Cohesins (black rings) near the centromere (yellow) hook a chromosome loop to itself. DNA can expand (red lines) as microtubules (green) shorten and tug on the loop.


DNA is a mechanical piece of the spindle, if results from Elaine Yeh, Kerry Bloom, and colleagues (University of North Carolina, Chapel Hill, NC) are any indication. The inherent springiness of DNA may be part of the spindle's tension-sensing mechanism.

During mitosis, tension between chromosomes indicates that each of two sister chromosomes is properly attached to its own spindle pole. The creation of tension requires a protein called cohesin, which glues sisters together until its cleavage at anaphase. But lots of cohesin is seen even at sister centromeres that are widely separated.

Bloom and colleagues set out to understand what cohesin is doing at these locations, which seem too far apart for cohesin to hook sisters together. They imaged dividing yeast cells containing labeled cohesin and found a cylinder of fluorescence encircling the entire central spindle, spanning sister centromeres.

The arrangement suggested that cohesins near a centromere were hooking a chromosome to itself rather than to its distant sister. In this model, the links would cause the centromeric region to loop out from the rest of the chromosome. The group looked for such loops by cross-linking DNA in its in vivo conformation and performing inverse PCR. The resulting products indicated that the centromere is at the apex of a ∼20-kb DNA loop.

The loops might act like a spring while microtubules at the centromeres grow and shorten during spindle assembly and bipolar attachment. “We are arguing that the centromere is dynamically unstable,” says Bloom. “When microtubules shorten, the loop stretches; when they grow, the loop contracts.”

Loops might be another place, along with kinetochores, where tension is sensed. “DNA is inherently an entropic spring, like a rubber band,” says Bloom. “If you have a rubber band, and you need to build a tension-sensing mechanism, you might as well use the rubber band.”

Yeast centromeres are only 120-bp long, but the entire loop might be a more accurate equivalent of the much larger mammalian centromeres, which have also been seen to form loops.


Yeh, E., et al.
Curr. Biol.