A proper understanding of early human development is crucial if we are to improve assisted reproductive technologies and prevent pregnancy loss and birth defects. However, studying early development is a challenge — few human embryos are available, and research is subject to considerable ethical and legal constraints. The emergence of techniques that use cells cultured in vitro to construct models of mammalian embryos therefore opens up exciting opportunities1. Two papers in Nature now make key advances in this field, showing that human embryonic stem cells2 or cells reprogrammed from adult tissues2,3 can be induced to self-organize in a dish, forming structures that resemble early human embryos. This is the first integrated human embryo model containing cell types related to all the founding cell lineages of the fetus and its supporting tissues.
In mammals, a fertilized egg undergoes a series of cell divisions over the first days of development, leading to the formation of a structure called the blastocyst. The blastocyst contains an outer cell layer called the trophectoderm, which surrounds a cavity containing a cell cluster called the inner cell mass (ICM). As the blastocyst develops, the ICM becomes segregated into two adjacent cell populations — the epiblast and the hypoblast (known as the primitive endoderm in mouse embryos). The blastocyst then implants into the uterine tissue, setting the stage for an event called gastrulation, in which epiblast cells give rise to the three basic cell layers that will form the entire fetus. The trophectoderm goes on to form most of the placenta, and the hypoblast forms some layers of a structure called the yolk sac, which is required for early fetal blood supply.
The first in vitro models to recapitulate blastocyst formation using cultured cells (structures known as blastoids) used mouse stem cells corresponding to the stem cells found in the epiblast, trophoblast and primitive endoderm in the mouse blastocyst4–6. However, the generation of similar blastoids from human cells has not been achieved until now1. Previous models of early human development used human stem cells developmentally similar to post-implantation, pre-gastrulation epiblast cells7–9. As such, although they could recapitulate some stages of post-implantation human development, they lacked lineages associated with the trophectoderm, hypoblast or both.
In the current papers, Yu et al.2 and Liu et al.3 describe human blastoids. The key to these technological breakthroughs seems to have been twofold: first, the use of cells representative of lineages in the human blastocyst; and second, optimization of culture protocols.
Yu et al. started with either human embryonic stem cells, which are derived from human blastocysts, or induced pluripotent stem cells, which are generated from adult cells. Importantly, both of these types of stem cell are developmentally similar to epiblast cells in the blastocyst, and can also give rise to lineages related to the trophectoderm and hypoblast. By contrast, Liu et al. reprogrammed adult skin cells called fibroblasts to form a mixed cell population that contained cells with gene-expression profiles similar to those of cells of the epiblast, trophectoderm and hypoblast. As in the mouse blastoid protocols4–6, both approaches involved seeding the cells in 3D culture dishes called Aggrewell plates, and treating them with liquid growth medium that contained chemical factors to control the signalling activities needed for blastocyst development (Fig. 1). Yu and colleagues treated the cells with two different types of culture medium in sequence, to promote differentiation of the cells into lineages representative of the trophectoderm and hypoblast.
Both groups found that human blastoids emerged after 6–8 days of culture, with a formation efficiency of up to almost 20%, comparable to the efficiencies of the mouse blastoid protocols4–6. The human blastoids were of a similar size and shape to natural blastocysts, with a similar total number of cells. They contained a cavity and an ICM-like cluster.
Detailed characterization of the blastoids (including genome-wide expression analysis and comparisons with human embryo data) showed that their cell lineages share molecular similarities with those of the pre-implantation human blastocyst. The spatial organization of the epiblast-, trophectoderm- and hypoblast-related lineages is consistent with that found in pre-implantation human embryos. The groups also demonstrated that the blastoid cells have key properties of blastocyst lineages — cells isolated from the blastoids could be used to generate various stem-cell types. Yu et al. showed that, if these stem cells were transplanted into mouse blastocysts, they gave rise to cells that could integrate with the corresponding mouse lineages in the mouse embryo.
Next, the researchers analysed further development of the blastoids using an established assay that mimics implantation into the uterus in culture dishes. Like blastocysts, when blastoids were grown in this assay for four to five days, some attached to the culture dish and continued to develop. In a portion of these attached blastoids, the cell lineage representative of the epiblast became reorganized into a structure enclosing a central cavity — reminiscent of the pro-amniotic cavity, which forms in the epiblast of post-implantation blastocysts. And in some blastoids, the trophectoderm-related cell lineage spread out and showed signs of differentiation into specialized placental cell types. Yu et al. also observed a second cavity in the hypoblast-related cell lineage in some blastoids, akin to the yolk-sac cavity.
Together, the groups’ data demonstrate that human blastoids are promising in vitro models of pre-implantation and early post-implantation blastocyst development. However, there are notable limitations to overcome. For example, development of the blastoids is inefficient, and varies between cell lines produced from different donors, and between experimental batches. In addition, the three lineages seem to develop at slightly different rates in single blastoids, and development of blastoids in the same dish seems unsynchronized. Spatial organization of the hypoblast-related lineage in blastoids remains to be improved. Furthermore, the blastoids contain unidentified cell populations that do not have counterparts in natural human blastocysts.
Another challenge is that development of the blastoids is limited in post-implantation stages, unlike in mouse blastoids4–6. Further optimization of culture and experimental conditions will be needed to improve post-implantation-stage culturing of human blastoids in vitro, up to the equivalent of 14 days in vivo. Strict ethical rules prevent the culturing of human embryos past this stage, when structures associated with gastrulation begin to appear. Three-dimensional systems for culturing human blastocysts10, which effectively promote post-implantation development, might help to improve our ability to culture blastoids up to this limit, by maintaining the normal 3D tissue architecture and spatial relationships between the different cell lineages in the blastoids.
Human blastoids are the first human embryo models that are derived from cells cultured in vitro and that have all the founding cell lineages of the fetus and its supporting tissues. As protocols are optimized, these blastoids will more-closely mimic human blastocysts. This will inevitably lead to bioethical questions. What should the ethical status of the human blastoids be, and how should they be regulated? Should the 14-day rule be applicable? These questions will need to be answered before research on human blastoids can proceed with due caution. To many people, the study of human blastoids will be less ethically challenging than the study of natural human blastocysts. However, others might view human blastoid research as a path towards engineering human embryos. Thus, the continuous development of human embryo models, including human blastoids, calls for public conversations on the scientific significance of such research, as well as on the societal and ethical issues it raises.
Nature 591, 531-532 (2021)