How do DNA duplexes pair up and stay linked? This togetherness is necessary for proper cell maintenance and division, such as during homologous recombination, but the mechanisms are not well understood. Now, in a recent paper (Hopfner, K.-P. et al. Nature 418, 562–566; 2002), researchers suggest how a set of proteins called the Mre11 complex may crosslink DNA.

Mre11 (a double-strand break repair nuclease) and Rad50 (an ATPase) are the primary components of the Mre11 complex, and both are found in all kingdoms of life. They form a heterotetramer in 2:2 stoichiometry and appear to function together in numerous processes including telomere maintenance, synaptonemal complex formation, meiotic DNA processing, non-homologous DNA end joining, and recombination.

Hopfner et al. focus on Rad50, which contains a long coiled coil. Near the middle of this region, there is a CXXC motif thought to be involved in protein–protein interactions. To test this hypothesis, the researchers analyze portions of Rad50 that include this sequence.

In solution, these Rad50 polypeptides readily form dimers — but only in the presence of zinc. An X-ray crystal structure shows that, in the dimer, each monomer (orange or yellow in the left panel) forms an antiparallel coiled coil, with the CXXC motif at the apex.

The structure confirms that the CXXC motif mediates Rad50 homodimerization. Four Cys residues (two from each Rad50 polypeptide) chelate a zinc atom, and this assembly is stabilized by additional interactions. Mutational studies provide in vivo evidence for the importance of this zinc-mediated partnership. When the first Cys in the CXXC Rad50 motif is mutated to Gly, deleterious phenotypes are observed even though the mutant protein can still bind to its Mre11 partner.

Doubling up of the Rad50 coiled-coil domains may provide a mechanism for linking DNA duplexes in the flexible fashion needed to carry out the diverse processes mediated by the Mre11 complex. In vivo, each coiled-coil domain is connected to the ATPase domain in the same polypeptide. The ATPase domains could bind to Mre11 proteins, which could interact with DNA (as depicted in the right panel). Hopfner et al. use electron microscopy to explore this idea further.

They image Mre11 complexes with and without zinc. With the metal, they see many dumbbell-shaped particles that seem to be hooked together in the middle (center panel). The ends of the dumbbell are interpreted as Rad50 ATPase domains bound to Mre11 nucleases (drawn as blue circles in the right panel) and the center portion is presumed to be pairs of coiled coils hooked via zinc in a limber tail-to-tail manner (as illustrated in the right panel). They estimate that the length of the center region of the human Mre11 complex dumbbell is ∼1,200 Å, which, nicely, is the predicted distance between sister chromatids in vivo.