Shifting early embryology paradigms: Applications of stem cell-based embryo models in bioengineering.

The use of bioengineering technologies to mimic the maternal environment is already underway [ 73 , 75 , 76 ]. Technologies to model the endometrial tissue of the maternal environment continue to improve as unbiased -omics-based approaches have been applied to characterize the cell types and transcriptional signatures of the endometrium and embryo-maternal interface [ 77 ], as reviewed at length here [ 70 ]. Further, endometrial organoids have been developed and have already been used in co-culture with natural embryos [ 78 – 80 ]. In recent years, microfluidics have been leveraged [ 76 , 81 – 85 ] as a method for modelling maternal tissues, including a recent study [ 76 ] that reported a micro-engineered, vascularized endometrium-on-a-chip that undergoes angiogenesis and hormonal responses like the natural phases of the menstrual cycle, circumventing the lack of vascularization issue observed in current endometrial organoids. Of note, a recent publication has offered a novel synthetic vascularization approach, enabled by a method they coin ‘soft 3D microfluidics’ which may be helpful to address needs for vascularizing endometrial organoids as well [ 86 ]. In combining such technologies with integrated embryo models, future platforms could assay embryo-maternal interface processes at a high-throughput scale. Since in vivo human investigations are impractical and unethical, this type of technology could be extremely useful to interrogate maternal pathologies that have adverse outcomes in pregnancy for unknown reasons in humans ( Figure 2 ). Although there are many maternal conditions that can influence fertility and/or embryo development and survival, below we propose a few conditions with unknown pathophysiology as examples to highlight the potential for targeted investigation towards future therapeutics. Two maternal uterine pathologies – endometriosis and uterine fibroids – are strongly associated with infertility for reasons that are not well understood [ 87 – 89 ], although there is evidence that implantation failure may occur due to structural and/or biochemical factors associated with these conditions [ 90 – 92 ]. The mechanisms underlying these associations are unknown but could be probed in detail from multiple angles (structurally, biochemically, epigenetically, metabolically) using a high-throughput integrated embryo platform mimicking the embryo-maternal interface as proposed above. Additionally, studies have identified an association between obesity and recurrent pregnancy loss [ 93 , 94 ], but underlying mechanisms remain unidentified. All three of these diseases could be modeled in a 3D embryo-maternal interface platform at high-throughput with human-specific tissues using the emerging technologies described in this review.

Integrated stem cell-based embryo models across multiple stages of development in both mice and humans showcase the inclusion of both embryonic and extraembryonic tissues to gain insights about intercellular crosstalk and to recapitulate the 3D spatial architecture reminiscent of the native embryo, as well as its developmental potential. The incorporation of bioengineering approaches in conjunction with classic developmental biology techniques has enormous potential to allow us to explore deeper questions about our own earliest stages of life through controlled embryo modelling. The existing abundance of approaches to model the maternal environment with microfluidics and organoids opens a wealth of possibilities for integration of extraembryonic tissue-containing stem cell embryo models to understand implantation and the embryo-maternal interface. There is currently a significant gap in our understanding of maternal pathologies that have adverse outcomes in fertility and pregnancy for unknown reasons, which represents a currently unmet need in this field. However, rapidly progressing technologies in both stem cell-based embryo modeling and platforms to model the maternal environment offer enormous potential towards understanding the mechanisms underlying these pathologies within the coming years.

The integrated stem cell-based embryo models that recapitulate the post-implantation developmental period so far have been reported in the mouse context, which we summarize below: The beginning of early post-implantation development in mice is defined by epiblast polarization and formation of the pro-amniotic cavity, followed by symmetry breaking. To recapitulate these events in a stem cell-based embryo model system, mESCs and mTSCs were first co-embedded in a hydrogel, mimicking the extracellular basement membrane – required to cue polarization and pro-amniotic-like cavity formation. This approach resulted in multicellular post-implantation-like structures called “ ETS embryos ”, representing the aggregation of the two cell types – embryonic (E) and trophoblast (T) stem (S) cells [ 56 ]. Follow-up studies replaced the use of hydrogel with the inclusion of mXEN [ 57 ] or mESCs induced (‘i’) to express the extra-embryonic fate transcription factor Gata4 [ 58 ] and led to models called “ ETX/iETX embryos ,” referring to mESC (E), mTSC (T), and mXEN (X) or Gata4 -induced (i) mESCs as a means of creating induced XEN cells. These ETX/iETX models are akin to the natural early post-implantation embryo and able to undertake early gastrulation events, such as the epithelial–mesenchymal transition, and consequential nascent mesoderm and definitive endoderm specification ( Figure 1 ). Despite their advantages, these ETS/ETX/iETX embryo platforms have a relatively low efficiency of ~20% owing to their aggregation design that relies on interactions of multiple stem cells in the correct number and ratio, creating a need to improve reproducibility. Bioengineering-inspired approaches potentially allow for higher degrees of control, opening the possibility to further developmental progression. A recent study [ 59 ] leveraged this principle by using hydrogel microwells to generate mESC-aggregates separately from mTSC-aggregates and then combined both aggregates in low-attachment wells to allow them to merge and undergo post-implantation morphogenesis. The resultant hybrids were called “ EpiTS embryoids ,” representing the combination of epiblast-like aggregates (Epi) with trophoblast aggregates (TS) to form embryoids. These models have a symmetry-breaking efficiency of up to 70%, undergo axial morphogenesis, and develop cell types found in the early midbrain and hindbrain. Building upon mouse post-implantation embryo models, which combine the three stem cell types (mESC, mTSC, mXEN), recent reports [ 60 – 62 ] have leveraged transcription factor programming and ex vivo gas-controlled roller culture methods to develop advanced integrated embryo models of gastrulation through post-gastrulation stages ( Figure 1 ). Notably, some mouse and human non-integrated high-throughput designs also demonstrated significant aspects of post-gastrulation development, such as somitogenesis or neural tube formation [ 63 – 68 ]. These recent integrated models, however, display remarkable co-features of the brain, neural tube, heart, foregut, somite, allantois, primordial germ cells, and yolk sac in space and time [ 60 – 62 ]. In these models, parental mESCs were genetically reprogrammed to produce mTSCs (via Cdx2 overexpression) and mXEN cells (via Gata4 overexpression), allowing the co-aggregation of all three cell types in microwells with the same genetic background [ 61 , 62 ]. Importantly, it was noted that the degree of Cdx2 expression in the induced mTSCs was a key determining factor in the overall self-assembly and development of these post-gastrulation models, citing that low Cdx2 expression resulted in almost no correctly formed aggregates [ 61 ]. This example highlights the importance of proper extraembryonic tissue contribution for the self-assembly of spatially correct development in integrated embryo models. Ensuring the genetic compatibility between the three starting cell types likely plays a substantial role in accomplishing advanced development than previous integrated embryo models. However, all three studies reported that only cells of certain genetic backgrounds could self-assemble and develop in these models, and the overall efficiency was extremely low. This limitation opens up space for future investigations to determine distinguishing features of genetic background in cell lines with respect to their developmental potential. Additionally, it appears that adjusted gas levels in the roller culture incubation was another key in overall success of these platforms. It was found that delivering gas mixtures at lower pressures (0.5 psi rather than 6.5 psi) than previously published values [ 69 ] was highly effective in promoting in vitro development. As extraembryonic tissues are critical for gas exchange between the embryo and maternal tissues to ensure development in vivo , these novel platforms highlight the impact of integrated embryo modeling in potentially achieving complex developmental features. Collectively, these integrated post-implantation embryo models represent significant strides in the field and open doors for translational approaches, such as developmental disease modeling – though such future directions will require increasing reproducibility in the model systems before translating this exciting technology to practice. If also developed at tractable levels in human systems, these models can be extremely valuable platforms to study mechanisms of normal development as well as embryonic/extraembryonic defects.

During early mammalian development, the successful implantation, patterning, and nutrition of the embryo depend on the proper establishment of the extraembryonic lineages in coordination with the pluripotent epiblast tissue. Pluripotency describes the developmental potential of epiblast cells to give rise to all cells of the future body [ 1 , 2 ]. Epiblast cells initially possess a “naïve” pluripotency state, which is lost upon implantation as the cells transition through a “formative” state and shortly after that become “primed” in preparation for gastrulation and subsequent organogenesis (pluripotency states discussed at length for both mouse and human in: [ 3 , 4 ]). The derivation and culture of embryonic or pluripotent stem cells (ESCs/PSCs) in vitro have been crucial for advancing our understanding of epiblast cells [ 4 – 9 ]. Recent breakthroughs in culture conditions have expanded the potential of PSCs to recreate embryonic epiblast states that span the pre- to post-implantation transition, and ongoing work continues to refine in vitro conditions, particularly at peri-implantation stages [ 4 ]. Proper epiblast development relies on critical crosstalk and support from the peripheral extraembryonic tissues, the trophectoderm (TE) and primitive endoderm (PE, or known as hypoblast in primates), which form the placenta and yolk sac, respectively [ 2 ]. Stem cell lines have successfully been established for these two extraembryonic tissues in mice: trophoblast stem cells (TSCs) [ 10 ] and extraembryonic endoderm (XEN) cells [ 11 ]. Multiple promising approaches have been developed to derive human extraembryonic stem cell lines, although some limitations persist. Current protocols include derivation from blastocysts [ 12 ], directed differentiation from hPSCs [ 13 , 14 ] or transcription factor reprogramming [ 15 – 17 ]. Traditionally, stem cells have been studied in 2D monolayer cultures. Although highly valuable, this approach lacks the crucial resolution of a developing embryo’s three-dimensional (3D) spatial organization. It has been envisioned that the development of methods to enable the assembly and differentiation of stem cell derivatives of mammalian embryo could allow for controlled exploration of embryonic organization and patterning using cell lines rather than embryos. Indeed, a recent surge in publications describes stem cell-based in vitro models with an ever-increasing similarity to the natural embryo and its developing organs in both mouse and human systems [ 18 , 19 ]. These models have been particularly valuable in understanding early human development, owing to the ethical and technical difficulties associated with human embryo research [ 20 ]. Additionally, bioengineering technologies have enhanced approaches to studying developmental questions by allowing scientists to control and manipulate biological systems to a higher degree [ 21 ]. Techniques such as 3D printing and bioprinting [ 22 ], microfluidics [ 23 , 24 ], and optogenetics [ 25 – 27 ] have enabled the design of highly complex and precise organoid structures and organ-on-a-chip devices with substantial control. These models promise to improve our understanding of the molecular mechanisms driving embryo-intrinsic processes, such as axis formation associated with the pre-and peri-gastrulating embryo. They may also facilitate a more comprehensive understanding of embryo-maternal interaction by producing models that include maternal components. In this review, we exclusively focus on the development of stem cell-based mammalian embryo models, which present both embryonic and extraembryonic tissues within the platform, termed “ integrated embryo models ” [ 28 ]. We discuss these recent strategies with a specific focus on how embryonic and extraembryonic lineages co-evolve to support morphogenesis in 3D. We further highlight unmet challenges and strategies to validate functionality. We also explore the feasibility and potential impact of using complementary platforms where the embryo-maternal interface can be modeled via bioengineering approaches.

During in vivo development, extraembryonic tissues are essential for generating the embryo-maternal interface, the physical and functional connection and the source of reciprocal signaling pathways between the developing embryo and the maternal endometrium [ 70 ]. Due to the in utero inaccessibility of mammalian embryo, little is known regarding the process of implantation and the cellular and molecular interactions at this interface. Particularly in humans, numerous limitations associated with the acquisition of relevant tissue material for study create a significant gap in knowledge of the mechanisms mediating the interconnection of the embryo and the mother [ 71 ]. Further, the biology of the placental and embryo-maternal interface varies significantly even among primates [ 72 ], restricting the translation of non-human species towards clinically relevant outcomes for human health. This area of research offers a number of significant novel avenues to investigate the intricate signals that occur at the interface, as well as allowing for interventions that are specific to the mother, the fetus, and the placenta. Thus, comprehensive model systems developed through bioengineered co-culture environments which reproduce key features of the embryo-maternal niche set the stage for the next big breakthroughs in the field, with the exceptional importance for human-centric stem cell-based models. Recent efforts in this field utilized mouse blastocysts in engineered biomimetic platforms to visualize key features of the implantation niche, shedding light on important cues of trophoblast invasion and the first embryonic interactions with the maternal vasculature [ 73 , 74 ]. It was shown that the use of customizable synthetic hydrogels in a microfluidic recapitulated key biomechanical features of the decidua, allowing for the discovery of novel mediators of implantation, such as Pdgf signaling cues as a means to promote the establishment of contacts between the invasive trophoblast and the endometrial endothelial cells [ 73 ]. The presence of extraembryonic tissues within integrated embryo models offers a future possibility to recapitulate and visualize these critical processes in vitro ( Figure 2 ). Previously, limited numbers of IVF-derived human blastocysts have been ex vivo co-cultured with endometrial cells in various platforms to gain insights on maternal environment interface [ 75 ]. Most recently, human blastoids demonstrated a similar capacity to attach to human endometrial epithelium on 2D surface, simulating implantation-like dynamics in vitro [ 45 ]. Neither aggregates that lack trophoblast nor trophospheres (lacking an epiblast) could initiate an implantation-like event in vitro , highlighting the importance of locally provided signals both from the embryonic and extraembryonic lineages [ 45 ]. Further work leveraging 3D bioengineering-based approaches with stem cell-based embryos will undoubtedly lead to an abundance of new understandings about this complex and critical tissue-tissue interactions, particularly in the human context [ 70 ]. Moreover, advancements in gene editing, developmental drug and toxicity testing will likely result in target-specific treatments with promising results in pre-clinical maternal and embryonic disorder models (discussed in the next section).