The mutational dynamics of the Arabidopsis centromeres.

Centromeres are essential for faithful chromosome segregation during cell division. Yet despite their conserved function, many centromeres contain highly variable but internally remarkably homogenized tandem-repeat arrays1-5 whose evolutionary dynamics remain poorly understood. Here, using replicated genome assemblies of mutation accumulation lines, we define the centromere-specific mutation spectrum in Arabidopsis thaliana. We find that kilobase-sized insertion-deletion mutations (indels) occur frequently and consistently preserve tandem-repeat arrays by adding or removing only complete repeat units. Point mutations accumulate at an almost tenfold higher rate than elsewhere in the genome, probably driven by non-allelic gene conversion between closely linked repeat units. These findings suggest a central role for homology-directed DNA repair in centromere evolution, further supported by the accumulation of more frequent and longer tandem-repeat-preserving indels in Arabidopsis lines that are deficient in the anti-recombinase helicase RTEL1. Forward-in-time simulations parameterized with the observed mutation spectrum show that kilobase-sized indels and point mutations alone are sufficient to generate the megabase-sized homogenized repeat blocks characteristic of natural centromeres. Together, our results show that centromere evolution is driven by a distinct mutational spectrum shaped by homology-directed DNA repair, providing a quantitative framework for understanding how mutational processes generate and maintain the large-scale architecture of centromeric DNA.