cells, known as neural mesodermal axial progenitors (NMPs), contribute to both spinal cord and paraxial mesodermal development [58, 91]. These NMPs arise from the epiblast associated with the primitive streak and are molecularly and functionally distinct from anterior neural precursors [91]. In vitro induction of NMPs, however, comes at the cost of anterior neural tissues, as the addition of exogenous CHIR is required to facilitate NMP differentiation [16, 19, 53, 58, 91]. The development of novel models that enable the differentiation of both anterior and posterior neural lineages is crucial for further understanding the mechanisms underlying embryonic neurulation.

The proportion of NMPs in different species may determine the ability of gastruloids to form neural structures. For example, the number of NMPs in conventional human gastruloids is significantly lower compared to mouse gastruloids, particularly those derived from the tailbud-like structure (TLS). This difference likely accounts for the greater difficulty in inducing neural tissue formation in human models, underscoring the need for improved methods to enhance neural precursor specification in human gastruloids.

Species-specific insights necessitate human gastruloid models

The study of mammalian gastruloids reveals intricate interplays between evolutionarily conserved mechanisms and species-specific developmental programs. While murine and human gastruloids share fundamental developmental logics, species-specific trajectories emerge during critical morphogenetic transitions. Core processes including symmetry breaking, axis patterning, and germ layer segregation exhibit both evolutionary conservation and human-specific adaptations, necessitating systematic cross-species comparisons.

Human gastruloids, first established by Moris et al. (2020), have emerged as a powerful in vitro model for investigating early human embryogenesis, particularly in the context of gastrulation and axial patterning [17]. These models complement the long-established mouse gastruloid systems, which have been pivotal in advancing our understanding of mammalian developmental mechanisms. While mouse gastruloids excel in modeling conserved processes such as mesoderm formation and body axis elongation, human gastruloids offer unique insights into species-specific aspects of development, including differences in gene regulation, cellular fate decisions, and tissue organization.

More specifically, recent findings represented by gastruloids have demonstrated the differences between humans and mice in embryogenesis with many aspects: In morphological changes, while mouse gastrulation occurs within a cup-shaped egg cylinder, with primitive streak forming at posterior end, humans represent a bilaminar disc, with primitive streak extending radially. In signaling pathway regulation, the requirement is also diverse. While mouse gastruloids are efficiently driven by Wnt activation in symmetry breaking, human models require more precise pathway balance and temporal regulation (WNT/BMP/NODAL coordination). Primitive streak formation in mouse relies heavily on Wnt/β-catenin and Nodal signaling. However, in humans, BMP signaling plays a more prominent role alongside Wnt and Nodal. In EMT behaviors, while DE cells in mouse gastrulid intercalate into extra-embryonic layer and maintain epithelial characteristics, human gastruloid EMT is accompanied by extensive cell migration and tissue remodeling. In neural induction, mouse gastruloids contain abundant NMP, facilitating robust posterior neural and mesodermal differentiation, and Wnt inhibition can effectively boost neural induction. Compared with mouse, humans exhibit lower induction efficiency in NMP. In somitogenesis, mouse gastruloids recapitulate the segmentation clock and wavefront to guide somite formation, while the fully oscillatory gene network has not been demonstrated in humans, RA signaling is reported to play a prominent role. Subsequent refinements in human gastruloid models have significantly enhanced their ability to recapitulate key developmental events, addressing the ethical and technical limitations that hinder direct research on human embryos.

By providing detailed examination of the molecular and cellular dynamics underlying early human development, human gastruloids bridge critical gaps in our understanding of both conserved and species-specific mechanisms. For instance, comparative studies between mouse and human gastruloids have revealed divergent roles of signaling pathways, such as FGF and WNT, further emphasizing the necessity of human-specific models to accurately interpret developmental processes. Beyond their contribution to basic developmental biology, these models hold immense potential for translational applications.

Application of gastruloids

Given the difficulty and high cost of accessing mammalian embryos, much of the developmental mechanism during the early embryogenesis remains poorly understood. In particular, human embryogenesis is constrained by ethical concerns and the “14-day rule”, which limits direct study of human embryos [2–4]. As a result, much of our understanding of human development is inferred from studies on mouse models. The maturation of gastruloid technology offers a valuable alternative, providing a platform that complements traditional models. It allows researchers to explore unsolved molecular mechanisms and signaling pathways, contributing to our understanding of birth defects, miscarriage, and other developmental disorders. Moreover, as gastruloids can be generated in a scalable, reproducible, and accessible manner, they hold considerable promise for drug research and regenerative medicine.

Applications of gastruloids to understand early embryogenesis

The development of gastruloids has provided an ideal in vitro model for studying early developmental processes. In developmental biology, gastruloids serve as a model to investigate the spatiotemporal dynamics, cellular processes, and molecular pathways active during gastrulation. Recent studies have leveraged omics technologies to gain deeper insights into early embryonic development. For example, single-cell RNA sequencing has been used to characterize mouse gastruloid formation and to improve anterior structure development through dual Wnt pathway modulation [81]. Additionally, multilayer proteomics has identified new cell subtype markers and potential transcription factors in mouse gastruloids [92].

In stem cell biology, gastruloids offer a valuable platform for exploring the regulatory network involved in the differentiation process. They can be used to better guide directed differentiation and trans-differentiation. For example, exposure to bisphenol A, an environmental toxicant, impairs axial elongation in gastruloids by perturbing the Wnt/β-catenin pathway [93]. Using single-cell RNA sequencing, another study found that transcription factor ETV1 disrupts germ-layer organization and causes ectoderm loss in human gastruloids by dysregulating PI3K/AKT signaling [94]. Gastruloids can also be employed to investigate the potential roles of specific molecules through gene knockout or environmental treatments, with follow-up studies using single-cell RNA sequencing to uncover the underlying regulatory network governing stem cell differentiation (Fig. 4).

Fig. 4.

The future applications of human gastruloids. Gastruloids have promising potential in drug screening, regenerative medicine, research applications, and disease modelling. Gastruloids enable scalable production of developmentally synchronized model systems, facilitating high-throughput drug screening. With advances in gastruloid techniques, more complex gastruloids with multiple early organ structures will be generated, providing potential transplantation or progenitor sources. By introducing specific genetic mutations or environmental factors into gastruloids, researchers can explore the mechanisms underlying various diseases, which could serve as a basis for therapeutic development. Furthermore, gastruloids generated from iPSCs derived from congenital patients may generate individual-specific disease models, which would be beneficial in formulating personalized treatment plans

In the field of evolutionary developmental biology (evodevo), gastruloids are useful for studying conserved and divergent mechanisms between species, shedding light on the potential evolutionary trajectories. Comparative studies between human and mouse gastruloids have revealed similarities in the timing and spatial localization of Hox gene expression, supporting the conserved role of Hox genes across species [95]. In the future, these comparisons may help to explain the evolution or degradation of specific structures in the human body.

Application of gastruloid in disease modelling

Initially, research in the field of gastruloids focused primarily on optimizing their culture conditions and understanding developmental mechanisms. More recently, there has been a shift toward utilizing gastruloids as a platform for modeling developmental defects and diseases. Gastruloids, which mimic the interaction of multiple germ layers and basic patterning of early embryogenesis, particularly during gastrulation and early organogenesis, provide an invaluable tool for studying developmental failures at these critical stages.

Gastruloids offer a platform to dissect the pathophysiology of developmental disorders by perturbing key signals, the abnormality of which can lead to lethality or deformity. In mouse gastruloids, the synergic roles of Wnt and Nodal signaling in axial elongation have been well-documented, with both pathways playing key roles in A-P axis formation [79]. Inhibition of Nodal or absence of Cripto, a coreceptor for Nodal, disrupts the A-P axis, reflecting the axial truncation phenotype observed in the corresponding mutant embryos [79, 80]. Similarly, inhibition of FGF signaling with the PD03 inhibitor leads to premature growth and shorter morphology, recapitulating the short-tail phenotype observed in vivo [54, 88]. In human gastruloids, inhibition of BMP or WNT signaling leads to failure in aggregate formation and elongation [17], highlighting the sensitivity of early human morphogenesis to these pathways. Additionally, treatments with teratogens such as RA or thalidomide lead to increased SOX2 and reduced T expression, directly modelling drug-induced teratogenic effects and embryopathies [17, 96, 97].

Studies have successfully modelled disorders of somitogenesis and neurogenesis in gastruloids, which are two crucial events with frequent birth defects during early organogenesis [54–56, 75]. For instance, modulating the extracellular matrix environment, such as increasing the Matrigel concentration to 10%, enhances the fidelity of somite-like structure formation and anterior-posterior polarity [55, 98]. The controlled system allows for the precise dissection of pathogenic mechanisms. When modelling somitogenesis in gastruloids, key signalling pathways implicated in vertebral segmentation defects have been perturbed: Somite formation requires a gradient of RA and the oscillation signaling of Wnt, Notch and Fgf pathways in mouse models, and the perturbation of the signaling is reported to cause abnormal vertebral segmentation like spondylocostal dysostosis (SCDO) and Alagille syndrome (AGS) in humans [99, 100]. In mouse gastruloids, these pathways were perturbed to mimic the abnormal segmentation. Excessive activation of Wnt signaling and inhibition of Bmp signaling leads to abnormal somite formation, creating a phenotype reminiscent of “a bunch of grapes” [55]. The MEK inhibitor induces premature growth leading to a posterior shift in somite formation, resembling the short-tailed phenotype in vivo [54]. Similarly, FGF-receptor inhibition produces outcomes similar to MEK inhibition, further supporting the correlation with the in vivo short-tail phenotype [54, 88]. In human gastruloids, the finding that early RA pulses also promote segmented somite formation confirms the shared mechanisms of congenital abnormalities between humans and mice [75], indicating the platform’s utility for studying human congenital diseases to teratogens and genetic variants affecting axial development.

In both species, neurogenesis relies on signaling cues such as BMP, Hedgehog and RA, while transcription factors such as TBX6 and PAX3 are crucial for neural tube formation [101]. Notably, gene-editing studies have begun to directly link mutations to gastruloid phenotypes. In a proof-of-principle experiment, Tbx6 knockout in mouse gastruloids resulted in ectopic neural tube formation at the expense of somite development, faithfully modeling the spina bifida-like phenotype of Tbx6 mutant embryos [55, 102]. RA-induced gastruloids in humans similarly demonstrate the essential roles of TBX6 and PAX3 in neural tube formation and localization [75]. This underscores the conservative function of these transcriptional factors across species, and following studies on perturbation of these two genes are expected to model corresponding human neural tube defects.

Looking forwards, the potential application of gastruloid in disease modelling lies in combining this validated phenotype-presenting platform with patient-specific genetics. While current gastruloid-based disease modelling predominantly focuses on introducing known genetic or environmental perturbation, future work will leverage patient-derived iPSCs harboring mutations in genes associated with specific congenital diseases to generate isogenic models. This approach assists in elucidating the underlying molecular mechanisms and their interplays with phenotypic manifestations, ultimately facilitating drug screening and personalized therapeutic strategies (Fig. 4).

Application of gastruloids for drug screening and drug development

Besides understanding basic regulatory mechanisms, gastruloids have emerged as an ideal system for initial drug screening and development, largely due to their embryo-like structures and high-throughput competency. Unlike conventional 2D ESC-based drug tests, gastruloids provide a more comprehensive picture of overall developmental processes, making them a superior model for assessing drug effects. As gastruloids grow in 96-well plates and allow for real-time imaging, they offer a dynamic and quantitative platform for monitoring developmental changes in response to drug exposure. This high-throughput capability enhances the ability to analyze drug effects across multiple parameters.

In human trials, Developmental and Reproductive Toxicity (DART) testing is often limited by ethical concerns [103]. Gastruloid-based models using hPSCs offer a solution by excluding species-specific effects and reducing the reliance on animal models (Fig. 4). Meanwhile, it evaluates the toxicological and teratogenic effects of drugs by monitoring a range of factors, including morphological and elongating perturbations, gene expression patterns, and cell viability [96, 104]. The application of high-throughput analytical technologies on toxicity tests in gastruloids enables to elucidate underlying molecular mechanisms of teratogenicity. A notable example is valproic acid (VPA), an embryotoxic compound causing neural tube defects such as spina bifida [105]. Treatment of VPA on human gastrulating stem cells (hGaSCs)-derived gastruloids, followed by single cell RNA-sequencing, reveals that SHH, a notochord-secreted signaling molecule critical for neural tube development, is remarkably downregulated [106]. Additionally, gastruloid-based platforms are uniquely capable of capturing species-specific drug effects. For example, while thalidomide is teratogenic in primates [107], rodents are not affected. In contrast, human gastruloids exhibit impaired elongation and increased SOX2 expression when exposed to thalidomide [96], demonstrating the necessity of evaluating human-specific drug responses. Furthermore, in scenarios requiring rapid drug testing, where time-consuming animal tests may impede the process, gastruloids offer an ideal alternative. Administration of N4-hydroxycytidine (NHC), a metabolized form of anti-COVID-19 drug, leads to reduced aggregate size in human gastruloids at plasma level corresponding to the clinically recommended dose [108]. However, limitations exist in gastruloid-based testing. For instance, drugs that bind to plasma proteins or require maternal metabolism may hinder accurate toxicity assessments and determination of the minimum toxic concentration [97]. Additionally, the limited concentration range and imprecise quantitative analysis methods can affect the accuracy of drug testing.

Beyond drug screening, gastruloid models also serve as a platform for the development of novel drugs. For instance, budesonide, a glucocorticoid used to treat asthma, impedes the exit from pluripotency of ESCs by promoting global cell-cell adhesion, although high local cell-adhesion tension in specific geometrical regions has been reported to promote germ layer specification in recent research [109]. Exposure of mouse gastruloids to budesonide maintains E-cadherin expression and prevents spontaneous symmetry breaking, leading to reduced elongation. The phenotype caused by budesonide suggests its potential in treating diseases involving transitions in cell migration, such as metastatic cancer [110]. Recently, a budesonide analog, CHD-030498, showed similar effects on mouse gastruloids at a lower dose, highlighting its potential as a novel therapeutic option [111].

Although gastruloids show high potential in screening the safety of drugs and detecting novel uses of typical drugs, they have not been used in large-scale clinical trials, probably owing to the lack of phenotypic identification criteria. In the future, more precise and sensitive gastruloid-based testing systems will be developed. Additionally, personalized gastruloid models for drug testing could pave the way for more tailored and effective therapeutic interventions.

The application of gastruloids in regenerative medicine

Gastruloids have emerged as a groundbreaking platform in regenerative medicine, offering unparalleled advantages in modeling early organogenesis and tissue-tissue interactions. These self-organizing 3D structures, derived from pluripotent stem cells, are proving to be more accurate and versatile than traditional systems in studying developmental processes and generating functional progenitor cells for therapeutic purposes. Their capability to recapitulate key aspects of embryonic development holds significant promise for advancing organ transplantation and cell-based therapies (Fig. 4).

In the field of cardiogenesis, gastruloids derived from both human and mouse models are capable of forming early cardiac structures, an achievement made possible through careful regulation of growth factors such as vascular endothelial growth factor (VEGF) and ascorbic acid [67, 71]. This ability to generate functional cardiac progenitors from pluripotent stem cells offers substantial advantages in understanding the mechanisms of heart development and enhancing protocols for cardiac regeneration. Notably, studies on the cardiopharyngeal mesoderm within mouse gastruloids have provided valuable insights into cardiac and skeletal muscle lineage differentiation, which could lead to improved treatments for cardiac diseases and muscle dystrophies [112]. The capacity of gastruloid system to model complex tissue interactions at early stages makes it an indispensable tool for exploring heart development and potential regenerative therapies.

Gastruloids also hold great promise in the field of hematopoiesis. The generation of early transplantable blood progenitors from mouse gastruloids demonstrates their remarkable ability to recapitulate hematopoietic differentiation pathways with high spatiotemporal accuracy [113]. This contrasts with traditional systems, such as embryoid bodies (EBs), which tend to show ectopic gene expression. Gastruloid-derived blood progenitors, on the other hand, follow a precise hematopoietic developmental trajectory, highlighting the advantages of the gastruloid system in modeling developmental processes with greater fidelity [114]. Furthermore, optimization strategies involving mechanic forces, metabolism states, and epigenetic regulation have been shown to enhance hematopoietic differentiation, making gastruloids a promising system for generating blood cell progenitors for clinical applications [72].

In the germ cell field, gastruloids have revolutionized the generation of PGCLCs. Unlike traditional methods, such as floating 3D aggregates, gastruloids can generate PGCLCs without the need for exogenous PGC-specific maturation factors. This ability to mimic the co-development and interaction of PGCLCs with somatic cells provides a unique advantage, as it allows for the endogenous signaling mechanisms required for optimizing germ cell identity transitions. Furthermore, gastruloid-derived PGCLCs reach a more advanced maturation status (~ E14.5 stage) compared to EB-derived PGCLCs (~ E9.5 stage), indicating their superior potential as an in vitro model for developing transplantable germ cells [115]. This development holds significant promise for addressing infertility and advancing reproductive medicine by generating functional germ cells for therapeutic purposes [73].

Gastruloids also offer unique opportunities in neural development. Mouse gastruloids have been successfully used to model early neural tube formation and its associated structures, providing insights into the intricate interactions between embryonic and extraembryonic tissues during neural development [56, 101].

While gastruloids are still limited in their ability to fully recapitulate late-stage organ development, their precision and versatility in modeling early-stage organ formation are unmatched. The faithful representation of early tissue interactions in gastruloids provides invaluable insights into developmental processes that were previously difficult to study in vitro. Their ability to generate tissue-specific progenitor cells that are more likely to integrate successfully in transplantation settings holds enormous potential for clinical applications. While research on human gastruloids is still in its early stages, the continuous refinement of this model will likely open up new avenues for regenerative medicine. The ability to generate complex, functional cell types from pluripotent stem cells with greater accuracy and predictive power will be key to unlocking the full therapeutic potential of gastruloids in treating a wide range of diseases and disorders.

Prospective

The significance of gastruloids extends beyond their practical applications to their potential for addressing key challenges in human embryo research. First, in vitro models of specific cell types, such as primordial germ cell-like cells, face issues regarding their authenticity. PGCLCs in these models may not fully mimic their in vivo counterparts, the PGCs. Gastruloids, by providing a more organized environment for PGCLC development, facilitate interactions with other germ layers and the integration of signaling pathways. As a result, PGCLCs derived from gastruloids exhibit higher validity and reliability. Second, gene editing in human embryos, particularly for studying the function of specific genes, raises significant ethical concerns. Consequently, most gene editing studies in developmental biology rely on animal embryo models. The advent of human gastruloids provides an ethical alternative, allowing for the study of gene function and underlying signaling pathways without the ethical complications associated with direct embryo manipulation. Finally, maintaining human embryo ex vivo beyond the implantation stage remains a significant challenge, with embryos typically curling before reaching 14 dpf. However, advances in gastruloid culture techniques are making it possible to extend their culture period, positioning them as a complementary platform for studying early embryonic development alongside ex vivo cultured embryos.

In terms of gastruloid culturing, challenges related to structure formation and efficiency remain significant. Gastruloids often fail to fully recapitulate the complex structures observed during embryogenesis. For example, while progress has been made in generating neural tube-like structures in both human and mouse gastruloids, the development of more advanced structures remains elusive [55, 75, 101]. Additionally, most current gastruloid protocols lack extraembryonic tissue, which limits the ability to model key embryonic events and patterns, such as the formation of the yolk sac, primitive streak, and trilaminar disc. Recent studies have attempted to incorporate specific extraembryonic tissue cells into gastruloid models. However, no protocol has yet achieved the inclusion of all types of extraembryonic tissue. For instance, the mouse ETS model contains similar in vivo structures [68], while a hPSC-based model known as the peri-gastruloid has shown promise in recapitulating the formation of the amniotic and yolk sac cavities without external CHIR (a Wnt pathway activator) signals [76]. Despite these advances, the growth conditions of gastruloids can vary considerably between different cell lines and laboratory settings. Even within the same experimental setup, outcomes, such as elongation and marker expression, can differ. Therefore, developing a stable and efficient gastruloid protocol that can be applied across various cell lines and conditions is essential. Such a protocol would enable the use of gastruloids at a larger scale, beyond small-scale trials.

Moreover, current gastruloid systems typically induce the formation of specific structures, such as neural tubes, somitoids, or cardioids. A more universal protocol that allows for the formation of all embryonic structures remains elusive. Notably, human gastruloids with early cardiac structures and local neural plexus have been observed, indicating that the creation of a more comprehensive, compatible system may be possible [71]. Looking ahead, microfluidic systems could play a critical role in generating more complete gastruloids, providing not only an opportunity to study interactions within and between germ layers and early organ formation, but also an alternative source for patients with organ failure.

In addition to technical challenges, ethical concerns surrounding the use of gastruloids also warrant attention. Currently, human gastruloids derived from PSCs cannot fully recapitulate the functions of post-implantation stage embryos and are primarily used as tools for studying developmental mechanisms. However, improvements in human gastruloid culture techniques may enable the generation of more accurate models, particularly those that incorporate pre-neural or extra-embryonic tissue. At that point, the relationship between gastruloids and embryos will need to be reassessed, particularly regarding the ethical implications of developing in vitro tools that closely resemble the structure and function of early embryos. Moreover, as the capabilities of human gastruloids mature, ethical constraints may become more stringent. Recent studies have also indicated that gastruloids develop at a faster pace than real human embryos, raising questions about the appropriateness of the 14-day limit on embryo research [116]. These ethical issues need to be carefully considered as gastruloid technology progresses.

Recently, beyond mice and humans, gastruloid technology also applies in primate species such as monkey. While human models have ethical concerns and mouse models are morphologically and mechanistically different from primates, the monkey model offers an evolutionarily closer, ethically tractable system that mirrors human post-implantation development in scale, timing, and architecture. Current studies have successfully mimicked late-gastrulation events previously inaccessible in vitro, including formation of foregut and hindgut-like structures, onset of yolk sac-like hematopoiesis and PGCLC migration [117]. Future comparative studies across mouse, monkey, and human gastruloids will help distinguish universal mammalian principles from lineage-specific adaptations, marking significant progress towards a fully integrated, cross-species understanding of embryogenesis.

In summary, gastruloids function as an in vitro complementary platform for mammalian embryos, which provide a promising alternative for embryology research. Despite the potential challenges, gastruloids face vast future applications as related technology advances and more sophisticated gastruloid structures.

Author contributions

K.L., H.C., Y.C., Z.L., P.T. wrote this manuscript draft. H.Y., K.L., H.C., Y.C., Z.L., P.T. revised the manuscript. D.C. conceptualized the manuscript and provided funding support. All authors read and approved the final manuscript.

Funding

This work was supported by the National Key R&D Program of China awarded to DC (Grant No. 2024YFA1108100), National Natural Science Foundation of China awarded to DC (Grant No. 32270835), and Zhejiang Natural Science Foundation awarded to DC (Grant No. Z22C129553).

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare that they have no competing interests.