Biochemistry: A glimpse of molecular competition

Single-molecule studies reveal how the DNA-repair protein RecA overcomes competition from another protein to bind to single-stranded DNA, and how other mediator proteins assist in this process

Chromosomes consist of two interwound DNA strands millions of base pairs in length. These long molecules inevitably suffer breakage, which can be induced by ionizing radiation, biochemicals or natural DNA processing within cells. Double-strand breaks in DNA are not normally lethal to a cell, and can be accurately repaired by a process known as homologous recombination1, 2, during which a broken DNA fragment searches for and pairs with an intact strand from another DNA molecule that carries an identical (homologous) sequence. The search process is remarkable in two respects: the homologous partner can be found even among the many millions of non-homologous segments, and the crucial base sequence can be recognized even when it is largely buried within a DNA double helix.

In all domains of life, this extraordinary search and pairing process is made possible by a class of structurally related proteins of which the RecA protein from the bacterium Escherichia coli is the most-studied member. To initiate the recombination process, a filament composed of many RecA molecules must form on single-stranded DNA (ssDNA). But in doing so, the protein has to compete for binding with another resident protein, the ssDNA-binding protein (SSB)3. On page 274 of this issue, Bell et al.4 present single-molecule images of fluorescent RecA as it binds to ssDNA and extends to form a filament*. Although there have been other single-molecule studies5, 6, 7, 8, 9, 10, 11, 12 of RecA and of the analogous protein Rad51 from humans, this is the first study to visualize RecA binding to its natural substrate: extensively SSB-coated ssDNA. It is also the first to examine the effects of recombination 'mediators', proteins that potentiate recombination in vivo by aiding RecA-filament formation13, 14, particularly on SSB-coated DNA.

The process of RecA-filament formation is complex. The RecA filament has a dynamic structure, which means that molecules continually join and leave it. This is much like the filaments formed from the proteins tubulin and actin, which make up the 'skeleton' within cells. But the situation for RecA is more complicated than that for tubulin and actin because RecA filaments must form on a molecular scaffold, ssDNA, and compete with SSB for access to this scaffold. RecA binding to ssDNA occurs in two kinetically distinct phases: nucleation, which is the initial binding to ssDNA; and extension, in which additional RecA molecules are recruited to generate a polymeric structure, and which can occur at different rates at either end of the resulting filament.

Bell and colleagues' real-time observation of single RecA molecules binding to SSB-coated ssDNA allowed the nucleation step to be distinguished from the extension step. The authors report that, in the presence of ATP أ¢â‚¬â€‌ a nucleotide cofactor for RecA أ¢â‚¬â€‌ or its analogues, nucleating clusters of RecA formed randomly and slowly (over a time course of minutes to hours) on SSB-coated ssDNA. The nature of the dependence of nucleation on RecA concentration indicates that a RecA dimer is the nucleating species (Fig. 1). This makes sense, because a RecA dimer is the smallest RecA oligomer that has a functional ATP-binding site, which is constructed at the interface of two adjacent RecA molecules

The latest study also clarifies the role of nucleotide cofactors in RecA nucleation. Bell et al. observed that nucleation was considerably faster when ATP was replaced with the nucleotides ATPγS or deoxy-ATP (dATP). The cofactors are hydrolysed (converted from a triphosphate to a diphosphate form) by RecA, but the rates at which the various nucleotides are hydrolysed differ widely. The authors found that a lowery propensity for hydrolysis fails to correlate with a nucleotide's ability to promote RecA nucleation. These results support the idea that nucleotide binding induces RecA to adopt a conformation that has a high affinity for ssDNA, and that the nucleotides differ in their ability to induce this conformational transition: ATPγS is better than dATP, which is better than ATP.

The authors observed that, after initial RecA nucleation on SSB-coated ssDNA, RecA filaments grew at a rate comparable to those measured in studies of the protein in bulk16, 17, and with a linear dependence on RecA concentration. These findings are consistent with a growth mechanism in which monomeric RecA molecules are added to the end of the growing filament.

By performing nucleation using RecA molecules that had been tagged with a red fluorescent label, and then performing extension using RecA bearing a green fluorescent label, Bell et al. showed that the RecA filament extends at both ends أ¢â‚¬â€‌ that is, in both the 5′-to-3′ direction of the ssDNA and the 3′-to-5′ direction (Fig. 1). However, extension in the 5′-to-3′ direction was about 50% faster than that in the opposite direction. This preference has also been found previously in an electron-microscope analysis of RecA filaments18 and a study of RecA-mediated DNA-strand exchange16. These studies initially led to the notion that RecA filaments extend only in the 5′-to-3′ direction. Bell and colleagues' single-molecule experiments, along with those of others5, 12, clearly show that this is incorrect: net growth is 5′ to 3′, but growth occurs in both directions. And the authors found that, unlike RecA-filament nucleation, RecA-filament extension is broadly insensitive to the nucleotide cofactor used.

When the authors performed experiments at physiological pH, they found that filament nucleation and extension on SSB-coated ssDNA was extremely inefficient. This might reflect the situation in vivo, in which mediator proteins are needed to initiate recombination13, 14. The requirement for mediators may help to avoid inappropriate RecA assembly on ssDNA that is transiently exposed during DNA replication, thereby restricting RecA to bona fide recombination substrates.

But perhaps the most noteworthy breakthrough of Bell and colleagues' work was the direct observation of RecA-filament formation stimulated by the mediator proteins RecF and RecOR. It had previously been proposed13, 14 that the recombination mediator proteins RecF, RecO and RecR specifically allow RecA to overcome competition with SSB for binding to ssDNA, and it is therefore imperative to view mediator effects in the context of SSB-coated ssDNA. Bell and co-workers' study provides the first demonstration of the effects of mediators on single molecules, and offers the opportunity to distinguish between such effects on RecA-filament nucleation and extension. The authors found that RecF stimulates RecA-filament formation on SSB-bound ssDNA through nucleation, whereas RecOR stimulates both nucleation and filament extension.

Compared with previous single-molecule studies11, 12 of RecA binding to ssDNA, Bell et al. draw different conclusions about the oligomers involved in RecA nucleation and extension and the influence of nucleotide cofactors. It remains to be seen whether these differences are due to the presence of competing SSB or result from other features of the experimental system.

In Bell and colleagues' experiments, RecA presumably binds to ssDNA that is transiently exposed as SSB releases it or slides away (Fig. 1). RecA and SSB have distinctly different binding modes: each RecA monomer binds three nucleotides of ssDNA that are in a stretched conformation, whereas tetramers of SSB bind and wrap 65 nucleotides of ssDNA around their surface. RecA does not bind to SSB directly and must therefore replace SSB on ssDNA using a passive mechanism. By contrast, the mediator protein RecO does interact directly with SSB and may actively aid its removal19. But the exact mechanism by which RecO and other mediator proteins enhance nucleation or filament extension remains unclear. Future single-molecule imaging studies are therefore needed to identify the mechanisms by which RecA and Rad51 filaments are modulated by different proteins. This is not just of intellectual interest أ¢â‚¬â€‌ the human breast cancer type 2 susceptibility protein BRCA2 belongs to the same class of recombination mediator protein as RecFOR, and so an understanding of RecA/Rad51-filament modulation might aid our understanding of cancers that involve BRCA2 mutations.