Will you live to 120? · The Counting Problem

What the mouse won’t tell you

It would be easy to read that last section as “telomeres don’t really matter in people.” That would be the opposite error, and the book owes you the other half.

We have a natural experiment that proves telomeres matter in humans: people born with mutations that cripple telomere maintenance — in the genes for telomerase or its partners. These telomere biology disorders (the best-known is dyskeratosis congenita) produce exactly what the theory predicts, and they produce it first in the tissues that divide the most. Bone marrow, which must manufacture blood cells around the clock, fails. The lining of the lungs scars (pulmonary fibrosis). The gut and skin suffer. Affected people show features of premature aging, concentrated in the highest-turnover tissues, sometimes decades early.

So in humans, telomeres are not a decorative correlate. When maintenance genuinely fails, people age prematurely and predictably — fastest where cell division is fastest. The truth sits in the uncomfortable middle. Telomeres are a real, sometimes rate-limiting part of human aging in specific tissues and specific people, and at the same time a weak, noisy yardstick for the gradual aging of a healthy person. Both halves are true. A reader who knows only one half will misread the news for the rest of their life.

What the mouse won’t tell you

There is one more result you will see cited — often as proof that aging is basically solved — and it is worth walking through slowly, because it teaches the lesson this entire book is built around.

In 2011, a group reactivated telomerase in mice and reversed signs of aging. Shrunken brains regrew, failing organs recovered, the animals visibly rejuvenated. It is a real, dramatic result. It is also widely misread, and the misreading is instructive.

Two things the headline leaves out. First, those mice were not normal. They had been genetically engineered to lack telomerase entirely, so that their telomeres collapsed and they aged prematurely; the experiment switched telomerase back on and rescued them. That is not “making a healthy animal younger.” It is curing an artificial deficiency the experimenters built in the first place. Restoring something you deliberately removed is a much weaker claim than improving a normal animal — and only the second would tell you anything about yourself.

Second, and more fundamental: the ordinary laboratory mouse is one of the worst possible models for human telomere aging — because in a mouse, telomeres are not the limiting factor at all. Lab mice carry enormous telomeres, tens to over a hundred thousand base pairs against roughly ten thousand in humans, and they keep telomerase switched on in many adult tissues. By the logic of the Hayflick limit, a mouse ought to be nearly ageless. And yet a mouse grows old and dies in two or three years. Telomere shortening simply is not what paces aging in a normal mouse; other mechanisms get there long before the telomeres do. So if you want to test the hypothesis “telomere shortening drives aging, and restoring telomeres reverses it,” the standard mouse is precisely the animal in which telomeres matter least. The 2011 result works because the mice were re-engineered to make telomeres limiting. It does not transfer to a normal mouse — let alone to a normal human.

This is the pattern to carry into every chapter that follows. Aging biology is full of astonishing demonstrations in worms, flies, and mice: lifespans doubled, organs regrown, clocks wound back. Some of those effects survive the jump to humans. Many quietly do not — because the model organism differs from us in exactly the way that made the experiment work in the first place. A reader who sees “reverses aging in mice” and hears “so it will work in me” has made the single most common mistake in this field. Telomeres are the right place to learn that lesson, because here the gap between the mouse and the human is not subtle. It is the whole story.

Where the machines come in

This is a book about how artificial intelligence is reshaping the search for medicines, so it is fair to ask where AI touches the telomere story. So far, mostly in measurement and scale. Telomere length can now be estimated computationally straight from genome-sequencing data, and machine-learning methods increasingly help analyze the noisy telomere assays and microscopy images researchers rely on. At the population end, the UK Biobank measured telomere length in the blood cells of roughly half a million people — which turned telomere biology into a large-scale machine-learning and genetics problem, letting researchers map the inherited variants that set telomere length and connect them, disease by disease, to risk.

But here is the honest framing, and it is the bridge to the next chapter. A better algorithm can sharpen the number. It cannot make the number mean more than it means. Run the most sophisticated model in the world on one person’s blood-cell telomere length, and you still have a noisy, weak, single-tissue predictor of how that person is aging. A precisely measured weak signal is still a weak signal. AI’s real leverage in aging will turn out to live mostly in finding better signals and better drug targets — not in squeezing certainty out of a measurement that never had much to give.

Which raises the obvious question. If telomere length is a noisy, low-resolution clock, is there a better one — a measurement that tracks biological age tightly enough to be worth acting on? There is a candidate, and it is built entirely out of machine learning. It is called the epigenetic clock, and it is where we go next — along with a trap that looks a great deal like the one in this chapter, only sharper.

✦ ✦ ✦

Carry these three things

1. A cell counts its divisions, and the counter is a disposable buffer at the end of each chromosome. Telomeres — TTAGGG repeated thousands of times — are sacrificed to the fact that linear DNA cannot copy its own ends, so that the genes stay safe. When the buffer runs critically short, the cell senses damage and retires. That is the Hayflick limit, in molecules.

2. The enzyme that keeps us young is the one cancer switches on to live forever. Telomerase can refill the buffer, but most body cells switch it off — because an unlimited division counter is an unlimited tumor. Aging and cancer protection are entangled, two ends of one trade-off. In this field, “just turn the clock back” usually means “remove a brake that was there for a reason.”

3. A measurement can be real across a population and nearly useless for one person. Telomere length tracks aging in a crowd of thousands and tells you very little about yourself: a noisy assay, a weak individual prediction, possibly the wrong tissue, and “longer” isn’t simply “better.” We can measure it is not you should act on it. That sentence is the discipline the rest of this book runs on — and the mouse that seems to “prove” telomeres drive aging is, read closely, the clearest warning of all that mouse biology is not human biology.

All 10 chapters of Will you live to 120? · Foundations · All courses