Background

The first attempts to model mouse embryonic development in vitro begun with a coincidental observation. When teratocarcinomas from neonatal mouse testes were analyzed, scientists noticed a structural similarity to the developing mouse embryo (Stevens, 1959). Hence, they called these tumors “embryoid bodies” (Pierce et al., 1960). It was further shown that embryoid bodies retained the tumor-forming capacity to form cell types ranging from cartilage to neural tissue, a multipotency that was later attributed to the embryonal carcinoma (EC) cells (Kleinsmith and Pierce, 1964). The use of EC cells to mimic embryonic development was rapidly replaced by embryonic stem cells (ESCs) following their isolation in 1981 (Evans and Kaufman, 1981). Pioneering studies have demonstrated self-organization potential of embryoid bodies to recapitulate, to a limited extent, gastrulation-like events and antero-posterior axis determination (ten Berge et al., 2008). However, this was not accompanied by axial morphogenesis.

More recently, gastruloids have taken the extent of self-organization potential of mouse ESCs (mESCs) to demonstrate that embryoid bodies could undergo axial morphogenesis (van den Brink et al., 2014). In this model, when treated with Wnt agonist CHIR99021, small aggregates of mESCs (~300 cells) were shown to break symmetry and demonstrated polarized T/Bra expression in a reproducible way. When cultured further, gastruloids elongated and established patterning along antero-posterior, dorso-ventral, and medio-lateral axes (Beccari et al., 2018). Moreover, the multi-axial patterning was linked to spatiotemporal activation of Hox gene clusters, a phenomenon that is conserved across many species (Santini et al., 2003). Such self-organization potential of mESCs, in the absence of any extraembryonic tissue, was remarkably similar to the developing post-occipital region of the mouse embryo; however, tissues mapping to anterior brain regions were completely absent in gastruloids.

Mechanical forces and extracellular matrix composition have significant influence on the developing mouse embryo (Hiramatsu et al., 2013). When gastruloids are embedded in a basement membrane substitute, they remarkably organize into structures bearing somites and a central neural tube-like tissue (Veenvliet et al., 2020). Even under these conditions, gastruloids fail to form any anterior neural tissues corresponding to the brain regions.

The epiblast is the domain that forms the embryo proper. However, contribution from extraembryonic tissues is required since development is halted in their absence (Donnison et al., 2005; Rodriguez et al., 2005). Studies accounting for this necessity have therefore established hybrid embryoid models, combining embryonic stem cells with trophoblast stem cells (TSCs) (Harrison et al., 2017; Rivron et al., 2018) and/or extraembryonic endoderm cells (XENs) (Sozen et al., 2018). More recently, mESCs were co-cultured with transdifferentiated TSCs and XENs to generate structures that almost completely recapitulate E8.5 embryos, including brain tissues (Amadei et al., 2022; Tarazi et al., 2022). However, the complexity of the tri-culture system and necessity of special equipment to grow them until late stages remain as the major limitations in generating synthetic embryos.

Previously, it was shown that embryos lacking extraembryonic Wnt source could still break symmetry and initiate gastrulation, suggesting an autonomous developmental potential of epiblast cells (Yoon et al., 2015). Moreover, overactivation of Wnt signaling in the epiblast was shown to deplete anterior neural progenitors in the favor of mesoderm derivatives; this phenotype could be rescued by inhibition of Wnt signaling (Osteil et al., 2019). Therefore, we reasoned that a similar trade-off could be happening in conventional gastruloids, owing to overactivation of Wnt signaling by CHIR99021 treatment.

Here, we report a new model system based on aggregation of mESCs to derive post-implantation epiblast-like structures (EPI aggregates). We formulated a serum-free epiblast-induction medium comprising Activin-A (Tgf-β agonist), Fgf2 (Fgf agonist), and knockout serum replacement, which promoted the acquisition of epiblast identity, followed by their spontaneous symmetry breaking and subsequent morphogenesis without any external Wnt stimulation. Moreover, inhibition of Wnt signaling during early stages of EPI aggregate formation helped to maintain anterior neural progenitors, which then committed to generate forebrain-, midbrain-, and hindbrain-like tissues. This protocol expands the cell type repertoire that can be generated within gastruloids.