Recombination and Repair

Three-dimensional structure of a Holliday junction, a central intermediate in homologous recombination.

Genome stability related to BRCA2 defects

Our main interest how homologous recombination repairs double-strand breaks during S-G2 phase of the cell cycle when the chromosomes have undergone replication continues to progress well.  We know that efficient recombinational repair requires a number of proteins including RAD51, RAD52, RAD54, the RAD51 paralogs (RAD51B, RAD51C, RAD51D, XRCC2, XRCC3), BRCA1 and BRCA2.  We have been successful in purifying many of these proteins and biochemical studies to determine their DNA substrate specificities are in progress.

Of particular importance in recombinational repair is RAD51, a homolog of the bacterial RecA protein, which catalyses the key reactions required for DNA pairing and strand exchange.  In response to DNA damage RAD51 localises to distinct sub-nuclear assemblies, known as foci, where the repair reactions are thought to take place.  The localisation of RAD51 to repair foci is dependent upon the breast cancer-associated tumour suppressor BRCA2, with which RAD51 interacts.  The current hypothesis is that BRCA2 controls RAD51 activity.

Individuals carrying mutations in BRCA1 or BRCA2 are predisposed to breast and/or ovarian cancers. Approximately, 20% of breast cancers are inheritable, and of these about one third have been linked to mutations in BRCA1 or BRCA2. In the United Kingdom alone, there are approximately 80,000 predisposed individuals of whom more than 70% will develop cancer. One defective copy of BRCA1 or BRCA2 in the genome is sufficient to confer cancer predisposition, and the loss of the second allele is commonly observed in tumour cells isolated from predisposed individuals. BRCA2 is a very large protein (mass 384 kDa) that contains a series of eight degenerate motifs, of which six have been shown to bind RAD51. These motifs, known as the BRC repeats, are approximately 30 amino acids long and are interspersed along a 1200 amino acid central region of the protein encoded by exon 11. Additionally, there is an unrelated RAD51 binding site located at the C-terminus of BRCA2. Studies in the mouse have shown that deletion of the very C-terminal region (exon 27) of BRCA2 confers a classical BRCA2 recombination- and repair-deficient phenotype and affects the localisation of RAD51 to repair foci in response to DNA damage, despite the fact that all the BRC repeats remained intact. Recently, we found that the RAD51 interaction domain at the C-terminus of BRCA2 contains a site (Serine 3291) that is phosphorylated during the cell cycle by cyclin-dependent kinases, or is dephosphorylated in response to DNA damage. Remarkably, phosphorylation of S3291 appears to act as a switch that regulates interactions of this region with RAD51. Activation of this site is therefore a critical modulator of the recombinational repair pathway (Esashi et al., Nature 434, 598-604 [2005]).

DNA repair defects and neurodegenerative diseases

Recently, our work defined the molecular defect associated with a neurological disorder known as Ataxia with Oculomotor Apraxia-1 (AOA1). We found that the product of the Aptx gene, Aprataxin, which is defective in individuals with AOA1, acts as a proofreader for abortive DNA ligation reactions (Ahel et al., Nature 443, 713-716 [2006]) . All cells, especially neuronal cells, are subjected to high levels of oxidative stress resulting in the formation of DNA strand breaks. When these breaks are repaired by a DNA ligase, it is not uncommon for the reaction to stall at an intermediate stage, such that ‘abortive ligation intermediates’ accumulate. It was found that Aprataxin specifically interacts with these intermediates and removes the AMP residue that ligase leaves covalently bound to the 5’-side of the unrepaired nick after abortive ligation.

The loss of Aprataxin activity is particularly evident in neurological tissue that, due to its non-proliferative nature, is unable to utilize alternative (replication-associated homologous recombinational repair) mechanisms to remove these lesions. Without Aprataxin, unrepaired nicks with 5’-AMP moieties will accumulate in the neuronal tissue over a period of years until they are sufficiently numerous to pose a significant block to transcription. Many neuronal cells will then undergo apoptosis and cell death leading to the characteristic neurological features associated with the disorder AOA1. Thus, Aprataxin plays a critically important cellular role in guarding the genome against DNA damages that would otherwise pose a block to normal cellular processes, illustrating the link between defective DNA repair and human disease.  

Involvement of RAD51C and XRCC3 in Holliday junction resolution

For many years, we have been interested in understanding how recombination intermediates are processed in somatic human cells. By fractionating cell-free extracts we were successful in enriching for a protein complex that catalyses the two critical Holliday junction-processing reactions, known as branch migration and resolution, but the identities of the protein components of this complex have remained elusive despite a decade of effort.

Recent studies in yeast implicated a role for Mus81/Eme1 in Holliday junction resolution in meiotic cells. These data stimulated us to clone the human homologs of Mus81 and Eme1 (Ciccia et al., J. Biol. Chem. 278: 25172-25178) and to compare their biochemical properties with the partially purified Holliday junction resolvasome fraction from somatic cells. We found that Mus81/Eme1 was actually a very poor Holliday junction resolvase, a result consistent with the work of others suggesting that D-loops or flap-like structures, rather than Holliday junctions, may be the true target of the Mus81/Eme1 complex during meiotic recombination.

In a significant step forward, however, we were recently successful in identifying RAD51C as the first component of the human Holliday junction resolvasome (Liu et al., Science 303: 243-246). This protein is a member of a family of proteins related by sequence to RAD51 and known as the RAD51 paralogs. RAD51C is present in two cellular complexes; the first contains RAD51B-RAD51C-RAD51D and XRCC2, while the second contains RAD51C with XRCC3. All five proteins are required for normal levels of recombination and repair. Using in vitro assays with partially purified resolvasome fractions, we showed that RAD51C was necessary for both branch migration and Holliday junction resolution. Moreover, extracts prepared from RAD51C (irs3) or XRCC3 (irs1SF) cells were shown to contain reduced levels of Holliday junction resolvase activity.

These observations provided a breakthrough that should open the door to an understanding of the mechanisms of Holliday junction processing in human cells. However, the functions of these proteins may not be restricted simply to Holliday junction processing, since XRCC3 plays a role in the modulation of replication fork progression following DNA damage (Henry-Mowatt et al., Molec. Cell 11: 1109-1117), and RAD51D has been visualised at the ends of chromosomes where it is thought to be required for recombination-mediated telomere maintenance (Tarsounas et al., Cell 117: 337-347). It is hoped that our future studies will enable us to define the precise roles of these proteins in recombinational repair, and to gain further insight into how recombination is required for replication fork stability and telomere maintenance.