In vitro gametogenesis

In vitro gametogenesis is the production of functional eggs or sperm from pluripotent stem cells in culture, without a gonad. In mice the full female cycle has been closed: embryonic stem cells have been converted to oocytes entirely in a dish, fertilised, and carried to healthy pups whose own cells could start the cycle again. In humans, the same protocols stall well before meiosis. The gap between those two sentences is the whole subject, and it has not narrowed as quickly as the mouse results led people to expect.

§How it works

Germ cells are set aside early in development and then follow a schedule no other lineage repeats: they erase most of their DNA methylation, re-establish sex-specific imprints, halve their chromosome number through meiosis, and, in the female line, grow enormously while stockpiling the maternal factors an embryo needs before its own genome switches on. Reconstituting that in culture means reproducing each stage in order.

The standard route has four steps.

1. Pluripotency. Start from embryonic stem cells or from Induced pluripotent stem cells made by reprogramming a somatic cell such as a skin fibroblast or a blood cell.
2. Germline specification. Push the cells through an epiblast-like intermediate and then, with BMP signalling and the right transcription factors, into primordial germ cell–like cells. In humans, SOX17 rather than the mouse's BLIMP1-first logic turns out to be the critical specifier, one of several places where the two species diverge.¹
3. Gonadal environment. Aggregate the germ cells with embryonic gonadal somatic cells to form a reconstituted ovary or testis. The somatic niche supplies the signals that license meiosis; germ cells will not do it alone.
4. Maturation. Grow the resulting follicles or spermatogenic cells to a fertilisable gamete. This is the step that has never been completed for humans.

Step three is the hidden bottleneck. Mouse experiments use somatic cells dissected from mouse embryonic gonads, which is not an option in humans; the field therefore also needs to make human ovarian and testicular somatic cells from stem cells, a problem being worked on in parallel and not yet solved to the point of supporting complete human folliculogenesis.

§Development history

1. 2011
   Mouse sperm
   Hayashi and Saitou reconstitute germ-cell specification in culture; the resulting cells produce fertile sperm after transplantation into mouse testes.
2. 2012
   Mouse eggs
   The same approach yields oocytes, though maturation still requires transplantation of reconstituted ovaries into a live animal.
3. 2015
   Human germ cells specified
   Two groups independently derive human primordial germ cell–like cells and identify SOX17 as the key specifier, showing the human pathway differs from the mouse one.
4. 2016
   The mouse cycle closes
   Hikabe and colleagues complete the entire female germline cycle in vitro and obtain fertile offspring, at low efficiency.
5. 2018
   Human oogonia
   Human germ cell–like cells are matured to oogonia inside xenogeneic reconstituted ovaries built with mouse gonadal somatic cells, then stop short of meiosis.
6. 2023
   Eggs from male cells
   Hayashi's group reports functional oocytes derived from male mouse cells, involving loss of the Y and duplication of the X, with pups born from bipaternal embryos.
7. 2020s
   Companies form
   Conception Biosciences, Gameto and others begin funding human work; none has reported a mature human gamete.

The programme is largely the work of two connected Japanese laboratories, Mitinori Saitou's at Kyoto and Katsuhiko Hayashi's at Kyushu and later Osaka, both descended from the reprogramming tradition established by Shinya Yamanaka in the same city. The 2011 and 2012 papers established that pluripotent cells could be routed into the germline reliably.² The 2016 paper by Orie Hikabe and colleagues was the milestone: oocytes generated wholly in culture, fertilised, transferred, and producing pups that grew into fertile adults.³ Efficiency was poor, with only a small fraction of the oocytes produced proving competent, but the pathway was closed.

The 2023 result attracted the most attention outside the field. Cultured male mouse cells lost the Y chromosome at some frequency, and the resulting X-monosomic cells were treated with a compound that promoted X duplication, yielding XX cells that could be taken through oogenesis.⁴ Embryos made from these eggs and ordinary sperm produced live pups. The yield was on the order of one percent of transferred embryos, which is the number that most reporting omitted.

§The human gap

Human germ cell–like cells are now made routinely. Getting them further has proved slow. Culture in xenogeneic reconstituted ovaries carries them to the oogonial stage, at which point they have undergone extensive epigenetic reprogramming but have not entered meiosis.⁵ Subsequent work has extended the reprogramming and expansion phases and clarified how the human germline erases and rebuilds its methylation marks,⁶ but as of 2026 no laboratory has publicly reported a human oocyte or spermatozoon made in vitro, and none has reported fertilisation with one.

Several things make humans harder. Human oogenesis takes months rather than weeks, and the follicular growth phase is correspondingly long and poorly reproduced in culture. Human germ cells specify differently, so mouse protocols do not transfer. And the ethical and legal constraints on obtaining human embryonic gonadal tissue restrict the shortcut that made the mouse work possible.

Distinguish the mouse claim from the human claim

Every complete demonstration of in vitro gametogenesis is in mice. Reports described in press coverage as human breakthroughs have concerned earlier stages of the pathway, ovarian support cells used to improve conventional egg maturation, or nuclear-transfer approaches that are not gametogenesis from pluripotent cells at all.

§What it would change

If human IVG worked, its effects would run through several other technologies covered here.

Infertility. Anyone with somatic cells could in principle produce gametes, including people whose gonads were removed or damaged by cancer treatment, people with premature ovarian insufficiency, and men with non-obstructive azoospermia. This is the application the companies name and the one most likely to reach a clinic first. It addresses gametes and not gestation, so absolute uterine factor infertility would still require Uterus transplantation or gestational surrogacy: every Artificial womb built so far takes over a pregnancy already underway, and complete ectogenesis has no demonstration in any mammal.

Ovarian aging. Oocyte quality declines steeply with maternal age, the central fact of Reproductive longevity. IVG would decouple the age of the gametes from the age of the person, though only if the somatic cells used as input have not themselves accumulated damage. That is an open question: Epigenetic reprogramming resets methylation marks but does not repair point mutations, and the germline's usual quality-control filters would be operating on cells that never passed through a gonad.

Same-sex genetic parenthood. The mouse work shows the male-to-female direction is possible in principle. The reverse, making sperm from XX cells, has not been achieved and faces the problem that spermatogenesis requires Y-linked genes.

Embryo supply. This is the consequential one. Embryo selection and Polygenic embryo screening are limited less by prediction accuracy than by the handful of embryos an IVF cycle yields. IVG would in principle supply hundreds. The gain does not scale linearly: because selecting the best of n draws from a normal distribution improves roughly with the square root of the logarithm of n, going from ten embryos to a thousand less than doubles the expected shift rather than multiplying it a hundredfold. Even so, it is the only route by which polygenic selection could produce effects large enough to matter, and it makes proposals such as iterated embryo selection, in which stem cells from selected embryos are used to make the next generation of gametes in vitro, technically coherent for the first time.⁷ The constraints on that idea are discussed in Genetic enhancement of cognition.

Editing. IVG would also make heritable Human germline editing more tractable, since edits could be made and verified in a cell line before a gamete is made, avoiding the mosaicism that plagues editing of zygotes. That combination is what most commentators mean when they use the term Designer babies, and it is the reason the technology attracts attention disproportionate to its clinical readiness.

Conservation. The same protocols applied to endangered species would allow gametes to be made from banked tissue, an approach pursued alongside the cloning-based methods discussed in De-extinction. Progress here has been limited by the same species-specificity problem: a protocol tuned for mice does not transfer to a rhinoceros any more easily than it transfers to a human.

§Risks and limitations

Efficiency is the first constraint and may be the least interesting one. The serious concerns are epigenetic.

Genomic imprinting, the parent-specific silencing of certain genes, is established during gametogenesis and is essential to normal development. Culture perturbs it; imprinting disorders are somewhat more common after conventional IVF, and IVG puts the entire imprinting cycle in a dish. Mouse offspring have appeared healthy, but mice tolerate imprinting perturbation differently from humans and the numbers examined remain small.

Somatic input cells carry mutations acquired over a lifetime, including the clonal expansions described in Stem cell exhaustion. Reprogramming does not repair DNA damage, and culture adds its own mutations and karyotypic abnormalities. A gamete made from a fifty-year-old's blood cells is genetically fifty years old even if it is epigenetically young.

There is also a consent problem with no precedent. Gametes could be derived from any nucleated cell, so a discarded biopsy, a blood sample, or in principle a hair follicle would be sufficient starting material to make someone a genetic parent without their knowledge. Legal scholars have flagged this as a gap in existing tissue-consent law.⁸

§Regulation

No jurisdiction currently permits the clinical use of gametes made in vitro. The UK's Human Fertilisation and Embryology Act defines "permitted" eggs and sperm as those produced by or extracted from a person's ovaries or testes, which excludes IVG gametes by construction; using them in treatment would require primary legislation, and the regulator has flagged the point in its law-reform advice. In the United States, an appropriations rider bars the FDA from considering applications involving heritable modification of embryos, which does not cleanly describe IVG but has been read to complicate it, and the agency would in any case treat the gametes as a biological product requiring a full development programme. The International Society for Stem Cell Research's guidelines place clinical IVG in the category of research that should not proceed pending evidence of safety, while permitting the laboratory work.

The governance question resembles the one described in Governance of human genome editing: a technology whose research phase is legal nearly everywhere and whose clinical phase is legal nowhere, in a field with an established pattern of reproductive tourism. The first human IVG birth, if it happens, is more likely to occur in a jurisdiction with weak oversight than in one that has debated the question, which is what happened with Mitochondrial replacement therapy and, in a more serious way, with the He Jiankui affair.

§Outlook

The near-term milestones are legible. A human oocyte that completes meiosis in culture would be the first; fertilisation and normal preimplantation development would be the second; and a human ovarian somatic cell population capable of supporting folliculogenesis without mouse tissue would be the enabling step for both. Sperm may arrive earlier than eggs, since spermatogenesis is shorter and the cells are smaller and simpler, though in vitro human spermatogenesis has its own long record of failure.

Forecasts from within the field have consistently been too optimistic; predictions of clinical use within a decade have been made repeatedly since 2016 and have not held. What has not been argued away is that the mouse demonstration is complete, which distinguishes IVG from most technologies on this wiki. The question is whether the human germline's longer, differently wired developmental programme is a matter of more culture optimisation or of biology that does not run outside a body at all.

§See also

• Embryo selection
• Polygenic embryo screening
• Reproductive longevity
• Induced pluripotent stem cells
• Human germline editing
• Artificial womb
• Human cloning
• Procreative beneficence

§References