Centromeric origin of Arabidopsis interstitial telomeric repeat domains

The mechanisms underlying the emergence of new DNA repeat families remain poorly understood. Interstitial telomeric repeat (ITR) domains compose a class of repetitive elements associated with genomic instability and disease that illustrates this knowledge gap. Commonly viewed as telomere-fossil byproducts of chromosome rearrangements, their origins and evolutionary dynamics have long been obscured by their highly repetitive nature. Here, combining a pangenomic survey of Arabidopsis thaliana with forward simulations of centromere evolution, we show that megabase-scale ITR domains arise in situ from canonical centromeric repeat arrays. Mutation-driven gains of telomeric motifs within centromeric satellites produce centromeric-telomeric mosaic satellites, which expand into higher-order structures via head-to-tail amplification. Despite their proximity and sequence similarity to centromeres, ITR satellites adopt an atypical chromatin state that excludes CENH3 incorporation, enabling escape from the functional constraints of centromere identity. Accordingly, ITR abundance varies abruptly among natural accessions, independently of telomere length, population relatedness or environmental variables, often via transposon-associated block duplications. Collectively, this study identifies centromeres as an unrecognized source of telomeric sequences and reveals an unexpected route shaping the rise of new satellite types.