Telomeres: The Clock That Isn’t Quite a Clock

Every time a cell divides, the machinery that copies its DNA runs off the end of each chromosome and leaves a short stretch uncopied. It is not damage and it is not a design flaw — it falls out of how the copying enzyme works. The copy comes back shorter than the template, every time, and biologists call this the end-replication problem. Everything interesting about telomeres and ageing follows from that arithmetic — including the parts that do not work the way the popular story says.

Second in a series on the biology of ageing: gene editing, telomeres, epigenetics, the Blue Zones, and cellular reprogramming.

Why every copy comes back short

A chromosome is one long DNA double helix with two ends. To copy it, the cell pulls the two strands apart and builds a new partner strand against each.

DNA polymerase, the enzyme that does the copying, works under two constraints:

It builds in one direction only — written 5′ to 3′, after the two chemically different ends of a DNA strand.
It cannot start from nothing, only extend something already in place: a short RNA primer, laid down first, whose free 3′-OH group gives the polymerase a chemical foothold.

A replication fork treats the two parental strands differently. The strand whose 5′ end sits at the terminus is the leading-strand template: synthesis against it is continuous, right to the last base, and the product ends blunt.

The other parental strand matters. It is G-rich, its 3′ end sits at the terminus, and it is copied backwards — in short pieces called Okazaki fragments, each started by its own RNA primer.

Primers are RNA, not DNA, so each is removed again and its gap filled in — for a middle primer at no cost, since the fragment behind extends forward and fills it. The terminal primer has no fragment behind it, so no upstream 3′-OH to fill from. Its gap stays open and the new strand ends short.

Human telomeres absorb that loss: the same six letters over and over — TTAGGG — ending in a single-stranded 3′ overhang, carrying no genes. Losing terminal sequence every division is survivable because what is lost is spacer, not instructions. Up to a point. This part is established, textbook mechanism.

Stages 1 to 3: only the lagging strand ends up with a gap it cannot fill; stage 4 is the fix. Mechanism and the 80-90% figure from the Nobel Assembly’s 2009 prize summary (source 1).

Telomerase, and why most of your cells switch it off

One enzyme puts the sequence back. Telomerase carries a short piece of RNA of its own and uses it as a template, writing fresh telomeric repeats onto the 3′ end.

Blackburn and Szostak showed in 1982 that telomere sequence protects chromosomes. Greider, then a doctoral student in Blackburn’s lab, saw the enzyme activity itself on Christmas Day 1984. The three shared the 2009 Nobel Prize.

So the cell has the fix, and most human somatic cells — the ordinary body cells, everything that is not the germ line — repress it anyway.

Some 80 to 90 per cent of cancer cells have abnormally high telomerase activity. A tumour needs to divide without limit, and a division counter running down brakes exactly that. Repressing telomerase is a trade: give up unlimited renewal, keep a ceiling on any one clone.

The ceiling Hayflick found

Hayflick and Moorhead established in 1961 that human diploid fibroblast strains stop dividing after a finite number of population doublings. Fibroblasts are ordinary connective-tissue cells; diploid means the normal two sets of chromosomes, not a cancer line’s scrambled count.

The number was around 50 doublings, plus or minus 10. What the cells do next is nothing: they stay metabolically alive and refuse to divide again. Something counts, and something calls a halt.

Why a worn-down end looks like a broken chromosome

A cell cannot tell a chromosome end that has worn short from a chromosome that has snapped in half. Both present the same thing: a loose end of DNA.

A broken chromosome is an emergency: the cell stops dividing until it is dealt with, and stitches the loose end to any other it can find. Applied to a real chromosome end, that response fuses two chromosomes together — a catastrophe. So a healthy end must be marked, continuously, as an end rather than a break.

Shelterin does the marking: six proteins — TRF1, TRF2, POT1, TIN2, TPP1 and Rap1 — that bind the TTAGGG repeats and stop a chromosome end looking like a broken one. Two do the suppressing:

TRF2 suppresses ATM signalling, one of the two master alarm kinases of the DNA damage response, and blocks non-homologous end joining — the repair route that would fuse two ends together.
POT1 suppresses ATR, the other master alarm.

Shelterin needs repeats to sit on. When they run low, loading falls and the end starts to register as a double-strand break: the alarm goes off and the cell stops dividing. That is the Hayflick ceiling from the inside — replicative senescence is that damage response, not tape running out.

Where short telomeres genuinely cause disease

Here is where telomere shortening demonstrably makes people ill.

Armanios and colleagues screened 73 probands — the first person in each family to come to medical attention — from a familial pulmonary fibrosis registry in 2007. Six, 8 per cent, carried heterozygous mutations in hTERT or hTR and had short telomeres. Heterozygous means one broken copy out of two; those genes encode telomerase’s protein subunit and its RNA template.

The same genes underlie autosomal dominant dyskeratosis congenita — dominant meaning one broken copy is enough. It kills through aplastic anaemia, in which the marrow stops making blood, and through pulmonary fibrosis, in which the lungs scar.

In these telomere biology disorders the causal arrow is not in doubt: half the normal dose of telomerase is not enough.

How to test cause without running a trial

The harder question is whether telomere length matters for common disease in ordinary people. Measuring patients cannot settle it: illness, smoking and inflammation shorten telomeres themselves, so a short telomere in a sick person could be cause or consequence.

The way past that is Mendelian randomisation. Its logic runs in three steps:

Some common genetic variants nudge telomere length up or down a little.
Which of them you carry was settled at conception, by which copy of each chromosome your parents happened to pass on.
That allocation is unrelated to how you later live: your genotype was not chosen by your diet, job or income.

So the population is already split into groups dealt slightly longer or shorter telomeres for reasons connected to nothing else — what a randomised trial does deliberately, here done by inheritance at a scale no trial could fund. If the long-telomere group gets more of a disease, length is the plausible cause, not the consequence.

The assumption that makes it work is also where it can fail: the variants have to affect disease only through telomere length. Genes are rarely that single-minded.

Longer is not better

The Telomeres Mendelian Randomization Collaboration used 16 SNPs — single-letter differences in the genome — across 10 genomic regions as instruments: germline variants, fixed at conception, together explaining 2-3% of the variance in leukocyte telomere length.

They were tested against 35 cancers and 48 non-neoplastic diseases, in 420,081 cases and 1,093,105 controls. Results come as odds ratios per standard deviation of genetically higher telomere length: 1 means no difference, above 1 raised risk, below 1 lowered.

Genetically longer telomeres raised the odds of 9 of 22 primary cancers. Two of them:

Glioma, a tumour of the brain: 5.27 — odds a little over five times higher per standard deviation.
Lung adenocarcinoma: 3.19.

The same longer telomeres were protective elsewhere:

Interstitial lung disease: 0.09, roughly one-eleventh the odds.
Coronary heart disease: 0.78.
Abdominal aortic aneurysm: 0.63.

Read across the dashed line at 1: right is raised risk, left lowered. Longer telomeres push cancers right, fibrotic and vascular disease left. Odds ratios and 95% confidence intervals from the Telomeres Mendelian Randomization Collaboration (source 4).

The reading is uncomfortable but coherent. A longer starting telomere buys more divisions before senescence closes a clone down, and more divisions mean more chances to pick up a driver mutation — the kind that pushes a lineage towards cancer.

Note the tissue specificity. Lung adenocarcinoma comes in at 3.19; squamous cell lung cancer, in the same organ, at 1.07 — indistinguishable from no effect. Whatever is happening is not a general property of cells.

Strong, not settled. Two limits, both real:

Mendelian randomisation assumes the instruments act on disease only through telomere length. That is an assumption, not a finding.
Among the cancers, only glioma and bladder had independent replication datasets.

A separate line of evidence points the same way, and it is preliminary. In 2023 DeBoy and colleagues described families with germline loss-of-function mutations in POT1, one of the six shelterin proteins, disabled from birth in every cell. Their telomeres sat above the 99th percentile and failed to shorten with age.

What came with them was not a long healthy life:

Clonal haematopoiesis in 8 of 12 carriers tested — blood increasingly dominated by a few expanded mutant clones.
Benign and malignant neoplasms across several tissues.

Twenty-three carriers, from a few kindreds. That is a family study, not a population study.

Why a telomere test tells you almost nothing

Telomeres do shorten with age, and the curve is well described. Flow-FISH on 835 healthy people from birth to 102 gives median lymphocyte telomere lengths of:

11.2 kb at birth
10.0 kb at one year
7.9 kb at 18
4.3 kb at 102

A kb is a kilobase: a thousand base pairs, a thousand letters of DNA. Most of the loss is early:

About 1,190 bp per year below age one.
126 bp/yr from 1 to 18.
Only 43 bp/yr after 19.

The curve drops steeply in infancy, then is close to flat: adult ageing barely bends it.

Now put one person on that curve — what a telomere test claims to do. Two numbers decide whether that works.

The spread between people at any given age, 10th to 90th percentile: 2.45 kb.
What one person loses across forty adult years at 43 bp/yr: roughly 1.7 kb.

The gap between an ordinary short-telomere and an ordinary long-telomere person, on the same birthday, is wider than everything you will lose between 25 and 65. So one measurement mostly tells you which percentile you were born into.

Measurement makes it worse. Ten blinded laboratories measured the same DNA by three methods — Southern blot, STELA and qPCR. Their coefficients of variation, the spread of repeated results as a percentage of the average:

24.2% between laboratories, as a median.
20.7% among the qPCR labs, 9.2% for Southern and STELA.
1.9% to 4.6% within a batch inside one lab.

Read the last line against the first. Each lab reproduces its own answer to within a few per cent, and lands a quarter away from the lab next door: precise, mutually inconsistent. Rankings survive that; absolute values do not.

The authors’ conclusion is worth quoting: “the” telomere length of an individual does not exist as a measurable quantity.

One more limit: what is measured is leukocyte telomere length — the telomeres of white blood cells. Blood, not brain or heart or lung.

Telomerase in a capsule

In February 2018 the US Federal Trade Commission charged Telomerase Activation Sciences and its CEO with lacking scientific evidence for claims made for two products, TA-65MD and TA-65 Skin. The claims:

reverse ageing
prevent and repair DNA damage
restore ageing immune systems
increase bone density
prevent or reduce the risk of cancer

The settlement bars those representations unless they are supported by competent and reliable scientific evidence.

That is a finding about evidence, not mechanism. The deeper problem is direction: the best causal human evidence says longer telomeres lower cardiovascular and fibrotic risk and raise cancer risk. Any claim that a supplement lengthens human telomeres enough to change health is speculative.

What to take away

The end-replication problem is real, telomerase is real, and the disorders caused by too little telomerase are lethal. What does not hold up is the extrapolation: that telomere length is a readable clock for a whole organism, or that more is better. The Nobel Assembly said as much in its 2009 summary — the tallying-device picture “gives an incomplete picture of ageing”.

A division counter and a tumour-suppressive brake, not a dial to turn.