The longest-lived cells in the human body accumulate damage, do not get younger, and never divide.
Neurons. Cardiomyocytes. Mature oocytes. Some skeletal muscle fibers.
They resist apoptosis, which is to say normal cell death, and they do not form bona fide tumors. They sit stably outside the cell cycle for the entire adult life of the organism they belong to. Every other tissue turns over on some schedule.
These do not.
The cells that already outlive
The trait that correlates most reliably with cellular longevity across mammals is not autophagy efficiency, not telomerase activity, not sirtuin dosage. It is cell-cycle exit.
Cell-cycle exit means the cell has permanently stopped dividing. The neurons you are reading this with are the ones you were born with. They have never been replaced and never will be, and that is precisely why they are still here.
A recent aging model1 puts the number. If every hallmark of aging were eliminated except somatic mutation, median human lifespan would rise to roughly 156 years, and post-mitotic tissue would be the ceiling. Neurons and cardiomyocytes are the bottleneck. Unlike the liver, which stays functional by replacing itself, a post-mitotic cell would run for tens of thousands of years untouched. Somatic mutation is the one thing it cannot clear.
The trait that correlates most reliably with cellular longevity across mammals is cell-cycle exit.
The durability outliers are already in the body. They share the same trait.
Which brings us to who, if anyone, is actually trying to do something about it. Cellular reprogramming does not edit your DNA. It resets the settings layered on top of the DNA — the chemical marks that decide which genes a cell reads and which it ignores. Those marks drift with age. Reset them, and the cell reads its genome the way a young cell does.
The problem is that the tool for resetting them was discovered doing something else. Yamanaka factors turn an adult cell back into an embryonic one, and they do both jobs at once — they erase the age markers, and they drive proliferation and dedifferentiation. Push a cell all the way through that doorway, and you no longer have a young liver cell. You have a cell that has forgotten it was ever a liver cell, and those grow into tumors.
Altos is trying to reset epigenetic state without pushing cells through the pluripotent doorway. Partial reprogramming is the engineering problem of taking the first effect without the second. And their cell annealing paper5 is an honest attempt to formalize that window.
The field's silent consensus
The window turns out to have a condition attached, and the condition is the whole story.
Multiple review papers now note that proliferation may be an essential requirement6 for reprogramming-induced rejuvenation. Neurons could only be epigenetically reset after being forced to divide. So the natural rejuvenation event in the mammalian embryo happens during peak proliferation. Twelve days of OSKM expression in adult mouse hearts is lethal.7 But drop it back to six days, and that can produce regeneration. Post-mitotic tissue does not tolerate what proliferating tissue absorbs.
Read that as a constraint and you get the field's current program: division is the price of admission, so buy your way in. The obvious reading is that you cannot reset a cell without letting it divide. This has led the field to treat post-mitotic status as an obstacle, and current cardiac work is explicitly trying to push cardiomyocytes back into the cycle so they can regenerate after they die once the blood supply fails to reach that area.
The other reading is the opposite.
The safest form of rejuvenation may be the one that resets state and then locks the cell out of division permanently.
What the deal costs
Take that second reading seriously, and it stops being a lab preference. It becomes a trade, with a bill.
You get youthful methylation. Restored transcriptional programs. Refreshed protein homeostasis. And a cell that can no longer produce descendants. In other words, the cell isn't dying. It's rejuvenated and fully functional, but what it can't do is produce a next cell. It's a terminal node. Every capability except succession.
That is not sterilization of the organism. It is sterilization of the somatic cell as a lineage.
You do not get younger. You get more like a neuron.
A body of such cells is durable. It is also a neuron. Every tissue starts behaving the way neurons and cardiomyocytes always have: damage accumulates where it lands, because nothing repairs it in place and turnover waits on a stem-cell compartment.
That eliminates the cancer risk because there's nothing left to transform. So is regeneration, because a cell that can't turn malignant can't turn into anything. Both come from the same fact. Division is what makes a tumor possible, and division is what makes healing possible.
A cut still closes at the skin. But a scarred heart won't rebuild its muscle; the liver stops regrowing, beta cells quit replacing themselves. And that leads to the working tissue inside your organs losing the ability to come back. Parenchymal recovery isn't slower. It's gone.
You don't get younger. You get more like a neuron.
Permissions, not immortality
That is the extreme version, where every cell in the body takes the same deal. Nobody has to build it that way.
The endpoint case is the post-mitotic body, and the practical case is more interesting.
The field's own data already shows that rejuvenation and reproductive authority can be decoupled cell class by cell class; we can choose which trait to impart on the process. Transient reprogramming of human muscle stem cells8 restored regenerative capacity while preserving identity. Senescent progenitor cells regained proliferation9 after partial treatment. Then retinal ganglion cells, they were rejuvenated10 without proliferating at all. Some cells lose the ability to divide after treatment. Some cells regain it. The engineering choice then becomes which authority to leave in place.
That is not one universal reset. That is a permissions system. Youth and the right to divide arrive as separate grants, and you can hand out one without the other.
The architecture that falls out looks like this. Long-lived post-mitotic cells get rejuvenated and replication-locked. Renewable mature cells get rejuvenated temporarily and permitted a bounded amount of division. Stem and progenitor cells get rejuvenated with their self-renewal intact, and are monitored closely. Genomically damaged cells get destroyed rather than reset. This leaves cells with uncertain identity to be excluded or triggered into apoptosis.
Every rejuvenation event carries a permission set. Youthful function, yes. Division rights, only if the cell's tissue role requires them. Descendants, only for the small population authorized to produce them.
Not immortal cells. Rejuvenated cells with constrained execution privileges.
That is what the technology probably looks like if it lands at all.
The same trade, one layer down
Here is the part that should be uncomfortable, because the same shape shows up in systems that have nothing to do with cells. The unit that lives longest is the unit that stopped transmitting.
Institutions that survive across centuries stopped producing new institutions. Founders outlast their firms by never releasing the systems that ran them. Capital concentrates because the exits stopped happening, and every layer of durability at the top of a system corresponds to a suppression of turnover below it. But it is never uniform suppression. It is tiered. Some units retain transmission rights. Most do not. The permissions get gated by whoever holds the authority to grant them.
That is the cellular architecture almost exactly. Long-lived tissue that no longer divides. A small stem compartment that still can, monitored heavily, permitted only because turnover requires it somewhere. Transmission rights concentrated in a subset small enough to control.
The trade is the same trade. Durability and transmission are opposed at every scale where you can measure them. Institutions. Founders. Capital. Cells. The pattern does not change when it drops a layer.
What is actually being sold
The marketable version of rejuvenation is younger and still fully alive, still fully oneself, still fully generative. Youth without the concessions. Healthspan extended and every capacity intact. It's a lovely story and what every influencer will sell.
But the technically defensible version is a permissions system. Durable, epigenetically reset, transcriptionally younger, and quietly gated at the cellular level. A body that has traded local regenerative autonomy for whole-body durability. A durability purchased by locking most of the parenchyma out of its own succession.
That is a coherent thing to build. It is not the thing being sold.
The field's silence on this framing is itself the tell. The safest form of the therapy is the form that trades general reproductive authority for whole-body durability, and that is a much less marketable promise than the one currently on the boxes.
As a result, the published work keeps the possibility open by refusing to frame it. The cell annealing describes escaping a bad attractor without committing to what stable state the cell settles into afterward. It could settle into a younger dividing cell. It could settle into a younger, permanently arrested one. The mathematics do not care. The engineering will.
And that leaves us with the safest path, which may be that a longer human life may run through the cell that stopped having descendants.