Ancient co-option of LTR retrotransposons as yeast centromeres

How selfish DNA became essential

Every time a cell divides, it must hand out one complete set of chromosomes to each daughter cell. This hand‑off depends on tiny structures called centromeres, which act like molecular handles to pull chromosomes apart. In brewer’s yeast and its close cousins, these handles are unusually small and precisely defined, and biologists have long wondered how such streamlined, hard‑wired centromeres evolved from the larger, more flexible forms seen in most other organisms. This study uncovers an unexpected answer: pieces of once‑selfish jumping DNA were repurposed over hundreds of millions of years into the very sites that now guarantee faithful chromosome inheritance.

From broad landing zones to pinpoint anchors

In many plants, animals and fungi, centromeres are broad, repeat‑rich stretches of DNA whose identity is set more by specialized proteins than by the exact underlying sequence. Yeasts in the group that includes baker’s yeast are different: each chromosome carries a tiny, roughly 125‑base‑pair “point centromere” whose sequence is rigidly specified and can attach to only one spindle fiber during cell division. Because such point centromeres are found in just one small branch of the tree of life, researchers suspected they evolved from older, repeat‑based forms, but the intermediate steps were missing. The authors turned to closely related yeasts whose centromeres were unknown, reasoning that these species might still carry telltale transition stages.

Discovering the halfway houses

Using chromosome conformation capture (Hi‑C), chromatin mapping and functional tests, the team charted centromere positions in several apiculate, or lemon‑shaped, yeasts. They found compact regions where a single centromere‑specific nucleosome sits over a short, A‑and‑T‑rich DNA core, flanked by short sequence motifs that are important for centromere function but arranged in a relaxed, flexible way. These sites can drive stable inheritance of plasmids, confirming that they function as genetic centromeres, yet they lack the strict three‑part layout seen in classic point centromeres. The authors dubbed them “proto‑point” centromeres: sequence‑encoded, single‑nucleosome anchors that still tolerate variation in their flanking elements.

Jumping DNA clusters as the missing link

The story became more surprising in species whose centromeres sit inside dense tracts of long‑terminal‑repeat (LTR) retrotransposons, especially an element called Ty5. Retrotransposons are pieces of DNA that copy and paste themselves around the genome; they are usually classed as selfish, yet here they mark—and in some cases make up—the centromeric regions. By comparing multiple strains and related species, the authors showed that Ty5 elements have occupied these centromeric neighborhoods for tens to hundreds of millions of years, continuously inserting, decaying and reshaping the local sequence while the centromere position itself stayed conserved. Across many yeast lineages, genes that sit near point centromeres today are also commonly linked to Ty5‑rich centromeres in more distant relatives, implying that Ty5‑clustered centromeres were already present in a shared ancestor.

Recycling selfish code into precise control

Delving into the sequences, the researchers found that the hallmark motifs of modern point centromeres—an A‑and‑T‑rich central core and two specific flanking elements—resemble patterns encoded within Ty5 LTRs. These LTRs are enriched for binding sites recognized by transcription factors that later became core centromere‑binding proteins, suggesting that early interactions between proteins and Ty5‑derived DNA laid the groundwork for a more hard‑wired centromere. Over time, as ancestral forms of the centromere‑recognizing complex (CBF3) took shape and machinery for traditional heterochromatin was lost, selection appears to have favored centromeres that relied less on broad epigenetic marks and more on precise DNA–protein partnerships. This gradual tightening of both sequence and protein architecture culminated in the rigid, three‑part point centromeres of modern budding yeast.

What this means for chromosome inheritance

The work provides a mechanistic route for how a “softly defined” centromere, maintained largely by chromatin state, can be converted into a “hard coded” one whose activity is specified by exact base pairs. In this scenario, ancient clusters of Ty5 retrotransposons first colonized ancestral centromeres, then slowly donated sequence motifs that could be recognized by evolving centromere proteins. The resulting co‑evolution between selfish DNA, chromosomal structure and protein machinery turned once‑parasitic elements into indispensable parts of the segregation apparatus. For a lay reader, the key message is that genomes are not just static instruction manuals: they are dynamic ecosystems where even genetic hitchhikers can, over deep time, be reshaped into vital components that keep our chromosomes—and our cells—running reliably.

Citation: Haase, M.A.B., Lazar-Stefanita, L., Baudry, L. et al. Ancient co-option of LTR retrotransposons as yeast centromeres. Nature 651, 1004–1011 (2026). https://doi.org/10.1038/s41586-025-10092-0

Keywords: yeast centromeres, retrotransposons, genome evolution, chromosome segregation, Ty5 elements