One resolution that's always broken is the conversion of Holliday junctions into linear duplex products. These four-stranded DNA crossover structures (see illustration), which are formed during DNA recombination and repair, are 'resolved' by the simultaneous introduction of breaks into two opposite strands. Several structure-specific endonucleases — 'resolvases' — are known to do this job in prokaryotes, but similar activities have been surprisingly difficult to pin down in eukaryotic cells.

Enter MUS81. Two new papers show that this endonuclease is an essential component of a resolvase activity in fission yeast and human cells. What's more, MUS81 seems to have evolved specifically to cope with the problems caused by stalled replication forks.

Fission yeast mus81 was first described as a protein that associates with the replication checkpoint kinase cds1. It is related to the XPF subunit of the XPF–ERCC1 complex — a structure-specific endonuclease involved in nucleotide excision repair. Because XPF acts as part of a complex, Paul Russell and colleagues wondered whether the same might be true for mus81, and they report in Cell that it is.

Using two different approaches, Russell and co-workers identified eme1 (for 'essential meiotic endonuclease 1') as a new binding partner for mus81. Genetic analyses revealed that mus81 and eme1 act in the same pathway of resistance to UV damage, and that they are also required during meiosis. These studies indicated that the meiotic defect might be due to failure of the chromosomes to segregate properly, which could, in turn, be due to unresolved recombination intermediates — such as Holliday junctions.

To test this possibility, the authors expressed a bacterial Holliday junction resolvase called RusA in mus81 mutants. This was enough to correct the meiotic defect in most cases, and it was dependent on the endonuclease activity of RusA. Consistent with this, the predicted endonuclease active site of mus81 was found to be essential for activity.

Finally, Russell and colleagues used synthetic Holliday junctions (made by annealing oligonucleotides) to show that an affinity-purified mus81–eme1 complex acts as a resolvase in vitro. Like the previously characterized bacterial resolvases, mus81–eme1 introduces paired incisions on opposing strands of the X-structure to form linear duplex products. However, unlike them, it cuts only 5′ to a double-strand/single-strand junction — the bacterial resolvases have no such requirement.

Russell's group next teamed up with Clare McGowan and co-workers to clone the human homologue of mus81. They report in Molecular Cell that, like the yeast protein, human MUS81 interacts with CDS1. Levels of MUS81 are increased in cells that are exposed to DNA-damaging agents (γ-irradiation or UV) and to hydroxyurea, which interrupts DNA replication. Given that some UV-induced lesions are also thought to block progression of replication forks, the observed increase in MUS81 could, say the authors, indicate “a role in cellular responses to blocked DNA replication”.

The authors next asked whether, like its yeast counterpart, MUS81 has a structure-specific endonuclease activity. They incubated synthetic junctions with MUS81 immunoprecipitates from HeLa cells, and observed the formation of defined cleavage products. The pattern of cleavage was consistent with the hypothesis that MUS81 cleaves on opposite strands, close to the double-strand/single-strand junction. But as MUS81 elutes from a gel-filtration column at a higher molecular weight than the monomeric protein, the authors believe that it probably functions as a heterodimer. Identifying its partner — possibly a human eme1 homologue? — will be one of the next steps.