Repair of DNA double-strand breaks

DNA double-strand breaks are potentially lethal to cells and around ten double-strand breaks occur each day in dividing human cells. These breaks have a number of causes including replication through a single-strand nick, chemical damage from reactive oxygen species or ionising radiation, inadvertent enzyme damage (e.g. abortive topoisomerase activity), and mechanical stress. Mistakes made in the repair of DNA damage, including double-strand breaks, can lead to genomic instability and hence give rise to cancers.

There are two main avenues for the repair of double-strand breaks. The first is non-homologous end joining (NHEJ). NHEJ is used by cells to join broken ends by simply religating them. However, this process is not perfect and sometimes incorrect ends are joined to give rise to chromosome translocations. In addition, processing of ends at a “dirty” break (e.g. when the chemical moieties terminating at the break are not easily rejoined) or polymerase activity can give rise to mutations. Related NHEJ systems are found in both bacteria and eukaryotes although the bacterial system is a much simplified version and is not universal.

A second option for break repair involves homologous recombination (Figure 1). In this case, the information from a second intact DNA duplex is used to act as a template to ensure fidelity of the repair. Bacterial cells often contain multiple partially replicated chromosomes when they are dividing rapidly so a sister chromosome is usually available. However, in eukaryotes it is only in S-phase when the DNA is replicated prior to cell division that a sister is available and this is when homologous recombination takes place. The broken end(s) are processed and resected prior to loading of a recombinase (RecA in prokaryotes or Rad51 in eukaryotes). Strand invasion by the recombinase filament produces a four-way “Holliday” junction which is migrated and processed to repair the damage.

Figure 1 – Mechanisms of homologous recombination to repair double-strand breaks.
(A) In prokaryotes, a break arising from replication can be repaired via RecA-mediated homologous recombination. The broken end is processed by the RecBCD complex prior to loading of RecA.
(B) In eukaryotes, there are several outcomes and processes but some major pathways are shown here for illustration. The ends of the break are processed (for details see below) prior to homologous recombination mediated by Rad51. Panel B is modified from a review by Mazin et al., (2010).

Double-strand break repair in prokaryotes

A major mechanism for the repair of double-strand breaks in bacteria involves the 320KDa RecBCD complex. RecBCD comprises two Superfamily 1 helicase subunits (RecB and RecD) that operate with opposite polarities. The RecB subunit also has a nuclease domain that is regulated by the RecD subunit. The RecC subunit has the role of recognising a specific eight base sequence called Chi at which RecBCD loads the RecA protein to initiate homologous recombination. RecBCD processes the broken ends of the DNA with the final product being a 3’-tail on the duplex that is coated with RecA protein. The overall mechanism for this process is outlined in Figure 2.

Figure 2 – Processing of double standard breaks by the RecBCD complex
RecBCD complex binds to the double–stranded DNA end to form an Initiation complex (A). The RecB helicase/ nuclease is in orange, the RecC subunit is in blue and the RecD helicase is in green. The two helicase motors translocate along the DNA and the RecB nuclease domain digests the DNA duplex (B). Upon encountering the eight base Chi site (C) the complex pauses and the activities of the complex alter such that the 5’ end is resected to leave a 3’ tail. The final stage of the reaction is to load RecA onto the 3’ tail (D) to form a RecA filament that is proficient to initiate homologous recombination.

We determined the crystal structure of the RecBCD complex bound to a DNA substrate which represents Intermediate (A) in Figure 2. This structure (Figure 3) provided the first glimpses of how the protein complex processes broken DNA ends but several questions remain and so this project continues to be a major area of our research. We plan to understand more about each of the different intermediates in the reaction pathway from both a structural and biochemical perspective.

Figure 3 – Crystal structure of the RecBCD complex bound to a DNA substrate.

The RecBCD complex is the main system for double-strand break repair in Gram negative organisms, in Gram positive species there is a related complex called AddAB that performs a similar task. However, whereas RecBCD has two helicase subunits and a single nuclease, AddAB has a single helicase but two nucleases, resulting in a different mechanism for the control of DNA end resection. AddAB also contains an iron-sulphur cluster that is found in the eukaryotic helicase/nuclease Dna2 (see below). We have determined the crystal structure of AddAB bound to a DNA substrate (Figure 4).

Figure 4 – Crystal structure of the AddAB complex bound to a DNA substrate.

Double-strand break repair in eukaryotes

Although we have learned a lot (and hopefully continue to learn more!) about double-strand repair in bacteria from studying RecBCD/AddAB, recent studies on eukaryotes are producing interesting glimpses of the proteins that are involved in humans. The RecBCD/AddAB enzymes are not conserved in eukaryotes and this may lead one at first to assume that things are done differently in eukaryotes. However, the main players have been identified in the last few years and reveal a fascinating comparison with the bacterial systems. As might be expected, the bacterial system is simpler than that in humans but nontheless has many similarities. It turns out that all of the enzyme activities found in RecBCD/AddAB (with the exception of Chi recognition) are found in the proteins that repair double-strand breaks in humans but they are distributed in different ways (Figure 5).

Figure 5 – Comparison of some of the proteins involved in double-strand break repair in bacteria and humans. 
Similar enzymatic activities are represented by related coloured bars. Most enzyme activities are common to both systems but are created in different ways. RID refers to Recombinase Interaction Domain, either RecA or Rad51, depending on the system. This is contained within the same domain as the exonuclease activity in RecB and AddA. In addition to the enzymes shown, the human system also requires Replication Protein A (RPA), and the TopIII/Rmi1 and Mre11/Rad50/Nbs1 complexes.

Consequently, our understanding of how RecBCD/AddAB work and load RecA will have major implications for understanding the related process in humans. Since several of the players (e.g. BLM helicase and BRCA2 [Rad51-loader]) are tumour suppressors, then understanding these processes will also have implications for understanding how tumours arise from defects in the activities of these proteins. In addition to functional equivalents of RecBCD, the human system also includes Replication Protein A (RPA), a complex between topoisomerase III and Rmi1, and the Mre11/Rad50/Nbs1 (MRN) complex. This means that the resection of DNA ends requires complexes of proteins with a combined mass well in excess of 1MDa, depending on their stoichiometries which at present is unclear in many cases. Some of these individual components have already been crystallised and their structures determined. However, there are several sub-complexes of components that need be studied structurally and eventually we would like to be able to visualise the entire process. Recent work has shown that the system can be reconstituted in vitro  from proteins made in a variety of recombinant systems. This recent work also shows that the proteins form three separate, stable complexes that interact with one another to effect resection.

The first of these complexes comprises the Dna2 and BLM helicases together with the heterotrimeric RPA complex (Dna2/Sgs1/RPA in yeast). This complex forms a stable entity and is effectively the eukaryotic version of the RecBCD/AddAB systems (Figure 5). Dna2 consists of two domains; an N-terminal FeS cluster nuclease domain, that is homologous with AddB, and a 5’-3’ RecD-like helicase domain. The BLM helicase has 3’-5’ helicase activity like RecB, although the proteins are of different helicase superfamilies and, also like RecB, BLM has recombinase interacting domains (RecA for RecB and Rad51 for BLM). While each of the proteins has independent enzyme activity, these activities are regulated by physical interaction with one another and further still by interaction with RPA. This complex is able to carry out end resection in vitro. However, further stimulation of activity is observed in combination with two other protein complexes (Top3/Rmi1 and Mre11/Rad50/Nbs1 [Mre11/Rad50/Xrs2 in yeast]). Our current work aims to understand how these proteins cooperate to effect DNA double-strand break repair.