Can Old Cells Become Young Again? The Experiment That Has Reached Humans

Somewhere in the United States, inside the eye of a person whose sight has been damaged by disease, an experiment is under way that reaches considerably beyond ophthalmology. The patient is taking part in the first human trial of a treatment intended not merely to protect ageing cells, but to persuade them to recover something they had apparently lost: the behaviour of younger cells.

For decades, ageing research has lived uneasily between respectable biology and extravagant promises. There have been diets, supplements, hormones, genetic theories, fortunes made and fortunes spent, and no shortage of predictions that the first person to live to 150 might already have been born. Much of it has proved easier to sell than to demonstrate. Yet beneath this noisy industry, a quieter transformation has been taking place in laboratories. Scientists have learned that the age of a cell is not quite as fixed as it once appeared, and that some of the molecular changes associated with growing old can, under experimental conditions, be pushed in the opposite direction.

Now another revolution has arrived at almost exactly the same moment. Artificial intelligence is allowing researchers to search biological possibilities at a scale that would have been inconceivable only a few years ago. Billions of hypothetical chemicals can be examined without first being manufactured. Vast collections of biological data can be searched for patterns no scientist thought to look for. Increasingly, computers are being asked not simply to perform calculations, but to suggest experiments.

David Sinclair, the Harvard geneticist who has spent much of his career studying ageing, believes these two developments belong together. His laboratory has helped develop methods for restoring youthful characteristics to old cells. At the same time, he and his collaborators are using computational systems to search enormous chemical landscapes for molecules that might reproduce those effects more simply.

The ambition is formidable. But the experiment in the human eye makes the underlying question much harder to dismiss. For the first time, a technology descended directly from partial cellular reprogramming has entered a human clinical trial.

The memory of a young cell

Sinclair’s theory begins with a distinction between the genetic code and the machinery that decides how that code is used.

Nearly every cell in the body contains essentially the same DNA. A nerve cell and a skin cell possess the same fundamental genetic library, yet one carries electrical signals and the other helps form a protective covering around the body. The difference lies partly in which books from that library each cell opens.

The system controlling those choices is known as the epigenome. Chemical modifications to DNA and to the proteins around which DNA is wrapped help determine which genes remain available and which are silenced. A liver cell therefore maintains one pattern of genetic activity, a neuron another, a muscle cell another.

Age disturbs these patterns.

Some genes that should be quiet become more active. Others become harder to read. The precise molecular identity that a cell maintained for decades gradually becomes less exact. Sinclair has compared the process to scratches accumulating on a compact disc, although the metaphor can be misleading because the original genetic sequence is often still present. What has deteriorated is the machinery deciding how to read it.

This led him to a provocative possibility. Perhaps a substantial part of ageing is an information problem.

The distinction is enormous. A structure that has been physically destroyed must be rebuilt. Information that has become disordered might instead be recovered.

Sinclair’s laboratory spent years trying to determine whether that idea could survive an experiment.

Making time run faster

One of the tests involved deliberately creating a small number of breaks in the DNA of mice. A broken chromosome is an emergency for a cell. Repair proteins move rapidly towards the damaged region because failing to repair it may kill the cell or help produce cancer.

The researchers were interested in what happened elsewhere while this emergency machinery was occupied. Sinclair had long suspected that repeated episodes of DNA repair gradually disturb the systems responsible for preserving the organisation of the epigenome.

The mice were not immediately transformed. Then, as time passed, animals subjected to the intervention began developing characteristics normally associated with older mice. Their tissues changed. Their patterns of gene activity changed. Measurements based on DNA methylation shifted towards an older state.

The work, published in 2023, strengthened the argument that disturbing epigenetic information could produce features of ageing without first having to rewrite the sequence of the genome itself.

It did not settle the question of what ageing is. There are many plausible and overlapping mechanisms. Mitochondria function less efficiently. Damaged cells accumulate. Chronic inflammation rises. Proteins are maintained less effectively. Stem cells decline. Mutations build up. Ageing may ultimately be the consequence of several failures reinforcing one another rather than the product of a single master mechanism.

But Sinclair’s experiment created a more interesting question than the one with which it began.

If disturbing cellular information could push an animal towards old age, what would happen if that information were pushed in the other direction?

Three genes

The answer emerged from one of the most important discoveries in modern cell biology.

Shinya Yamanaka showed that mature cells could be returned to an embryonic like state by activating a small group of genes. The discovery overturned the assumption that a mature cell had travelled down a biological road from which there was no return.

For regenerative medicine, however, complete reprogramming creates an obvious difficulty. A neuron that forgets it is a neuron has not been rejuvenated. It has lost its identity. Push adult cells too far backwards and the result can also be dangerous, including the possibility of uncontrolled growth.

Sinclair and his colleagues therefore experimented with three factors known as OCT4, SOX2 and KLF4, collectively called OSK. Their objective was not to return a mature cell all the way to its embryonic beginning. It was to move it part of the way backwards while allowing it to remain itself.

The most striking early experiments were conducted in the eye.

In a study published in Nature in 2020, researchers introduced OSK into retinal ganglion cells in mice. These are neurons connecting the retina to the brain through the optic nerve. The treated cells developed younger patterns of gene activity and DNA methylation. Damaged axons regenerated. Visual function improved in old mice and in an experimental model of glaucoma.

That mattered because mature neurons in the mammalian central nervous system are notoriously reluctant to regenerate. Damage an optic nerve or spinal cord and the adult body usually does a poor job of rebuilding the connection.

Yet these neurons appeared to recover a capacity associated with younger tissue.

Sinclair began wondering what exactly the cells were remembering.

The backup copy

In his recent interview, Sinclair described what his laboratory informally calls the observer, a hypothetical mechanism preserving information about the younger state of a cell.

The language is seductive, perhaps too seductive. No scientist has opened a cell and discovered a biological hard drive containing an untouched copy of youth. The observer remains a hypothesis.

But the phenomenon requiring explanation is real enough to keep laboratories interested. Mature cells can be reprogrammed. Some youthful molecular patterns can return. Certain lost functions can reappear. The cell therefore seems to possess, or be capable of reconstructing, information that its aged condition no longer visibly expresses.

Nature already performs a still more dramatic version of this trick every generation. The reproductive cells of older parents can create an embryo whose biological age does not begin at the age of its mother or father. During early development, cellular state is reset.

How a cell knows where to return remains one of the great questions.

Sinclair thinks understanding that process may eventually matter less than learning to control it. Medicine has often learned to manipulate nature before completely understanding it. Human beings flew aeroplanes without calculating the movement of every molecule of air around the wings. Doctors used powerful anaesthetics long before all of their molecular actions were known.

The same may prove true here. A complete simulation of a human cell may not be necessary if researchers can identify a small number of reliable controls that move it from one stable condition towards another.

The experiment crosses into humans

In January 2026, the United States Food and Drug Administration cleared an experimental treatment called ER 100 to enter human trials. In June, Life Biosciences announced that the first participant had received it.

The trial is studying people with open angle glaucoma and non arteritic anterior ischemic optic neuropathy, conditions involving damage to the optic nerve.

ER 100 uses a modified viral vector to carry instructions for OCT4, SOX2 and KLF4 into cells of the eye. Expression of the factors is controlled using doxycycline, providing researchers with a means of switching the system on.

The eye is a logical place to begin. It is accessible, closely observed and relatively contained. A powerful experimental treatment can be delivered locally without immediately attempting to alter cells throughout the body.

There is another reason for caution. This is a Phase 1 trial. Its first purpose is safety and tolerability. Researchers will also look for changes in visual function, but the treatment has not yet been shown to restore human vision, much less reverse human ageing.

This is the sober boundary around an extraordinary experiment. An old mouse recovering vision is scientifically important. A human being doing the same would be considerably more important. They are not the same result.

The history of medicine contains many spectacular cures for mice.

The search becomes almost infinite

While this work was moving from animals towards humans, artificial intelligence was altering the scale at which laboratories could search for the next treatment.

Drug discovery has always faced a numerical problem. There are vastly more possible molecules than scientists could ever manufacture and test individually. Even large pharmaceutical companies performing enormous screening programmes explore only a minute corner of the available chemical universe.

Computational screening changes the economics of that search.

A molecule can be imagined before it is manufactured. Its shape, electrical properties and possible interaction with biological targets can be modelled. Most candidates can be rejected without ever entering a bottle.

Sinclair says his collaborators have examined roughly eight billion virtual chemicals in their search for molecules capable of reproducing aspects of cellular rejuvenation.

The number is almost difficult to comprehend. A physical programme testing eight billion compounds would be absurd. A computational search can at least contemplate it.

Sinclair asked an artificial intelligence system to estimate how long his laboratory might have needed to conduct an equivalent experiment by conventional means. It suggested about 160 years and billions of dollars.

The figure is illustrative rather than scientific, but the change in scale is not.

The laboratory of the future does not necessarily manufacture a million failures. It asks a computer to eliminate them before they are born.

For Sinclair, this matters because gene based rejuvenation is cumbersome. Viral delivery is technically difficult and expensive. Even if ER 100 eventually works, such a procedure is unlikely to be the final form of a technology intended for widespread medicine.

What he wants is chemistry.

A molecule that could produce some of the same effects would be easier to manufacture, easier to administer and potentially far cheaper. His laboratory has already experimented with chemical combinations in cells and animals. The larger objective is to find something simpler and more precise, perhaps ultimately a single compound.

This turns ageing research into a search problem, and search is precisely where artificial intelligence has become formidable.

When the computer stops being a calculator

Yet the more interesting role for artificial intelligence may not be screening molecules at all.

Sinclair described working with systems in which several artificial intelligence agents analysed ageing data, tested approaches and developed models. In one recent preprint, researchers used tens of thousands of biological samples from numerous human tissues to construct measures of ageing from patterns of gene activity.

What surprised Sinclair was that the system did not merely reproduce the procedure that experienced scientists expected it to use. It explored another way of analysing the information.

This is a subtle but consequential change.

For most of the history of computing, a scientist decided what calculation should be performed and the machine performed it. A more autonomous scientific system can instead be given a body of evidence and asked what might be worth calculating.

That does not make the machine a scientist in the full sense. Statistical novelty is not discovery. A computer can find a pattern that is meaningless, confounded or simply wrong. Biology still has the final vote.

But the division of labour is beginning to move.

A tireless machine capable of reading millions of experiments can notice relationships no individual scientist could keep in mind. It may also be able to ask questions produced by combinations of knowledge that have never previously occupied the same human brain.

This is the point at which artificial intelligence stops merely making science faster and begins changing what science can examine.

The enormous distance between a mouse and immortality

The temptation is to race ahead.

If an old cell can become younger, perhaps an organ can become younger. If an organ can be rejuvenated, perhaps the body can. If the treatment can be repeated, perhaps ageing itself eventually becomes a chronic condition rather than an unavoidable direction of travel.

Sinclair is willing to contemplate versions of that future. He talks about restoring ageing brains, improving failing organs and eventually developing medicines that could periodically reset aspects of biological age.

But the distance between the experiment now taking place in an eye and a rejuvenated human body remains immense.

Ageing is not simply the epigenome. DNA acquires mutations. Cells disappear. Tissues scar. Arteries change. The immune system evolves. Cancer risk changes. Mitochondria deteriorate. Proteins accumulate damage. The architecture of whole organs is altered by decades of life.

Even if partial reprogramming proves able to restore one important layer of cellular information, it may not restore all the others.

There is also the uncomfortable possibility that the mechanism works beautifully until it does not. A programme powerful enough to alter the identity and behaviour of a cell is not something medicine can afford to control approximately. Cancer, abnormal growth and inappropriate changes of cellular state remain obvious concerns. Effects that look safe for months may look different after years.

This is why the first patient matters and why the first patient proves almost nothing.

A different way of thinking about disease

What remains compelling about the research is not the promise of immortality but the possibility that medicine may have been looking at some diseases from the wrong direction.

Age is the largest risk factor for many chronic illnesses. The older brain becomes vulnerable to Alzheimer’s disease. The older eye becomes vulnerable to glaucoma and macular degeneration. Older arteries develop disease that would be unusual in children. Older muscles lose strength. Older immune systems behave differently.

Medicine ordinarily treats these as separate destinations.

Sinclair’s proposition is that at least some of them may share part of the same road.

If ageing makes a neuron vulnerable, perhaps strengthening the neuron by restoring a younger cellular programme is more powerful than attacking each consequence of its deterioration separately. Instead of asking only how to remove one abnormal protein or block one inflammatory pathway, the question becomes whether the cell itself can once again be made competent enough to deal with the problem.

That idea may ultimately prove too simple. Biology usually punishes simple theories. But it is testable, and increasingly it is being tested.

Two kinds of information

There is something almost symmetrical about the moment.

Artificial intelligence is a revolution in our ability to process information outside the body. Cellular reprogramming is an attempt to understand and manipulate information inside it.

One allows scientists to search billions of possibilities. The other suggests that an old cell may contain possibilities that had been assumed to be permanently lost.

For most of modern history, ageing was treated less as a biological process than as the background condition of existence. Medicine could repair some of its consequences, replace a joint, bypass an artery, remove a cataract or suppress a tumour, but the movement of the organism itself remained in one direction.

The experiments of the past few years have complicated that certainty.

They have not shown that human beings can become young again. They have not shown that Alzheimer’s disease can be erased, that spinal cords can be regenerated in people or that a pill can return an eighty year old body to twenty five.

What they have shown is narrower and, for that reason, more interesting.

Some cells possess a degree of biological reversibility that was once thought impossible.

And now that proposition has left the mouse laboratory.

Inside a human eye, researchers have begun asking whether it survives contact with us.