The Part of Your Chromosomes That Cannot Afford to Fail Is Also the Fastest-Mutating DNA in the Human Genome

A centromere is the pinch point on a chromosome — the place the cell's division machinery grabs to pull copies apart so each daughter cell ends up with the right amount of genetic material. Get it wrong and the cell miscounts chromosomes. It is one of the most safety-critical structures in biology, and until recently it was also one of the last large blind spots in the human genome.

A study published in Nature has now filled in a great deal of it, and the headline finding is not what a safety-critical structure is supposed to look like. Using long-read sequencing and custom-built computational tools, researchers assembled 2,110 complete centromeres from individuals spanning 28 population groups across five continents, then compared them against 5,747 centromeres already assembled by the Human Pangenome Reference Consortium. They found 226 major centromere haplotypes — distinct genetic patterns — and 1,870 repeat variants that had never been recognized.

Centromeres mutated faster than any other part of the human genome. The rate was not uniform: certain chromosomes accumulated changes at up to 20 times the rate of others. And within the centromere, the fastest-changing spot was the kinetochore attachment site — the exact anchor point where the chromosome latches onto the machinery that separates it during cell division. The most functionally constrained piece of the structure is the piece rewriting itself most aggressively.

The reason nobody had mapped this before is technical. Centromeres are built from long stretches of nearly identical repeated DNA called alpha-satellites. Standard sequencing chops DNA into short fragments and reassembles them by matching overlaps, which works well until every fragment looks like every other fragment. Long-read methods, which sequence far larger continuous stretches, are what made the region legible at all — and the study makes clear that widely used reference genomes are missing a substantial slice of real human diversity in these regions.

To watch the process rather than just photograph it, the team also sequenced a four-generation family, tracking how individual centromeres changed as they were passed from parent to child. That let them observe the mutations arising in real inheritance rather than inferring them from population comparisons.

The work has immediate practical edges. Centromere errors are implicated in chromosomal disorders and in the chromosome instability that characterizes many cancers, and a reference that omits most centromere variation cannot support that kind of diagnosis. The researchers also documented a recombined chromosome 13/22 in one cell line, confirmed by fluorescence imaging showing satellite probes from both parent chromosomes landing together in every cell examined — the kind of rearrangement that the old references were structurally incapable of resolving.