Next-Generation Stem Cell-based Human Embryo Models Through Bioengineering

The study of early human embryogenesis, particularly the post-gastrulation trunk development, is constrained by significant technical limitations and ethical guidelines. Stem cell-based in vitro models have emerged as a powerful alternative, but current approaches suffer from two major drawbacks including high variability and an inherent conflict between achieving tissue complexity and maintaining experimental modularity. This PhD thesis aims to overcome these limitations by developing two distinct, next-generation human embryo models based on bioengineering principles. First, to address the issue in conventional human gastruloid protocol, we investigated the pluripotency mismatch between robust mouse and variable human protocols. By resetting hPSCs to a naive-like (RSeT) state, we demonstrate that the confounding 2D CHIR pre-treatment can be omitted in favor of a simplified, more robust 3D-only protocol that is biologically analogous to the mouse system. These RSeT gastruloids exhibit superior anteroposterior organization, develop advanced structures such as an endodermal gut tube, and can be cultured for extended periods. Mechanistically, our scRNA-seq analysis reveals that the pre-pulse phase constitutes a critical, cell-line-specific formative transition where aggregates mature to a competent identity. Second, to resolve the conflict between complexity and modularity, we developed a programmable, bottom-up bioengineering platform. We created Human Assembled Trunk Organoids (HATOs) by pre-specifying and subsequently combining distinct progenitor modules, specifically neuromesodermal progenitors (NMPs), notochord progenitors (NOTOs), and primitive streak-like cells (PS). This modular assembly strategy enabled the generation of a spectrum of distinct organoid types with tunable composition. These modules fuse and self-organize into complex, multi-lineage organoids containing a neural tube, segmented somites, a notochord, a gut tube, and a cardiac compartment. Crucially, the HATO platform remains fully modular, enabling the precise dissection of tissue-tissue interactions. As a validation, we show that the notochord module successfully induces ventralization in the adjacent neural tube, an event absent in simpler models. Furthermore, the HATO system permits targeted signaling perturbations on individual progenitor modules. Together, this thesis provides two complementary and powerful platforms that offer greater robustness, fidelity, and experimental control, enabling a more precise, bottom-up engineering approach for modeling human trunk development and early organogenesis.