Our Scientists

The cell's response to DNA damage is central to maintaining the integrity of the genome. Defects in the pathways that detect, signal and repair DNA lesions can lead to mutations, chromosomal abnormalities and cancer. Mutations in these pathways are associated with inherited cancer predisposition syndromes, and genetic instability is a hallmark of many types of sporadic cancers.

DNA damage is ubiquitous, caused by exogenous agents such as background radiation, ultraviolet (UV) light and environmental carcinogens, as well as by endogenous metabolites. The resulting DNA lesions are repaired by a handful of specialized pathways. Some lesions, such as simple base modifications, can be recognized and repaired readily, either by direct reversal or by base excision, with the missing information copied from the complementary strand. Most single-strand lesions, however, involve a wide range of chemically and structurally diverse modifications, and their detection can present a major challenge.

Nucleotide excision repair (NER) is the main pathway that repairs these lesions, which range from bulky adducts of environmental carcinogens to UV-induced intrastrand crosslinks. Lesions affecting both DNA strands, such as DNA double-strand breaks (DSBs) or interstrand crosslinks (ICLs), present additional challenges for repair, as the missing information has to be copied from the sister chromatid or the homologous chromosome by homologous recombination (HR). These lesions elicit a global cellular response characterized by cell cycle arrest, and by the activation of transcription programs, signaling networks and repair pathways.

Lesion Recognition in NER
A major question in NER research has been how diverse lesions are recognized by just two proteins, the xeroderma pigmentosum C protein (XPC) and xeroderma pigmentosum E protein (XPE). XPC is required for detection of all NER lesions, while XPE is additionally required for the initial sensing of a subset of lesions that evade XPC surveillance. By determining the crystal structures of XPC and XPE bound to damaged-DNA substrates, we find that both proteins recognize lesions that destabilize the Watson-Crick double helix over two base pairs.

XPC recognizes the lesion indirectly, by sensing the propensity of the DNA duplex to open up, with two complete base pairs flipped out and a large XPC β hairpin inserted through the duplex. For undamaged DNA, this open conformation has a substantial energetic penalty due to loss of base stacking and pairing. But if a lesion lowers this energetic cost so that it can be paid for by the XPC-DNA interactions, then it will be recognized as a lesion. This explains how certain lesions that do not destabilize the duplex can evade recognition. Additional XPC crystal structures reveal a non-specific DNA binding mode and associated conformational change that suggest a mechanism for XPC probing the DNA for lesions (Movie 1). XPE, by contrast, inserts a small β turn through the DNA duplex, and expels only the two damaged nucleotides, while the opposing bases remain stacked within the helix. The energetic penalty of inducing this DNA conformation is smaller, consistent with the heightened sensitivity of XPE for lesions that destabilize DNA only marginally. To achieve this, however, XPE interacts with lesion-specific features, such as a precompressed phosphodiester backbone characteristic of intrastrand crosslinks. Consequently, it can recognize only a limited set of lesions with these features.

DNA Double-Strand Break Repair
Our research in this area focuses on proteins involved in the sensing, signaling and repair of DSBs and related lesions. These include the ATM and ATR kinases, which are required for sensing DSBs and stalled/collapsed replication forks, respectively. ATM/ATR then phosphorylate and activate effector kinases and other substrates that bring about a global response. For repair by HR, the DSB is first resected to produce an overhang of single stranded (ss) DNA, followed by the coating of the ssDNA by the RAD51 recombinase. In earlier work, we showed that this step requires BRCA2, which binds to the resected DSB, recruits RAD51, and facilitates the nucleation of RAD51-ssDNA filament formation, the rate-limiting step. The RAD51-ssDNA filament then binds to donor double stranded (ds) DNA, searches for homology, and catalyzes strand exchange, the key reaction in recombination.

To address the mechanism of strand exchange, we determined a series of crystal structures of minifilaments of RecA (the prokaryotic RAD51 homolog), bound to the ssDNA substrate and also to the strand-exchanged heteroduplex product. We do not yet know the structure of RecA-ssDNA-dsDNA, due in part to its transient nature, but we have a rough model based on mutational data. Movie 2 shows an interpolation of three structures and one model, illustrates how we think the homology search and strand exchange proceeds. Key aspects of the strand exchange mechanism include (i) RecA holding the ssDNA in a conformation that resembles the dsDNA product, with the edges of the bases exposed for homology sampling by the incoming DNA; (ii) the donor dsDNA is melted locally by stretching and unstacking, freeing one of its strands for homology sampling, which occurs through Watson-Crick pairing; (iii) the ATP-dependency of DNA binding gives a means to release the DNA after the clock runs out (ATP hydrolysis). If strand exchange has occurred, the reaction is irreversible, otherwise, the filament reforms and samples homology with dsDNA at a new register.

Interstrand Crosslink Repair
Repair of ICLs requires the Fanconi anemia (FA) pathway of at least 13 proteins, as well as components of translesion synthesis, NER and HR pathways. The FA pathway is activated when a replication fork encounters an ICL and collapses. ATR then phosphorylates the downstream FANCI-FANCD2 complex and several other FA proteins. This is accompanied by the monoubiquitination of FANCI and FANCD2 by an ubiquitin ligase formed by seven FA proteins. Monoubiquitination is essential for repair, and it is required for the recruitment of at least two FA-associated nucleases.

The structure of the ~300 kDa FANCI-FANCD2 complex shows the two proteins are evolutionarily related, and they are made of α-solenoid helical repeats. The structure shows large basic grooves, one on each protein, that could accommodate dsDNA, and adjacent narrower basic grooves that could accommodate ssDNA. In vitro assays showed that the individual proteins prefer to bind to DNA structures containing both ssDNA and dsDNA. We confirmed this structurally with low resolution data from crystals of FANCI bound to ss-dsDNA. Based on the pseudo-twofold arrangement of the FANCI and FANCD2 DNA-binding sites, we propose the model that the complex recognizes twofold symmetric DNA structures resulting from the collapse of replication forks converging onto the ICL from opposite directions (Figure 1). Once there, the complex can direct the FA-associated nucleases to excise the ICL from one sister chromatid. The structure also showed that the monoubiquitination sites and activating phosphorylation sites map to the interface between the two proteins, and together with biochemical data suggested that these activating modifications cooperate to induce FANCI-FANCD2 association.

This work was supported in part by grants from the National Institutes of Health.

As of November 16, 2012