Shinya Yamanaka asks the Japanese government for rules: what cellular reprogramming really holds up

A scientist knocks on his own government's door

There is a gesture that rarely happens in the history of research: a scientist walks into a ministry not to ask for funds, nor to defend a project, but to ask that someone set limits on him. "I went to the Japanese government and asked them to establish guidelines for this type of research": that is the sentence this issue starts from, and it is not a request for money. The field he speaks of has a precise name and an ambition the reader has already met in a thousand headlines: cellular reprogramming. Under that headline, however, there are two very different things — and the difference between them is the difference between a therapy and a tumor.

The question that matters, for the reader, is just one: how much of this promise is already proven, and in which animals. Because every line of solid evidence gathered for this issue carries a species next to it, and almost none of those species is ours. The tally, given up front: six claims verified one by one, four confirmed by the primary literature, two uncertain — not because they are false, but because the missing piece of verification is personal attribution, not substance. And in the middle a date, 2026, that shifts everything.

Two roads that start from the same cell

In 2006 four genes — later called Yamanaka factors — showed that a mature cell can be turned back all the way to the embryonic state. The product of that complete journey is called an iPS cell, and today it is a laboratory tool of the first order: the literature calls it "an indispensable tool" for reproducing neurodegenerative, cardiovascular, metabolic and autoimmune diseases in a test tube ⁴.

But if the same journey is made inside a living body, the result changes in nature. Complete reprogramming in vivo, the 2024 Aging Cell review writes, "has proven problematic due to dysplastic cellular proliferation and teratoma formation in multiple organs" ². A teratoma is exactly what happens when a liver cell forgets it is a liver cell and no longer receives instructions on what to become.

From here comes the second road, the one the speaker talks about: partial reprogramming. The same genes are switched on, but briefly: enough to erase the marks of age, not enough to erase the cell's identity. "The goal is not to make iPS cells, the goal is to rejuvenate cells ex vivo or in vivo." The distinction is not a specialist's nicety: it is the hinge on which the rest of this issue rests. The field's literature treats it as standard — partial reprogramming has its own name, its own dedicated review and its own stated objective, rejuvenation ². And this is the first thing to hold on to: a window exists. Before that window the cell is old; after it, it is no longer itself. The problem is that no one yet knows, in humans, exactly where that window closes.

The evidence, one by one, with its species alongside

The strongest result is dated 2024 and has a tail. Aged mice treated with a partial reprogramming system showed a 109% increase in median remaining lifespan: 142.5 weeks versus about 133 weeks for controls, with no gross teratoma formation observed ¹. The number is true and should be stated in full: the percentage refers to the life that remained to be lived at the time of treatment, not to total lifespan. It is the kind of figure that becomes "lifespan doubled" in headlines and remains what it is in protocols: an effect measured in one species, in one experimental window, with no long follow-up, no repeated doses, none of the range of tissues a human therapy would have to cross. The claim "at least in mice it worked" is precise precisely because it stops where the data stop.

The second block of evidence comes not from an experiment but from a table. In 2025 the HESI international committee for cell therapies published a consensus document on teratoma risk assessment in products derived from pluripotent cells, and in that document there is a threshold with numbers inside it ⁶. The third block is the experiment that went around the world, September 30, 2025: eggs obtained from skin cells, at Oregon Health & Science University ¹². We will come back to it, because its cascade of numbers tells a different story from the headline.

The fourth is a single line, and it is the only human line in the entire dossier: NCT07290244, phase 1, authorized in 2026. It is the very first human trial of partial epigenetic reprogramming, and no cellular rejuvenation therapy has yet passed that phase ⁸. A protocol number, not a result. Laid out in a column, this evidence has a regularity the text can only repeat sentence by sentence, while a figure shows it at a glance: the "species" column is almost entirely mouse or culture dish.

Risk has a number, and it changes when you enter a body

In most popular science articles the risk of cancer appears as an adjective: "risky," "to be handled with care." In the field, by contrast, it has a scale. The 2025 HESI consensus records that contamination of 0.1% undifferentiated pluripotent cells does not induce teratomas, whereas 1% and 10% do ⁶. That threshold serves a precise craft: manufacturing a cellular product outside the body — heart, pancreas or retina cells obtained from stem cells — and then certifying how much undifferentiated residue can be tolerated before injecting it. It is a declared, measured tolerance, negotiated between researchers and reviewers.

Then there is the other situation, the one the speaker means when he says in vivo: nothing is manufactured outside, the genes are switched on inside the patient. And here the same literature writes a sentence that leaves no margin: "even a single fully reprogrammed cell is already too many cells at risk of teratoma, and this is a serious challenge for the translational potential one has in mind" ³. The two quantities must be read together, and that is the point of the figure below: on one side a tolerance on the order of one part in a thousand, on the other a tolerance of zero. The passage from ex vivo to in vivo is not a problem of scale, it is a change of regime. There is a biological reason the threshold goes to zero. Outside the body, the product can be counted, filtered, discarded: the wrong cell is found beforehand. Inside, no — the Yamanaka factors are described as pro-oncogenic, and precisely for this reason, Aging Cell says, "their therapeutic applications are currently limited" ².

It is worth unpacking a word that slips by in press releases: "safe." In the first human trial's dossier, safety was built with data on mice, non-human primates and toxicology studies before authorization — "we went in with a lot of safety data" ⁸. Safe, here, means: enough to start trying in humans. It does not mean: proven in humans. And indeed what a regulator would use to decide is still missing: there is no official numerical threshold from the FDA, EMA or PMDA for teratoma rate or for epigenetic reversal in this class of therapies, because "regulatory pathways are not fully established" ¹⁰. The number exists in science, not yet in the rule.

The gamete branch: where the chain really stops

The other branch of the talk is the one that touches birth. If a skin cell can become an iPS cell, and an iPS cell can in principle become an egg or a sperm, then gametes — and children — could be made without using human embryos. "At least in theory, we could generate eggs and oocytes from iPS cells." The conditional is not a stylistic caution: it is exactly the point where the claim holds. On September 30, 2025 the group of Shoukhrat Mitalipov at OHSU announced functional eggs obtained from skin cells using a technique dubbed mitomeiosis ¹².

Then the funnel begins. Eighty-two engineered eggs; about 9% reaching the blastocyst stage at six days ¹². Most embryos stop earlier, at the four-to-eight-cell stage, due to chromosomal abnormalities. And on the last step the column collapses to zero. None of the 82 fertilized eggs had the correct number of chromosomes; what was missing was genetic recombination ¹³. Mitalipov himself said it in a sentence worth more than a press release: "it partially works, and partially doesn't" ¹³.

On the male side the picture is even further behind, and it must be said because in popular accounts "eggs and sperm" always travel in pairs. From human iPS cells researchers have produced spermatogonia and pre-pachytene cells, that is, stages preceding the start of meiosis: mature, functionally validated human sperm remain an unreached goal ¹⁷. The authors of the Portland experiment estimate "at least a decade" before a possible clinical trial ¹². The UK fertility authority commented that it is "only a proof of concept" and that more research on safety and efficacy is needed ¹⁴. The driest line comes from the American Society for Reproductive Medicine, in its 2026 ethics opinion: "no data, including safety data, exist today for the use of in vitro gametogenesis in humans" ¹⁵. And in mice, where the technique is more mature, only 1-3% of embryos developed in vitro and fertilized come to term.

The rules and the science, on the same time scale

Japan, on paper, is the country that has written the most rules on this ground. The guidelines of the ministry of education and science on creating germ cells from iPS cells and tissue stem cells were revised on February 13, 2026 and entered into force on April 1, 2026 — previous revisions dated March 2022 and April 2019 ²⁸. Earlier still, in July 2025, the Cabinet Office bioethics commission reached substantial agreement to allow the creation of human embryos from iPS cells, with a culture limit set at fourteen days ³⁰; the August 2025 report carries a title that says it all: the creation of human embryos with germ cells derived from stem cells ²⁹.

Who pushed for those rules to exist, however, the collected material does not say. The sources document the guidelines, the dates, the culture limits; none names the speaker as their promoter. It is the difference between the fact and the signature on the fact. At the international level, the claim that "national laws are not enough" stops being rhetoric the moment you look at the numbers. A study on the global governance of human genome editing examined 106 countries: 96 have policy documents relevant to embryos, gametes or precursor cells, but only 23 explicitly ban and 11 explicitly permit research on genetically modified embryos in the laboratory ²⁷. The others stay silent, or speak of something else.

The guidelines of the international society for stem cell research are the most-cited reference, but they remain voluntary and "subject to the local legal and regulatory jurisdiction of the different countries," in a framework described as "complex and evolving" ²⁵. And Japan, which would seem the natural candidate for harmonization, diverges instead of converging on one decisive point: it explicitly bans the production of embryos from gametes obtained by in vitro gametogenesis, while the 2021 international guidelines classify as category 1B research on generating gametes from stem cells when no fertilization is attempted ²⁴.

In the United States the block is of yet another nature: the FDA is by law barred from accepting clinical trial applications in which the embryo is created or modified to include heritable modifications — a budget rider introduced in 2016 and renewed since. At the state level the picture is fragmented: 18 permissive states, 11 prohibitive, 21 silent ²¹. Stacked one atop the other, the two timelines visibly slip out of phase: the advances have precise dates and a protocol number, the rules mostly have requests, national revisions and a global table that does not exist.

On what to do, those entitled to say it speak with caution and in writing. The ethics committee opinion of the American Society for Reproductive Medicine, published in 2026, argues that in vitro gametogenesis should not be considered for reproductive purposes in humans until robust studies in non-human primates with reassuring outcomes have been conducted, and that in the meantime it remain under the sole oversight of research ethics committees ¹⁵. The international society for stem cell research, for its part, places in vitro gametogenesis for human reproduction in the category of currently prohibited activities until safety and ethics questions are resolved, and points as the most promising route to the one starting from frozen immature follicles for fertility preservation, not full derivation from somatic cells ¹⁶.

The speaker adds a consideration that is his and remains his: overcoming the need to use human embryos to obtain eggs and sperm solves one ethical obstacle but opens another. And on how to untie that knot, he says, national laws are not enough: a global confrontation is needed.

Opinion reported as such, not verified by the editorial team.

What holds up, enthusiasm removed

Anyone reading this issue looking for a date for their own health will not find one, and it is worth saying without detours: 2026 brought an authorization, not a result. Phase 1 serves to establish whether something can be administered, not whether it works. The rest, however, is not smoke. The distinction between partial and complete reprogramming is established in the field, not a defensive formula; the effect in mice is quantified and published; the risk of teratoma is not a generic worry but a parameter with thresholds that someone has measured. The interesting twist is another one, the one the threshold figure makes visible: the field has given itself a numerical tolerance for cellular products manufactured outside the body, and at the same time a zero tolerance for the same technology switched on inside a living body. Two standards speaking different languages coexist within the same craft.

On the gamete branch the verification gives a sharp outcome in its nuance: the claim holds in the theoretical conditional in which it was formulated, and falls the moment it is translated into the indicative. Eighty-two eggs, none with the right chromosomal makeup, mature sperm never obtained: the chain breaks before the last link, that is, before the one that matters for making new life. On the rules, the only thing the data allow one to say is that convergence is not underway. Twenty-three explicit bans and eleven explicit permits across 106 countries do not describe an international order: they describe a mosaic with many blank tiles. An asymmetry remains that no source resolves. The most authoritative guidelines are voluntary; the binding laws are national; and research — the dates say so — moves faster than both.

Notes from the notebook

There is a detail in the 2024 mouse paper that almost no headline reports and that changes the size of the result: that 109% is not the mouse's life, it is the life it had left. An animal already old at the time of treatment reaches a median of 142.5 weeks against about 133 for controls. In weeks it sounds less sensational than in percentage — and it is the same thing. The most honest sentence in the whole dossier was not written by a regulator but by the researcher who had the most to gain from not saying it. Mitalipov, facing the result newspapers were already calling historic, commented: "it partially works, and partially doesn't." Then he added, on the chromosomes' makeup: "there's no way around it."

In the consensus document on teratomas hides a small lesson in method. The 0.1% threshold is not a precautionary principle: it is the point at which, experimentally, the tumor stops appearing. The difference between a prudential limit and an observed limit is everything that separates a rule that holds from one that reassures. Then there is the story of the American rider. The ban preventing the FDA from even looking at a trial application on embryos with heritable modifications is not a bioethics law: it is a clause inside the spending law, born in 2016 and copied over every year since. An entire field of research stays closed by a clause that technically expires every twelve months.

The Japanese archive, for those with the patience to open it, tells a rhythm. The guidelines on germ cells from iPS were revised in April 2019, in March 2022 and then on February 13, 2026: three rounds in seven years, each with well over a year between the decision and the entry into force. It is slow compared with a laboratory, and it is very fast compared with any international treaty. Finally, a note on a sentence no one can verify and that is nonetheless the heart of the story. The original quote on the "two medical applications" contains a passage the transcriptions render as "region dimension": almost certainly a mishearing of "regeneration." The dossier confirms that disease modeling is indeed an established application of iPS cells; what the second one was, and who worked on it, remains outside what public sources allow one to say.

The man who asked for a limit

Let us return to the ministry door. A scientist walks in and asks that someone write the rules of his trade. In 2025 and 2026, in that country, the rules did arrive: a revision date, an entry-into-force date, a fourteen-day culture limit. In the same span of time, on the other side of the Pacific, a review committee gave the green light to the very first human trial of partial reprogramming. Phase 1, a protocol number — NCT07290244 — and no result. The question carried into that ministry arrives at the table exactly now. Except the table he spoke of — the global one — does not exist yet: ninety-six countries with documents, twenty-three with a ban, eleven with a permit, and in between an arithmetic that does not add up to an agreement.