The carefully orchestrated recruitment over time of a large number of repair and checkpoint proteins to the site of DNA damage was studied by Rodney Rothstein and colleagues in live yeast cells. By visualizing fluorescently tagged checkpoint and repair proteins in a series of mutant strains, they identified which proteins were recruited in response to DNA double-strand breaks (DSBs) or replication stress, and dissected the order of their recruitment.

The MRX complex (which comprises Mre11, Rad50 and Xrs2) is the first to be detected at a DSB site, which confirms its status as the DSB sensor. The Tel1 kinase is recruited next, followed by replication protein A (RPA), which recognizes and protects single-stranded DNA ends. In turn, RPA directs the recruitment of several checkpoint proteins including the Rad24–Rfc2–5 clamp loader, the Ddc1–Mec3–Rad17 clamp, the Mec1–Ddc2 kinase, the Rad9 mediator and the Rad53 kinase.

The actual DNA-repair proteins — the homologous recombination (HR) machinery — are recruited last, and only in the S and G2 phases. Their recruitment depends on Rad52 and coincides with the disappearance of MRX from the site of repair. Sae2 seems to faciliate this transition from the damage-recognition phase to the repair phase, as the disassembly of MRX and the recruitment of recombination proteins are both delayed in an sae2 mutant strain.

Only a subset of the proteins that are recruited to DSBs also assemble at stalled replication forks, as MRX and Rad52 (and, therefore, all other recombination proteins) are lacking. However, when replication forks collapse, MRX and Rad52 are recruited, which, presumably, is triggered by exposed DNA ends.

These new insights raise an intriguing question: why do some proteins (MRX, RPA and others) bind to the DNA-repair site irrespective of the cell-cycle phase, whereas the recruitment of others (the HR machinery) is restricted to the S and G2 phases?