The Keystone Symposium on Genome Instability and DNA Repair took place between 1 and 6 March 2009, in Taos, New Mexico, and was organized by M. Jasin, S.J. Elledge & M.S. Neuberger.
See Glossary for abbreviations used in this article
DSBs occur when both strands of a DNA double helix are damaged by either the normal metabolic processes of the cell, or environmental factors such as ultraviolet light or radiation. Cells have evolved mechanisms to repair DSBs, to ensure that large pieces of the genome are not lost; however, the aberrant repair of DSBs can lead to genome rearrangements that scramble the information encoded in the DNA. The 2009 Keystone Symposium on Genome Instability and DNA Repair featured mechanistic studies of DSB repair, host responses to DSBs and recent phenomenology in human genomic disorders. This report highlights some of the key findings presented at the meeting. Space limitations have prevented us from discussing every contribution, and we apologize to those whose work is not highlighted in this report.
DSBs: formation, repair and signalling events
One way to understand the consequences of aberrant DSB repair is to identify and characterize genomic rearrangements. M. Stratton (Cambridge, UK) used second‐generation whole‐genome DNA‐sequencing technologies to characterize the most common types of DNA rearrangement in a range of cancer cell lines and primary tumours. In addition to finding a large number of point mutations in these samples, preliminary analyses of the data indicate that many rearrangements result from intrachromosomal events, and, less frequently, interchromosomal breakage and rejoining events. Importantly, many of these rearrangements had not been seen previously because they are too small to be detected by conventional karyotype analyses. The analysis of junction sequences by the Stratton group also revealed that classical repair processes such as HR and NHEJ, probably have a role in the formation of these rearrangements. However, he also observed a distinct pattern of insertion events, which involved nearby, unlinked or unknown segments of DNA that might result from non‐classical DSB repair events such as single‐strand annealing, template switching or break‐induced replication. Therefore, the mutations and DNA rearrangements that occur in tumour cells are more varied and complex than previously thought.
Although intriguing, Stratton's findings are based on inference. Yet, understanding DSB repair necessitates learning more about what happens to these lesions soon after they arise. Important contributions towards this goal came from L. Symington (New York, NY, USA) and G. Ira (Houston, TX, USA), who presented studies on the MRN complex, which has a 3′–5′ exonuclease/endonuclease activity in vitro that seems to be incompatible with the proposed 5′–3′ exonuclease activity that is needed to promote 3′‐overhang formation in vivo (Mimitou & Symington, 2008; Zhu et al 2008). Symington and Ira demonstrated that this paradox might be resolved by the activity of the SGS1 helicase in vivo, which unwinds DNA and displaces single‐stranded DNA in the 5′–3′ direction. The single‐stranded DNA is processed initially by the MRN complex and then by the DNA2 exonuclease/endonuclease, which acts redundantly with ExoI to degrade the 5′ strand (Fig 1). Their findings seem to have resolved a longstanding question in DSB processing and explain how 3′ overhangs are generated for use in HR‐based DNA repair.
Another clue about the role of the MRN nuclease came from D. Ferguson (Ann Arbor, MI, USA). Ferguson used knock‐in technology to create mice containing a nuclease‐deficient allele of Mre11 (Buis et al, 2008). Remarkably, these mice exhibited phenotypes that were indistinguishable from those of mice lacking the Mre11 gene, including early embryonic lethality and genomic instability; however, in contrast to Mre11‐knockout cells, ATM signalling is not affected in MRE11 nuclease‐deficient cells. Therefore, unlike simple eukaryotes such as Saccharomyces cerevisiae, higher eukaryotes probably have a low tolerance for DSBs and use a range of recombination‐based processes to repair these lesions rapidly in order to avoid cell death.
HR has been studied extensively and involves the formation of a Holliday junction (HJ) intermediate. The symmetrical cleavage of the HJ by an enzyme known as resolvase allows for the cleavage of the resultant duplexes (Liu & West, 2004). S. West (South Mimms, UK) reported the identification of the human (GEN1) and yeast (Yen1) HJ resolvases (Ip et al, 2008). Unlike previously identified nucleases—for example, Mus81–Mms4/EME1—GEN1 and Yen1 cleave model HJ substrates at symmetrical positions, yielding nicked duplexes that can be ligated without further processing. The identification of GEN1/Yen1 is a crucial breakthrough that will make its mark on many processes that involve HR‐based repair.
Several talks also addressed how DSBs are formed during the normal cell cycle. Torsional stress can occur during DNA replication‐fork elongation, especially when two replicons fuse together at termination or when the replication forks encounter transcription units. The topoisomerase‐mediated resolution of these constraints then allows the completion of S phase, and subsequent chromosome condensation and segregation. However, interfering with this process can result in entangled DNA molecules and DNA breaks, leading to intra‐S‐phase DNA lesions and/or segregation abnormalities. Using genomic approaches, M. Foiani (Milan, Italy) characterized regions where replication termination occurs in S. cerevisiae and identified cellular pathways that are required for the resolution of two converging forks. Importantly, these replication‐termination zones might represent replication‐risk elements that are prone to genome rearrangements, unscheduled recombination events and even DNA breakage.
The efficient activation of the DNA‐damage response might also be important for inducing chromatin modifications that promote an effective repair of DSBs. Following a DSB, ubiquitination of the damaged chromatin attracts BRCA1 complexes. How BRCA1 acts at the sites of DNA damage remains ill defined. The Ludwig laboratory (New York, NY, USA) showed that the ubiquitin ligase activity of BRCA1 is not essential for HR or the response to DNA‐damaging agents, and therefore seems to be dispensable for DSB repair (Reid et al, 2008). However, it is clear that BRCA1 forms distinct subcomplexes. S. Elledge (Boston, MA, USA) reported the identification of a new component of the BRCA1 A‐complex, NBA1, which is required for the recruitment of BRCA1 to sites of DNA damage. NBA1 also has a crucial role in ionizing‐radiation resistance and cell‐cycle control. Intriguingly, analyses of NBA1 and the other components of the BRCA1 A‐complex revealed marked similarities to the lid complex of the 26S proteasome (Wang et al, 2009). Does the BRCA1 A‐complex recruit the 26S proteasome to DSBs? If so, this groundbreaking finding is likely to shed light on the role of BRCA1 in DNA repair and cancer.
Chromosomal translocations are also a characteristic of many cancers. To identify proteins that affect translocation, M. Jasin (New York, NY, USA) used the rare endonuclease I‐SceI to introduce DSBs at two different chromosomal loci in either wild‐type mouse cells or mutants lacking proteins that are important for NHEJ—such as XRCC4 and Ku70 (Weinstock et al, 2007). Jasin reported a three‐ to fivefold increase in translocations in both Xrcc4‐deficient and Ku70‐deficient cells, and found that the translocation frequency could be suppressed to wild‐type levels after the reintroduction of XRCC4 and Ku70, respectively, into the mutant cells. The sequence characteristics of the translocation junction—which include microhomology—were similar for the wild‐type and mutant cell lines, pointing to the involvement of alternative pathways of NHEJ in translocation. Jasin also reported on a new technique that uses engineered zinc‐finger nucleases to induce and quantify translocations in human cells without the need for cell cloning. By using this system, Jasin could determine that the aberrant repair of the DSBs leads to translocations—as occurs in mouse cells—approximately one‐half of which show microhomology at the breakpoint junctions (Brunet et al, 2009). Therefore, this system can be used to analyse the role of DNA‐repair proteins in human cancers.
DNA repair in lymphocytes
A characteristic of B lymphocytes is their ability to recognize a vast array of pathogens. This diversity is achieved, in part, by V(D)J recombination and CSR (Fig 2). Both processes proceed through DSB intermediates and are among the rare instances in which DNA lesions are deliberately introduced at defined genomic regions. The regulation of the enzymes that generate these lesions and of the proteins that repair the resultant DSBs were discussed in depth at this meeting.
Artemis is an endonuclease that cleaves hairpin intermediates at coding ends during V(D)J recombination (Sekiguchi & Ferguson, 2006). Inactive alleles of Artemis lead to severe combined immunodeficiency, whereas hypomorphic alleles of Artemis can lead to immunodeficiencies of varying severity (Revy et al, 2005). J. Sekiguchi (Ann Arbor, MI, USA) generated knock‐in mice that phenocopy a premature‐truncation mutation in the human Artemis protein. The mice had a severe reduction in B and T lymphocytes, and a pronounced accumulation of hairpin‐coding ends. The mutant cells also exhibited DSB‐repair defects, indicating that the carboxy terminus of Artemis is essential for V(D)J recombination and the maintenance of genomic stability (Huang et al, 2009).
CSR occurs between repetitive switch regions, which precede each immunoglobulin constant‐region gene (Chaudhuri et al, 2007). The B cell‐specific AID protein deaminates cytidines to uridines within transcribed switch regions, initiating a series of reactions that leads to the generation of the DSBs required for CSR (Muramatsu et al, 2007). Mutations in AID that block its nuclear export severely impair CSR, but do not affect cytidine deamination. To determine whether nucleocytoplasmic shuttling is required for CSR, M. Neuberger (Cambridge, UK) mutated the AID nuclear‐export signal and/or replaced it with one from a heterologous protein (Geisberger et al, 2009). Many of the mutant proteins underwent nuclear export, but were severely impaired for CSR, which showed that export is necessary but not sufficient for CSR. However, some mutants that allowed nuclear export and CSR showed increased protein stability, which indicated that the AID nuclear‐export signal might mediate interactions with other proteins—for example, exportin and/or Crm‐1—to generate the high density of lesions at switch regions required for CSR.
Chromosomal translocations between the immunoglobulin loci and oncogenes are characteristics of B‐cell lymphomas. To determine why some genes are targeted for translocations, M. Lieber (Los Angeles, CA, USA) assembled a database of breakpoints from the most common chromosomal rearrangements (bcl1, bcl2 and E2A) in human leukaemias and lymphomas. His analyses revealed that CpG nucleotides, which constitute only 1% of the human genome, account for 40–70% of the translocations in early‐stage, but not late‐stage, B‐cell lymphomas (Tsai et al, 2008). These translocations probably arise owing to AID activity on methylated cytidines (Morgan et al, 2004), followed by RAG1/RAG2 activity at the mispaired bases.
The c‐Myc oncogene is also a common translocation partner of the IgH in patients with Burkitt lymphomas and mice with plasmacytomas (Ramiro et al, 2007). To determine whether AID generates DSBs at c‐Myc, D. Robbiani (New York, NY, USA) used I‐SceI to introduce DSBs upstream of the IgH‐switch region and/or the c‐Myc locus. In the absence of I‐SceI, all c‐Myc/IgH translocations were AID dependent. By comparison, c‐Myc/IgH translocations were observed in activated AID‐deficient B cells only when I‐SceI‐induced DSBs were generated at both loci. Interestingly, Robbiani showed that the breaks at c‐Myc are limiting under physiological conditions (Robbiani et al, 2008). In this context, F. Alt (Boston, MA, USA) found that c‐Myc expression in the translocated IgH/c‐Myc allele requires a transcriptional enhancer at the 3′ end of the IgH locus.
The role of NHEJ in CSR remains ill defined. Both Alt and J‐P de Villartay (Paris, France) showed that conditional inactivation of Xrcc4 or DNA ligase IV in mice allows significant levels of CSR. K. Yu (East Lansing, MI, USA) arrived at a similar conclusion by studying a B‐cell line in which DNA ligase IV was inactivated. In each study, most switch junctions showed DNA‐sequence microhomology. Therefore, an alternative end‐joining pathway is also active in CSR. Indeed, Alt found that Xrcc4‐deficient splenic B cells accumulate AID‐dependent DSBs at the IgH locus on chromosome 12 and RAG‐dependent DSBs at the Ig light‐chain gene on chromosome 16, which could lead to chromosome 12/16 translocations by alternative end joining. Therefore, although it is not strictly required for CSR, classical NHEJ is important for maintaining genome stability during CSR.
Together, these talks reiterated that although targeted DNA alterations are essential for immune‐system development, mistargeted RAG and/or AID activity could lead to the chromosomal translocations that are characteristic of B‐cell lymphomas. Similarly, DNA‐repair proteins that process the DSBs during lymphocyte differentiation have to be finely choreographed, as aberrant repair might have catastrophic consequences.
Meiotic control of recombination
During meiosis, recombination is initiated at DSBs that are created by the conserved Spo11 protein (Keeney & Neale, 2006). Repair of these DSBs by HR leads to either a reciprocal exchange of DNA between homologues (a crossover) or a non‐reciprocal transfer of genetic information from one homologue to another (a non‐crossover). Crossovers are important as they provide a physical link between homologues that ensures correct chromosome segregation. The regulation of meiotic recombination ensures that each pair of homologues receives at least one crossover event.
Several talks focused on genome‐wide approaches to analyse the initiation and/or resolution of meiotic‐recombination events. Previous work in several organisms showed that meiotic DSBs are not distributed randomly throughout the genome, but tend to cluster in ‘hotspots’. What causes this non‐random distribution remains poorly understood. S. Keeney (New York, NY, USA) described a high‐resolution method to analyse global DSB distributions in S. cerevisiae using Spo11‐associated oligonucleotide sequences. These experiments refine previous genome‐wide analyses of meiotic DSBs, paving the way for a more detailed analysis of their regulation. Although significant overlap exists between the crossover and DSB maps, there are distinct differences, indicating that crossover placement is subject to additional levels of regulation. Phenomena that can potentially account for these differences include crossover interference or crossover homeostasis, as discussed by J. Fung (San Francisco, CA, USA), and interference between non‐crossover and crossover events as presented by E. Mancera (Heidelberg, Germany). Interference imposes a more regular spacing of crossovers, whereas crossover homeostasis limits the number of crossovers to a tight fluctuation about the average.
Two talks highlighting recent work in mice suggested that trans‐acting factors can influence crossover distribution. B. de Massy (Montpellier, France) and P. Petkov (Bar Harbor, ME, USA) reported the identification of overlapping regions of chromosome 17 that influence the activity of recombination hotspots elsewhere in the genome. These loci, Dsbc1 (Grey et al, 2009) and Rcr1 (Parvanov et al, 2009), neither act through a general activation or suppression of all hotspots, nor alter the overall number of crossovers; instead, they seem to affect the activity of specific hotpots differentially. In both cases, the loci were shown to affect the initiation of recombination before the crossover/non‐crossover decision is made.
Copy‐number variants in the human genome
Several presentations also focused on how CNVs have influenced the human genome. CNV is a broad term used to describe DNA rearrangements that arise from chromosomal duplications, deletions, inversions, translocations and mobile genetic‐element insertions. J. Lupski (Houston, TX, USA) introduced the concept of genomic disorders and discussed how non‐allelic recombination events between highly similar, but non‐identical, low‐copy number repeated DNA sequences can result in distinct genomic disorders (Lupski, 2009). He then proposed a fork‐stalling and template‐switching mechanism that might be important in generating non‐recurrent DNA rearrangements in human cells (Lee et al, 2007). Interestingly, the junction sequences at many genomic rearrangements were composed of a patchwork of sequences derived from various intrachromosomal locations—a phenomenon that was also touched on in Stratton's talk. It will be interesting to see whether some mutations observed in tumours occur by a similar mechanism.
M. Hurles (Cambridge, UK) discussed how his colleagues at the Sanger Institute (Cambridge, UK), in Boston (MA, USA) and in Toronto (Canada) have amassed a near complete map of common CNVs in the human genome. The map comprises approximately 11,000 CNVs that are more than 1 kb in length and some generalizations are becoming evident. First, it seems that CNVs are relatively common and might arise by distinct molecular mechanisms—for example, non‐allelic homologous recombination, fork stalling and template switching, and/or retrotransposition—that often leave signatures in the resulting alleles. Second, genome architecture might predispose certain regions of human DNA to these types of rearrangement.
The idea that sequence context might predispose certain chromosomal regions to rearrangements was addressed by B. Emanuel (Philadelphia, PA, USA), who reported that nearly perfect palindromic AT‐rich DNA‐repeat sequences on human chromosome 22q11 might form stem–loop and/or cruciform‐like structures, predisposing them to disease‐producing meiotic DNA translocations with other genomic regions such as 11q23 and 8q24.1. These translocation partners also contain AT‐rich palindromes, providing support for the role of palindromic sequences in genomic instability. S. Lewis (Boston, MA, USA) discussed in vitro findings in yeast and mammalian cells, and showed that extended palindromic sequences could be extruded from DNA to form cruciform structures. The processing of these cruciforms in the cell could lead to DSBs, resulting in both subtle and large‐scale genomic DNA rearrangements. Lewis also presented a model called palindrome revision, and suggested that subtle rearrangements at extruded palindromic sequences might act to ‘disable’ them before they can lead to more serious genome disruptions. It will be interesting to see whether small central deletions at other palindromic sequences in human DNA render them ‘immune’ to subsequent gross chromosomal rearrangements.
The 2009 Keystone Symposium on Genome Instability and DNA Repair provided an opportunity to present data, exchange ideas and spend time with like‐minded colleagues. The meeting highlighted the many pathways used by cells to deal with genotoxic stress effectively, and how they are used to repair deliberate lesions introduced during the processes of immune‐gene diversification and meiosis. Indeed, the knowledge obtained from in vitro experiments and in vivo studies in model organisms—only a few of which could be mentioned here—should be instrumental in determining how the complex genomic rearrangements observed in cancer and human genomic disorders arise. We predict that this will be a fertile area of research for years to come, and thank M. Jasin, S. Elledge and M. Neuberger for bringing us together to discuss cutting‐edge issues on this topic.
Conflict of Interest
The authors declare that they have no conflict of interest.
V.C. is supported by Cancer Research UK, the European Research Council, the Lister Institute for Preventive Medicine and the European Molecular Biology Organization Young Investigator Programme. J.C. is supported by the Damon‐Runyon Cancer Research Foundation, the Bressler Foundation and the National Institutes of Health (1R01AI072194‐01). J.C.F. is supported by an American Cancer Research Scholar Award (RSG CCG 110,688). J.V.M. is supported by the Howard Hughes Medical Institute and the National Institutes of Health (GM60518).
- protein complex containing BRCA1 and NBA1, and other proteins
- activation‐induced cytidine deaminase
- ataxia telangiectasia mutated
- B‐cell leukaemia lymphoma
- breast cancer 1
- avian myelocytomatosis virus oncogene cellular homologue
- copy‐number variant
- cytosine guanine
- chromosome region‐maintenance 1
- class‐switch recombination
- DNA nuclease 2
- double‐strand break
- double‐strand break control 1
- E2A immunoglobulin enhancer‐binding factors E12/E47; also known as transcription factor 3 (TCF3)
- essential meiotic endonuclease 1
- exonuclease I
- XPG‐like endonuclease
- homologous recombination
- immunoglobulin heavy‐chain locus
- a rare cutting restriction endonuclease encoded by the Saccharomyces cerevisiae mitochondrial intron, omega
- 70 kDa Ku antigen
- methyl methanesulphonate sensitive 4
- meiotic recombination 11
- Mre11–Rad50–Nbs1 complex
- methyl methane sulphonate and ultraviolet sensitive 81
- new component of the BRCA1 A complex
- non‐homologous end joining
- recombination‐activating gene 1/2
- recombination regulator 1
- slow growth suppressor 1
- sporulation 11
- variable‐diversity‐joining recombination
- X‐ray cross complementation 4
- yeast endonuclease 1
- Copyright © 2009 European Molecular Biology Organization