Stem Cell-Based Embryo Models as Pre-Clinical Toxicity Platforms

Translational · Collaboration

Summary

A central and unresolved problem in drug development is predicting whether a therapeutic compound will harm an embryo during early development. Traditional pre-clinical safety studies rely on rodent and rabbit animal models, but for a growing class of modern therapeutics, these models are fundamentally unreliable — not because the experiments are poorly designed, but because the biology differs too much between species. This translational gap affects patients directly: promising drugs may be unnecessarily delayed or restricted, or conversely, human-specific risks may go undetected until clinical exposure.

We are developing a new pre-clinical platform that uses human stem cell-based embryo models (SCBEMs) — specifically human gastruloids — as a biologically faithful, ethically acceptable, and scalable system for developmental toxicity assessment. This work sits at the intersection of developmental biology, translational medicine, and regulatory science, and is carried out in collaboration with the laboratory of Drs. Li Chai and Jun Liu at Brigham and Women’s Hospital, Boston.

Why animal models are failing for a new class of drugs

The problem is particularly acute for targeted protein degraders — a rapidly expanding class of small molecules that harness the cell’s own degradation machinery to eliminate disease-causing proteins. More than 30 cereblon (CRBN)-based degraders are currently in clinical trials, targeting diseases ranging from cancer to sickle cell disease and β-thalassemia. CRBN is the same molecular target through which thalidomide caused devastating birth defects in the 1950s and 60s: thalidomide-class drugs bind CRBN and trigger the degradation of developmental transcription factors including SALL4 and TP63, leading to limb and organ malformations. This teratogenicity is uniquely human: the downstream targets of CRBN differ substantially between species, meaning the molecular machinery that makes these drugs dangerous in human development is simply absent in rodents and rabbits.

Because standard embryo-fetal development studies in these species cannot reliably flag human CRBN teratogenicity risk, a 2025 consensus document by the IQ Consortium — representing the pharmaceutical industry’s leading toxicologists — concluded that no validated developmental toxicity paradigm currently exists for this entire drug class [1]. This is a systemic regulatory gap affecting every CRBN-based therapeutic program in development.

Human gastruloids as a solution

Human gastruloids are self-organized, three-dimensional aggregates of human embryonic or induced pluripotent stem cells that reproducibly recapitulate the earliest steps of human embryogenesis: symmetry breaking, axial elongation, and the specification of the primary germ layers — mesoderm, endoderm, and neuroectoderm — within 72 hours [6,11]. Critically, they do so using human biology, making them sensitive to human-specific teratogenic mechanisms that animal models miss.

A landmark study by Mantziou et al. demonstrated that human gastruloids display thalidomide-specific sensitivity at sub-micromolar concentrations, mirroring human teratogenicity, while mouse gastruloids exposed to the same compound show no effect [8]. Subsequently, Huntsman et al. validated a gastruloid-based developmental toxicity assay against 24 reference compounds from the ICH S5(R3) toxicology guideline, demonstrating sensitivity comparable to in vivo rodent data for 25 of 26 drugs tested [12]. A recent review highlighted gastruloids as a leading path toward humanized 3D developmental toxicity platforms [13].

These advances coincide with a significant regulatory shift. The FDA Modernization Act 2.0 (2022) removed the blanket requirement for animal testing prior to human clinical trials, explicitly creating a pathway for validated human-relevant alternatives — referred to as New Approach Methodologies (NAMs) — to be accepted in regulatory submissions [10]. Regulatory agencies and the pharmaceutical industry are now actively seeking validated human-relevant platforms that can complement or partially replace traditional animal embryotoxicity studies.

Preliminary data and planned approach

In collaboration with the Chai lab at Brigham and Women’s Hospital, we have established robust human gastruloids from hESCs and patient-derived hiPSCs as a developmental toxicity readout platform. As an initial proof-of-concept, we tested SH6 — a novel non-IMiD ZBTB7A protein degrader developed by the Chai lab for β-hemoglobinopathy treatment [5,7] — in wild-type human gastruloids. SH6 induces dose-dependent disruption of gastruloid morphology and germ-layer marker expression at low micromolar concentrations, demonstrating that the assay is sensitive and quantitative, and that it can detect developmental effects of CRBN-engaging compounds in a human cellular context.

A central next step is implementing a paired wild-type / CRBN-knockout (CRBN-KO) comparison strategy. By generating isogenic CRBN-deficient hESC and hiPSC lines and testing candidate compounds side-by-side with their wild-type counterparts, we will be able to mechanistically dissect whether observed developmental toxicity is CRBN-dependent — and therefore specifically human-relevant — or arises through off-target pathways. This distinction directly addresses what regulators and drug developers need answered, and goes beyond existing gastruloid toxicity studies that lack this mechanistic control.

Where we’re headed

Beyond the CRBN degrader application, we see gastruloid-based developmental toxicity screening as a broadly applicable framework. The same paired isogenic strategy — testing compounds in cells with and without the relevant molecular target — can in principle be applied to any drug mechanism where human-specific developmental toxicity is a concern. Directions we are pursuing or exploring include benchmarking the platform against established reference teratogens and non-teratogens; profiling compounds from distinct chemical classes side-by-side to develop quantitative safety metrics that could inform go/no-go decisions in early drug development; and working toward engagement with regulatory agencies to explore formal qualification of gastruloid-based assays as complementary tools in developmental and reproductive toxicology workflows. Longer-term, we are interested in extending the platform to therapeutic areas beyond degraders — including any program where standard animal studies are likely to underestimate or miss human-relevant risk.

In plain terms: Some of today’s most promising new medicines work by instructing cells to destroy disease-causing proteins — a strategy called “targeted protein degradation.” These drugs are showing remarkable results for blood diseases, cancer, and more. But there’s a serious problem: the same molecular machinery these drugs use is also critical for normal embryo development. Standard safety tests in rats and rabbits can’t detect this risk, because the underlying biology differs too much between other species and humans. We are building a new type of test using human stem cells that form tiny, embryo-like structures in a dish. Because these structures use real human biology, they can detect the kind of developmental harm that animal tests miss — and do so early, cheaply, and without requiring human embryos. Our goal is to help drug developers identify safer compounds earlier, and ultimately to help patients access effective medicines with greater confidence in their safety.

Interested in collaborating?

We are actively seeking collaborations with academic laboratories and industry partners who are interested in early developmental safety assessment of therapeutic compounds. This includes groups developing CRBN-engaging or other targeted protein degraders, as well as programs in any therapeutic area where human-specific developmental toxicity is a concern and standard animal models may not provide adequate safety information. If you would like to discuss a potential collaboration, please get in touch.

Selected references

[1] Loberg LI, Proctor WR, Burdick AD, et al. Nonclinical teratogenicity safety assessment of CRBN-engaging targeted protein degraders: IQ Consortium points to consider. Regul Toxicol Pharmacol. 2025;158:105793.

[5] Liu J, Shen Z, Park SY, et al. Discovery of a small-molecule non-IMiD degrader of ZBTB7A for the treatment of β-hemoglobinopathies. bioRxiv. 2025. doi:10.1101/2025.09.17.676148

[6] Moris N, Anlas K, van den Brink SC, et al. An in vitro model of early anteroposterior organization during human development. Nature. 2020;582:410–415.

[7] Liu BH, Liu M, Radhakrishnan S, et al. Targeting transcription factors through an IMiD-independent zinc finger domain. EMBO Mol Med. 2025;17(6):1393–1416.

[8] Mantziou V, Baillie-Benson P, Jaklin M, et al. In vitro teratogenicity testing using a 3D, embryo-like gastruloid system. Reprod Toxicol. 2021;105:72–90.

[10] FDA Modernization Act 2.0, Pub. L. No. 117-328, §3209 (2022).

[12] Huntsman MC, Kurashima CK, Marikawa Y. Validation of a mouse 3D gastruloid-based embryotoxicity assay in reference to the ICH S5(R3) guideline. Reprod Toxicol. 2024;125:108558.

[13] Branco MA, Nunes TC, Cabral JMS, Diogo MM. Developmental toxicity studies: the path towards humanized 3D stem cell-based models. Int J Mol Sci. 2023;24(5):4857.