Loading
Old version — revision 1
This is a fixed snapshot of Stem-cell-based embryo models, saved by Import as part of the initial corpus import. It is not edited and it is not updated; the article may have changed since.
Edit summary: Initial import of content/synthetic-embryos.md — the filesystem corpus, unchanged. Not an edit.
Structures grown by aggregating stem cells that reproduce stages of early embryonic development without fertilisation, and the regulatory categories they fall outside.
Stem-cell-based embryo models are structures assembled from cultured stem cells that reproduce features of early embryonic development without an egg, a sperm, or fertilisation. They are widely called synthetic embryos, a term nearly every researcher in the field objects to. None has produced a live animal in any species, and the most advanced human versions self-organise for a period roughly equivalent to the first two weeks of development before deteriorating. Their significance is that they fall outside the legal definition of an embryo almost everywhere, while resembling one closely enough that the exemption has become difficult to defend.
All of them start from pluripotent cells: embryonic stem cell lines, or Induced pluripotent stem cells made by factor-based reprogramming of adult tissue. The models divide by which parts of the conceptus they include.
Efficiency is low throughout. Typically a small percentage of aggregates form structures with correct morphology, the rest producing disorganised tissue, which is itself a reason to doubt that the successful ones are following the same developmental logic as an embryo rather than arriving at a similar appearance by another route.
Terminology"Synthetic" is doubly misleading: nothing is chemically synthesised, and the cells are ordinary human or mouse cells. "Embryo model" is the term used by the International Society for Stem Cell Research and in the UK code of practice. Popular coverage in 2022 and 2023 that described mouse models as embryos "created without eggs or sperm" prompted public corrections from several of the researchers involved.
The period between implantation and the fourth week of human gestation is the least understood phase of human development, for a straightforward reason: it happens inside a uterus, is inaccessible, and is when a large share of pregnancies fail. Donated IVF embryos are scarce, ethically constrained, and cannot legally be cultured past fourteen days in most jurisdictions.
Embryo models supply an alternative that is renewable, genetically tractable, and produced in quantity. Concrete applications include studying the causes of early miscarriage, which rise steeply with maternal age for reasons examined in Reproductive longevity; mapping the signalling that specifies the body axes; testing whether a drug or environmental exposure is teratogenic at a stage no other model reaches; and examining the origin of congenital anomalies. They also intersect with Embryo selection, since much of what a clinic infers from an embryo's morphology rests on developmental assumptions that models can now test directly.
The models are close cousins of Organoids, which self-organise from stem cells into tissue-like structures, and share both their strengths and their ceiling: real spatial organisation, real gene expression programmes, no vasculature, and limited maturity.
The fourteen-day limit on culturing human embryos originates in the 1984 Warnock report in the United Kingdom and was written into the Human Fertilisation and Embryology Act 1990.4 It was chosen because the primitive streak appears around day fourteen, after which twinning is no longer possible and individuation can be said to have begun. It was also chosen because at the time nobody could culture an embryo anywhere near that long, which made it a rule that cost nothing. The rule has since been adopted, formally or in practice, in most countries that fund human embryo research, and it is one of the few pieces of bioethical line-drawing that survived four decades intact.
Two developments undermined it. Human embryos were cultured to around thirteen days in 2016, making the limit binding for the first time. And embryo models arrived, which are not embryos under the statutory definitions and are therefore not covered by the limit at all. The UK Act defines an embryo as a live human embryo "where fertilisation is complete"; a structure that was never fertilised does not qualify.
The International Society for Stem Cell Research revised its guidelines in 2021, moving culture beyond fourteen days from a prohibited category into one requiring case-by-case specialised review and public consultation, and placing integrated embryo models under oversight while leaving non-integrated models such as gastruloids largely unrestricted.5 In 2024 a UK-based group published a voluntary code of practice for stem-cell-based embryo models with an independent oversight committee, an instance of the self-governance approach whose founding precedent is the Asilomar Conference on Recombinant DNA and whose weakness is the same: it binds only those who agree to be bound. Whether a research community should be able to relax a limit it originally proposed is the recurring objection, and it is the point at which arguments from the Precautionary principle enter the debate.
What would make a model an embryoProposals for a bright line fall into three families. A provenance test asks whether fertilisation occurred, which is clear but arbitrary. A morphological test asks whether the structure has the parts of an embryo, which is vague. A potentiality test asks whether it could develop into a person if transferred, which is the criterion most philosophers regard as morally relevant and the hardest to apply, since testing it directly requires the transfer nobody permits.
No embryo model has produced a live animal. Mouse models transferred to a uterus have implanted in some experiments and then failed. Whether this reflects a fixable technical deficit or a categorical difference between a model and an embryo is unresolved, and it is the single most consequential open question in the field, because a model that could develop to term would collapse the regulatory distinction entirely.
Fidelity is uneven. Single-cell transcriptomic comparisons with real primate and human embryos show that some cell types in models match their in vivo counterparts closely and others do not, with extraembryonic lineages generally poorer matches than embryonic ones. Timing is often wrong, and models frequently produce cells in the wrong proportions. The structures also lack a maternal interface, so nothing about implantation biology that depends on the endometrium can be modelled properly, which is a limitation shared with attempts at Ectogenesis and every version of the Artificial womb.
The moral-status question is unavoidable and unsettled. If moral status attaches to a developmental potential rather than to an origin, then a model that grows better will at some point acquire whatever status an embryo has, and the field will have arrived there gradually and without a decision. If status attaches to fertilisation, models never acquire it however lifelike they become. Neither position is fringe, and the disagreement is not empirical.
Downstream concerns are more tractable. Consent from the donors of the original stem cell lines rarely mentioned embryo models, since the lines predate them. Public trust is affected by terminology, and the field has been damaged by coverage it did not write; the reception of Human cloning in the late 1990s is the standing example of how a research programme can be defined by a word it did not choose. Transfer to a human or animal uterus is prohibited under every existing guideline, and the enforcement question resembles the one described in Governance of human genome editing: rules that bind funded academic laboratories in a handful of countries do not bind a well-resourced private effort elsewhere.
A further prospect complicates the picture. If In vitro gametogenesis succeeds, gametes could be made from stem cells and used to create real embryos, at which point the interesting entity is unambiguously an embryo and the model debate becomes moot for reproductive purposes. Some researchers have suggested models could instead be a source of tissue, connecting them to Lab-grown organs and interspecies chimera work, though nothing at the current level of organisation supports that.
The near-term question is narrower than the philosophy suggests: whether funders and journals will require embryo-model work to pass through review bodies designed for embryo research, and whether a voluntary code with no statutory backing survives contact with a laboratory that decides not to participate.
paperRivron, N. C. et al. "Blastocyst-like structures generated solely from stem cells." Nature, 2018. ↩
paperYu, L. et al. "Blastocyst-like structures generated from human pluripotent stem cells." Nature, 2021. Published alongside Liu, X. et al., "Modelling human blastocysts by reprogramming fibroblasts into iBlastoids," Nature, 2021. ↩
paperTarazi, S. et al. "Post-gastrulation synthetic embryos generated ex utero from mouse naive ESCs." Cell, 2022.↩Mouse cells only, and the paper's own title uses the word synthetic that most of the field now objects to.
reportWarnock, M. Report of the Committee of Inquiry into Human Fertilisation and Embryology. HMSO, 1984. ↩
reportInternational Society for Stem Cell Research. Guidelines for Stem Cell Research and Clinical Translation, 2021, subsequently revised in its provisions on stem-cell-based embryo models.↩Guidelines written by the research community they govern; they carry no statutory force and bind only laboratories that choose to follow them.