In-Vitro Gametogenesis

Established In-vitro gametogenesis means building an egg or a sperm cell from a pluripotent stem cell, with no gonad involved at any step. The stem cell can itself be reprogrammed from a skin or blood cell, so the end state of the technology is a route from a cheek swab to a gamete. That is a different claim from growing a foetus outside a body, the subject of Artificial Wombs, and different again from editing a genome, the subject of Genetic Engineering. IVG supplies cells; the other two do things with them.

Established The mouse version is finished and the human version is not, and the distance between them is the whole subject. In 2016 a Japanese group reconstituted the entire female mouse germline in culture — pluripotent cell to fertilisable oocyte to live, fertile pup. Established In human cells the same laboratory lineage has reached oogonia, a foetal stage, after roughly four months of culture, and has not got past it. Frontier What is missing is not a machine but meiosis and oocyte growth inside a human-specific somatic environment nobody has yet built from human cells.

Frontier The binding wall is epigenetic, which is why honest timelines here are long. A germ cell is defined less by its shape than by its methylation state: it must erase inherited parental imprints and lay down a new, sex-correct set, on a schedule that in humans takes months. Established Imprinting error is the observed failure mode of the last shortcut in assisted reproduction, intracytoplasmic sperm injection, which entered routine use in 1992 and whose imprinting signal was characterised a decade later, on children. Handwave The claim that IVG will be safe because mouse pups looked normal is where the argument runs on assertion: mouse imprinting control regions are not human ones, and a mouse is never asked about its cognition at forty.

Frontier What the technology would change is access to gametes, not the mechanics of reproduction. Eggs are the scarcest input in human reproduction, which is why a market in donor oocytes exists at all. Speculative An unlimited autologous supply would dissolve that scarcity, and with it the age ceiling on genetic motherhood and the practical ceiling on how many embryos a couple can generate to choose among. Established That last consequence is why IVG appears in enhancement arguments it has no scientific connection to, and why this brief keeps the science and the ethics in separate sections rather than letting each supply the other’s missing evidence.

Established A note on sourcing. This brief was commissioned in September 2026 from the Institute’s research base. Reading-list entries without links are cited from the bibliographic record rather than re-fetched, and claims are dated no later than early 2026 unless carried by a linked source.

Established The founding mouse results are a decade old and were reached in stages. In 2011 Hayashi and colleagues induced primordial germ cell-like cells from mouse pluripotent stem cells and showed that, transplanted into infertile testes, they produced functional sperm and healthy offspring; in 2012 the female-side equivalent produced oocytes after aggregation with foetal ovarian somatic cells and transplantation. Established In 2016 Hikabe and colleagues closed the loop entirely in culture, producing metaphase-II oocytes that were fertilised and yielded fertile pups whose own germ cells could re-enter the same system. That result is why the field exists.

Established It is far less efficient than the headline suggests, and the efficiency number is the one most often dropped in retelling. The 2016 protocol ran about five weeks and produced live pups from on the order of three per cent of transferred embryos. Established The oocytes were smaller than in-vivo oocytes, matured abnormally in a substantial fraction, and carried elevated aneuploidy. Frontier Later work has raised yields without changing the shape of the problem: the culture produces a population of which a minority are competent, and no marker identifies that minority before use.

Established The most striking mouse result since produced pups with two genetic fathers, and its efficiency is the argument against reading it as a preview. In 2023 Murakami and colleagues took XY pluripotent cells, exploited spontaneous Y loss and X duplication to obtain XX cells, and ran them through in-vitro oogenesis; fertilised with sperm from a second male, the embryos gave seven live pups from roughly 630 transfers, about one per cent. Frontier A 2025 Chinese report separately described bi-paternal mice made by correcting around twenty imprinted loci in androgenetic embryos, a few surviving to adulthood. Established Both results are about imprinting, from opposite directions: parent-of-origin methylation is the gate, and getting it right is the experiment rather than a detail of it.

Frontier On the human side the field has a foetal-stage germ cell and no meiosis. Human primordial germ cell-like cells have been induced reproducibly since 2015. In 2018 Yamashiro and colleagues showed those cells progressed to oogonia over roughly four months when aggregated with mouse foetal ovarian somatic cells, with the expected genome-wide demethylation. Frontier Work reported in 2024 from the same lineage reconstituted the human germline reprogramming step in vitro more faithfully than before. Established What has not been reported anywhere, by anyone, is a human oocyte that entered meiosis, grew, matured to metaphase II and proved fertilisable, starting from a pluripotent cell.

Frontier A third route bypasses pluripotency and reached human cells first. Nuclear-transfer approaches place a somatic nucleus into an enucleated donor oocyte and then induce a reduction division to make the result haploid. Frontier A 2025 report from a US reproductive-biology group described functional human oocytes made this way, with a small percentage reaching blastocyst after fertilisation and, on the authors’ own account, pervasive chromosomal abnormality. Speculative The route inherits the donor-oocyte scarcity IVG is meant to abolish, since it needs an enucleated egg per attempt, and it sits at the edge of the period this brief can verify.

Established The funding picture is small, private and concentrated, which is itself a finding. Human IVG is pursued by a handful of academic laboratories, principally in Japan, and by a few venture-funded firms explicit that their objective is unlimited autologous eggs; no government runs a named IVG programme with a published budget.

Frontier Can a human oocyte grow to full size without a human ovary? The mouse succeeded because foetal ovarian somatic cells could be supplied, and later made from mouse pluripotent cells. Frontier The human equivalent — a human ovarian somatic cell-like cell competent to organise a follicle — has not been demonstrated to that standard, so human reconstituted ovaries still depend on mouse tissue, which is clinically disqualifying, or scarce human foetal tissue, which is unavailable in most jurisdictions. This is the field’s most specific unsolved problem.

Speculative Is human meiosis in culture a timing problem or a signalling problem? Human germ cells enter meiosis in the foetus and arrest until ovulation decades later. One reading is that culture has not been run long enough with the right cue sequence; a competing reading is that arrest and re-entry need signals only a somatic follicle gives. No published experiment separates the two.

Frontier Is direct conversion a shortcut or a dead end? In 2021 Hamazaki and colleagues induced oocyte-like cells directly from mouse pluripotent cells with a small set of transcription factors, skipping the germline trajectory; the cells supported fertilisation and early cleavage but yielded no offspring. Handwave The assertion that stacking enough factors will make the epigenome follow the transcriptome is exactly the step the imprinting data says is not free.

Frontier What is the acceptance criterion for a gamete that was never in a gonad? No validated assay predicts whether a given oocyte will produce a healthy adult. Frontier Candidate panels exist — methylation at imprinted loci, transcriptome comparison against in-vivo references, structural-variant calling — and none is anchored to an outcome, because anchoring requires the outcome.

Established Culture duration is the first-order constraint and it scales against the technology. A mouse protocol runs about five weeks. Human germ-cell development runs on a multi-month to multi-year clock, and published human work already needs around four months to reach a foetal stage. Frontier Every extra month multiplies contamination risk, medium-batch variance, genetic drift and cost per attempt, and there is no known way to run human developmental time faster without changing what is being made.

Frontier The somatic niche is the second, and it is a supply problem disguised as a science problem. Mouse-derived ovarian somatic cells are convenient and clinically disqualifying; human foetal ones are legally constrained nearly everywhere and cannot be manufactured to a standard. Frontier The only route that scales is deriving the niche from the same pluripotent stock as the germ cells, done in mouse and not, to a demonstrated standard, in human.

Established Genomic integrity is the third, and it worsens the longer the first bottleneck lasts. Pluripotent lines acquire culture-adapted mutations and copy-number changes with passage number, and long protocols require passages. Frontier A gamete is the one cell type where a somatic mutation is not confined to the person carrying it, so the tolerable mutation burden is lower than for any cell therapy — and nobody has stated that tolerance in numbers.

Speculative Cost is the fourth. A months-long GMP culture with a bespoke niche, per patient, has the cost structure of an autologous cell therapy rather than of an IVF add-on, which is a commercial problem large enough to decide which indications ever exist.

Established IVG depends on induced pluripotency at clinical grade, which is routine but not free. Reprogramming is a twenty-year-old technique with established GMP practice, and the chemistry is covered in Cellular Rejuvenation. Frontier What IVG needs beyond that is line-level epigenetic characterisation, because a line with an aberrant imprint is disqualifying here and merely suboptimal elsewhere.

Frontier It depends on single-cell epigenomics reaching a sensitivity the clinic can use. Establishing that a derived oocyte carries a correct, sex-appropriate imprint set means reading methylation at named loci from very small inputs; the research assays exist and validated, non-destructive, regulator-acceptable versions do not.

Established It does not depend on genome editing, and the field is damaged by the assumption that it does. IVG is a cell-derivation technology; editing is a separate capability that IVG would make easier to apply at scale, for the mundane reason that screening many embryos is easier than screening few. Established Genetic Engineering records that delivery, not nuclease performance, is the binding constraint on editing, and nothing about IVG relieves it.

Established In the clinic it depends on embryo culture and transfer being unchanged. An IVG-derived oocyte still has to be fertilised, cultured to blastocyst and transferred — conventional IVF with conventional success rates, which caps end-to-end yield at the live-birth rate of the transfer step however good the gamete is.

Frontier The decisive experiment is a human oocyte, derived entirely from pluripotent cells in culture, that reaches metaphase II and is shown to be fertilisable, reported with its efficiency denominator and its methylation state at imprinted loci. Nothing else would change this brief’s assessment as much, because every downstream question — safety, regulation, cost, ethics of use — is currently argued about a cell that does not exist. Frontier It is a research experiment, not a clinical one, and no transfer is involved. Its horizon is this decade rather than this year, and the gating item is the human ovarian somatic niche rather than any instrument.

Frontier The second experiment makes the first interpretable: a human ovarian somatic cell-like cell from human pluripotent stem cells, competent to organise a follicle around a human oogonium. The mouse equivalent has been reported and is what let the mouse system become fully in vitro. Until the human version exists, any human oogenesis result carries mouse or foetal tissue inside it, and is therefore not a product and arguably not a clean experiment.

Speculative The third is a multigenerational imprinting study in a long-lived species, which nobody has funded. Imprinted-gene phenotypes in humans appear as growth disorders, metabolic disease and neurodevelopmental outcomes on decade timescales, so mouse data cannot answer the question that matters. Speculative A non-human primate IVG line followed to reproductive age and into a second generation is the only responsive preclinical design, and would take fifteen to twenty years in a facility that does not exist for the purpose.

Established A natural experiment is already running and is under-used. Intracytoplasmic sperm injection has been performed millions of times since 1992, and those cohorts are the best available model of what happens when a reproductive shortcut is adopted ahead of its safety characterisation; registry follow-up produced measurable signals one to two decades after adoption. Frontier An IVG regime that did not pre-commit to follow-up of the same kind would repeat an error whose cost is documented.

Established The physical plant is unremarkable and is not what is missing. IVG runs on incubators, microscopes, micromanipulators and standard cleanroom practice. Frontier What is genuinely hard is process control over a months-long, multi-stage aggregate culture with no in-process release criterion — closer to a fermentation problem with a single irreplaceable batch than to a device problem.

Speculative Automation is the credible lever and is being pulled in the adjacent field first. Robotic handling, computer-vision grading and closed-system culture are being commercialised for conventional IVF to cut operator variance; IVG would need that maturity as a precondition, because variance introduced over three days becomes a dominant term over four months.

Frontier The reagent supply chain is a real constraint and is single-sourced in places. Long germ-cell cultures depend on recombinant cytokines, defined matrices and serum replacements whose lot-to-lot variation already limits reproducibility between laboratories, and a clinical process would need those at GMP grade under multi-year commitments. Handwave Claims that IVG will be cheap because stem cells divide indefinitely treat the expensive part — differentiation, maturation and certification — as though it were the cheap part.

Established Ectogestation is the technology IVG is most often confused with and overlaps with almost not at all. Artificial Wombs records that partial ectogestation has a decade of large-animal data while complete ectogenesis lacks an implantation substitute and a synthetic placenta. IVG supplies the beginning of a pregnancy and ectogestation the end; neither shortens the other’s critical path.

Established Embryo selection is the adjacency that matters, and the arithmetic is unkind to the enthusiasts. Biological Enhancement records that the peer-reviewed expected gain from choosing the best of ten embryos on a cognitive polygenic score is under a year of education, with a prediction interval spanning zero, and that most of the score is not the child’s own direct genetic effect. Established IVG would raise the number of embryos to choose among; gain from selection grows roughly with the square root of the logarithm of that number, so ten to a thousand does not multiply the effect by a hundred.

Frontier Stem-cell-based embryo models are the near neighbour that will reach regulators first. Integrated embryo models built from pluripotent cells already reach post-implantation stages in culture and share IVG’s legal problem — entities that behave like embryos but were not made from gametes. Whichever jurisdiction resolves that first sets the template.

Established Mitochondrial donation is the precedent worth copying, the one case where a novel reproductive technology was legalised deliberately. The United Kingdom created a statutory route in 2015, restricted it to a single licensed centre, and reported the first cohort of children in 2025 with mixed but broadly reassuring results. Narrow indication, one centre, mandatory follow-up, published outcomes: nothing about IVG makes that template unavailable.

Established In the United Kingdom, IVG-derived gametes are excluded from treatment by a definition rather than by a ban. The Human Fertilisation and Embryology Act permits only “permitted eggs” and “permitted sperm”, defined as cells produced by or extracted from ovaries or testes and not otherwise altered; a gamete derived from a skin cell is not such a cell, so no licence could authorise its use. Established The regulator’s 2023 law-reform recommendations flagged this and asked Parliament for a future-proofing mechanism. Research derivation, by contrast, is licensable.

Frontier In the United States the position is a patchwork with a fiscal rider at its centre. Since fiscal 2016 an appropriations rider has barred the FDA from considering applications involving a human embryo intentionally created or modified to include a heritable genetic modification. Frontier Whether unedited IVG falls inside that language is genuinely contested; the conservative reading is that it does not, but that an IVG-derived gamete would still be a regulated biological product requiring a licence. No sponsor has tested either question.

Frontier Japan hosts most of the science and permits less of the application than outsiders assume. National guidelines have allowed derivation of gametes from pluripotent stem cells for research while restricting their use in fertilisation, so the laboratories best placed to run the decisive experiment may need a guideline change to run it — an institutional gating item, not a scientific one.

Established The professional-society layer is restrictive and, in a field this small, is most of what actually binds. International stem-cell research guidance treats clinical use of in-vitro-derived gametes as not currently permissible while treating derivation and study as reviewable research; it has no force of law but determines what journals publish and what funders pay for.

Speculative The gap no jurisdiction has filled is consent to the derivation itself. Almost any cell can be reprogrammed, so a discarded biopsy could in principle become someone’s gamete without their knowledge. Frontier Tissue-consent frameworks were written for research uses, not for reproduction, and the remedy — a specific, separable consent for gametogenic use — is cheap to legislate and has been legislated nowhere.

Established What follows reports positions rather than adjudicating them, and several of the strongest are ones this brief would not endorse. The literature is polarised because IVG touches embryo status, parenthood, the body and the market at once, with argument running ahead of evidence in every direction.

Frontier The access argument is the field’s strongest and is usually stated too narrowly. IVG would offer genetic parenthood to people who cannot now have it: same-sex couples wanting a child related to both, people made infertile by childhood cancer treatment, people with premature ovarian insufficiency. Established These are the existing patient base of donor-gamete programmes, and their interest in a genetic tie is the interest the law already recognises when it regulates donor anonymity.

Frontier The embryo-protective objection is the most consistent position against, and it depends on no empirical question this brief could settle. On views assigning full moral status from fertilisation, IVG is objectionable not for what it does to gametes but because its point is to generate embryos in quantity, most of which will not be transferred. Established Catholic teaching and a range of secular embryo-status arguments reach that conclusion independently of safety data; no improvement in imprinting fidelity answers them, and treating the objection as a safety claim that better science will retire misreads it.

Frontier Feminist analysis divides sharply and both branches are serious. One welcomes IVG as the end of ovarian stimulation and of a donor-egg market in which young women accept medical risk for money, a market criticised as coercive for thirty years. Frontier The other argues that IVG completes the extraction of reproduction from women’s bodies into a laboratory-and-market system, that the burden merely migrates to gestation, and that abundant eggs will increase rather than reduce pressure on women to produce children on a schedule set by others. Both argue from the same history and reach opposite conclusions.

Speculative Solo and multiplex reproduction attract the attention and are not the cases that would arrive first. Unibiological reproduction — a child from one person’s cells alone — is possible in principle by deriving both gametes from one donor, and would produce an extreme consanguinity with predictable recessive-disease consequences. Frontier Multiplex parenthood through iterated rounds of IVG is coherent on paper and requires generational cycling no mammal has shown past the first round. Handwave Treating either as near-term is this subject’s clearest case of an ethical argument borrowing certainty the science has not supplied.

Frontier The child-welfare argument cuts uncomfortably and deserves stating at full strength. A person created from an IVG gamete cannot consent to the risk, and the risk may not be visible for decades; ICSI was adopted rapidly on thin preclinical evidence, and the cohorts later showed small but measurable increases in imprinting disorders. Frontier Against this, the non-identity objection is forceful — the alternative for that person is not a healthier life but no life — which is why reasoning of this kind converges on a threshold of decent prospects rather than on comparison to an unharmed counterfactual.

Frontier The enhancement argument is where IVG is most often conscripted, and this brief treats the conscription as unsound. Proponents of strong germline modification argue explicitly that IVG is the enabling step for iterated embryo selection, because cycling generations in a dish would compound gains that are trivially small in one round. Established The published single-round gains are small with intervals including zero, and cycling additionally requires deriving gametes from embryonic cells reliably, shown in mice and in no other species. Handwave The step that runs on assertion is the assumption that the epigenetic cost of cycling is zero, never tested past one generation in any mammal.

Frontier The disability-rights objection applies to the selection use and not to the access use. The expressivist claim is about what IVG would make easier, not about IVG itself, and the fact that one laboratory process supports both uses is why the governing question is indication control rather than a technology ban.

Speculative The largest demographic claim made for IVG is that it decouples fertility from age, and it is half true at most. Oocyte quantity and quality dominate late female fertility, and a derived oocyte would carry the nuclear genome of the person as she is now rather than as her ovary is now. Frontier But a somatic cell from a fifty-year-old carries five decades of accumulated mutation, and gestation at that age carries obstetric rather than gametic risk. Handwave The claim that IVG abolishes the reproductive clock treats the ovary as the only clock in the system.

Frontier A real but under-discussed consequence is the collapse of the donor-gamete economy. A global market in donor eggs exists because oocytes are scarce; it involves cross-border travel, payment structures that vary by jurisdiction, and a documented set of harms to donors. Frontier Abundant autologous oocytes would remove its reason to exist — one of the few unambiguously positive consequences on offer, and one neither enthusiasts nor critics foreground.

Handwave Claims that IVG changes the human species are the weakest in the file. Established Population-scale genetic change requires adoption by a large fraction of reproducing people, sustained across generations and pointed consistently; assisted reproduction accounts for a low single-digit percentage of births even where it is used most, after four decades and at far lower cost and burden than IVG would carry. Handwave A months-long, expensive, mostly unavailable technology has no species-scale mechanism, and asserting one is where this subject stops being about evidence.

These horizons track what would have to be demonstrated for a claim to change status, not predictions of arrival.

• 10 yr: Frontier A human oocyte matured to metaphase II entirely from pluripotent cells is plausible on this horizon and has no announced schedule; the gating item is a human ovarian somatic niche, not an instrument. Frontier Human in-vitro spermatogenesis is the less-watched half and may arrive first, because the male trajectory has a shorter developmental clock. Speculative A first clinical use anywhere would require statutory change wherever gametes are defined by their origin.
• 25 yr: Speculative If human oocytes are made at all, this is the horizon on which safety is decided rather than argued: a first cohort of children, a mandatory registry, a first decade of follow-up, with first indications narrow because there the counterfactual is no genetic child at all. Frontier On the same horizon the donor-oocyte market either shrinks or does not, which is the cleanest observable test of whether the technology reached routine use.
• 50 yr: Speculative Same-sex genetic parenthood, if it happens, is a fifty-year item rather than a ten-year one, because it needs the harder gamete made from the wrong starting karyotype and then needs a legal system to have caught up. Handwave Generational cycling in vitro — the technical core of iterated selection — remains a mouse result on this horizon unless somebody shows the epigenome survives a second round in a long-lived species.
• 100 / 250+ yr: Handwave Claims about species-level genetic change, engineered lineages, or reproduction detached from human bodies belong here, and belong here flagged. Handwave Each requires adoption fractions, generational persistence and directional consistency no reproductive technology has achieved, and none follows from making a gamete in a dish.

• Depends on Less than the framing suggests. IVG inherits clinical-grade reprogramming and pluripotent-cell manufacture, whose chemistry is covered in Cellular Rejuvenation, and inherits nothing from Genetic Engineering, whose binding constraint is delivery to tissues IVG never touches. It does not wait on Artificial Wombs. No result in another brief on this map sits on this one’s critical path.
• Requires (not on this map) First, a human ovarian somatic cell-like cell from human pluripotent stem cells, shown to organise a follicle — the missing piece that keeps every human result dependent on mouse or foetal tissue. Second, an assay reading imprinted-locus methylation in a single oocyte without destroying it, since certification otherwise consumes the product. Third, a statutory definition of a usable gamete that does not turn on the organ it came from, since the UK Act and its imitators exclude derived cells by definition. Fourth, a consent instrument separating gametogenic use of stored tissue from research use, drafted by no jurisdiction and the cheapest of these to obtain. Fifth, GMP-grade cytokines, matrices and serum replacements under supply commitments long enough to survive a multi-month culture. Sixth, a payer for a months-long autologous process in a sector that mostly bills patients directly.
• Enables An unlimited autologous oocyte supply: genetic parenthood after gonadal loss, an end to the oocyte constraint on mitochondrial donation and nuclear-transfer research, and more embryos available for screening. It enables, mechanically rather than morally, the selection arguments treated in Biological Enhancement and Neurogenetics — both of which record that the gains are small and slow-growing in the number of embryos, so the enabling relation is real and the payoff is not.
• Adjacent Stem-cell-based embryo models share this subject’s legal problem and will reach regulators first; in-vitro maturation of patient-derived oocytes is the commercially active neighbour routinely mis-reported as IVG; mitochondrial donation is the procedural precedent worth copying. Artificial Wombs is the adjacency most often mistaken for a dependency, and Precision Medicine supplies the general lesson that predicting well says little about acting well.

Established “Scientists have made human eggs in the lab.” No. What has been made is a human oogonium, a foetal-stage germ cell, after about four months of culture in an aggregate containing mouse ovarian tissue; the steps between that and a fertilisable egg are the majority of oogenesis by duration and have been completed for human cells by nobody. Frontier Reports saying otherwise usually describe in-vitro maturation of a patient’s own retrieved oocytes using stem-cell-derived support cells — a real and useful product now in clinical use in several jurisdictions, and a different thing.

Established “It worked in mice, so it is an engineering problem now.” This is the field’s characteristic error. The mouse protocol works because its somatic niche could be supplied, because mouse oogenesis runs on a weeks-long clock, and because mouse imprinting control regions are the ones it was tuned against. Each is species-specific.

Frontier “IVG lets parents pick from thousands of embryos and dramatically raise their children’s traits.” The arithmetic does not support the adverb. Expected gain from selecting the best of n embryos grows approximately with the square root of the logarithm of n, so a hundredfold increase in embryos buys well under a doubling. Established The effect being multiplied is itself small: the peer-reviewed best-of-ten estimate on a cognitive score is a fraction of a year of education with an interval spanning zero, and a majority of the predictive signal is not the child’s own direct genetic effect. Frontier This is the rare case where the speculative technology and the speculative payoff can be evaluated separately, and the payoff fails on its own terms.

Frontier “IVG is banned.” Mostly it is undefined, which is a different and more consequential state. Research derivation is licensable in several countries today, and a technology that is unregulated rather than prohibited tends to arrive first where the law is least developed. That, rather than prohibition, is the governance risk attaching to this subject.

Speculative “Companies are three to five years away.” Treat this as a fundraising statement until an efficiency denominator appears. Several venture-funded firms give near-term human timelines; none has published a human metaphase-II oocyte, a human ovarian somatic niche from pluripotent cells, or an imprinting characterisation of a derived human germ cell. Frontier The field’s academic leaders give decade-plus estimates and name the same missing pieces. Where founders and investors differ by a factor of three on timing, the disagreement is about incentives rather than data.