|Home | About | Journals | Submit | Contact Us | Français|
Genome integrity is frequently challenged by DNA lesions from both endogenous and exogenous sources. A single DNA double-strand break (DSB) is lethal if unrepaired and may lead to loss of heterozygosity, mutations, deletions, genomic rearrangements and chromosome loss if repaired improperly. Such genetic alterations are the main causes of cancer and other genetic diseases. Consequently, DNA double-strand break repair (DSBR) is an important process in all living organisms. DSBR is also the driving mechanism in most strategies of gene targeting, which has applications in both genetic and clinical research. Here we review the cell biological response to DSBs in mitotically growing cells with an emphasis on homologous recombination pathways in yeast Saccharomyces cerevisiae and in mammalian cells.
Advances in automated live cell fluorescence microscopy, digital image analysis, and engineering of spectral variants of green and red fluorescent proteins over the last decade has greatly contributed to our understanding of the cellular response to DNA damage . The ability to induce a single DSB in a site-specific manner by mega-nucleases or a subnuclear cluster of DSBs by laser micro-irradiation and other approaches has further facilitated the study of DSB repair. Many proteins are recruited to a DSB in multiple copies, resulting in a high local concentration of the protein, which is visualized cytologically as a focus using either GFP-tagged (green fluorescent protein) fusion proteins or immunohistochemistry (Fig. 1). The recruitment of checkpoint and repair proteins to foci is being used to study the choreography of DSB repair inside cells.
Two major pathways exist for repairing DSBs: non-homologous end-joining (NHEJ) and homologous recombination (HR) (Fig. 2). In general, NHEJ is the preferred pathway in the G1 phase of the cell cycle, while HR is favored in S and G2 phases [5,6]. Moreover, NHEJ makes a greater contribution to DSBR in higher eukaryotes with large intergenic regions and many repetitive sequences, whereas HR is preferred in yeast and other eukaryotes with small compact genomes of short intergenic regions and low redundancy [7,8]. Biochemically, NHEJ consists of a religation of the two ends of a DSB. Since ligation is often preceded by nucleolytic processing of ends, NHEJ is prone to causing deletions . By contrast, DSBs that are repaired by HR do so without loss of genetic information, because the break is resealed by copying genetic information from an intact donor sequence.
The monitoring of DNA damage-induced foci is a useful tool for studying HR in living cells. Foci are dynamic giga-Dalton structures typically containing hundreds to thousands of copies of a given repair protein. The high local concentration of proteins at the DNA lesion is likely to facilitate the efficiency of individual biochemical steps during recombination. For example, budding yeast Rad52 is present at approximately 50-fold higher concentration within foci than elsewhere in the nucleus. At the same time, the low abundance of repair proteins at undamaged DNA likely constitutes a barrier to untimely recombination or potentially deleterious interactions with intact DNA. For each specific DNA repair protein in question, its accumulation at a DNA lesion results from a combination of novel binding sites being generated and a change in the residence time of the bound protein, as elegantly demonstrated by fluorescence recovery after photobleaching (FRAP) and fluorescence loss in photobleaching (FLIP) analyses in mammalian cells . Finally, it is important to keep in mind that although HR foci are likely to reflect an attempt to perform recombination, they are not evidence of successful completion of the process. Moreover, it is likely that some HR events do not form cytologically detectable foci, because the reaction is fast and therefore completed before the accumulation of a detectable concentration of repair proteins at the DNA lesion.
In budding yeast, DNA double-strand breaks are initially sensed independently by the Ku (Yku70-Yku80) and MRX (Mre11-Rad50-Xrs2) complexes [11-13], both of which bind directly to DNA ends [14-16]. The relationship between these complexes is a question under active investigation, but it is reported that the Ku complex delays the onset of 5′ end resection likely creating a window of opportunity for NHEJ , which itself is stimulated 3-fold by Ku in vitro . In contrast, MRX promotes the onset of resection using its intrinsic 3′-exo- and endonuclease activity [18-24]. Likely, the competition between Ku and MRX for binding to DSBs is important for determining the choice between NHEJ and HR during DSB repair.
The two ends of a DSB are held together by a mechanism that is partially dependent on MRX and Sae2 [16,25-27]. For this reason, the two ends of a DSB form a single Mre11 focus. Further, the MRX complex interacts with the Tel1 kinase and is required for its recruitment to foci . The Tel1 kinase phosphorylates histone H2A, which is a chromatin mark specific for damaged DNA in most eukaryotes [29-31] (Fig. 3). The recognition of DSBs by MRX occurs independently of cell cycle phase and other factors, consistent with the notion that the complex binds directly to free DNA ends [13,16]. Importantly, the modification of chromatin by H2A phosphorylation facilitates binding of Rad9 to histone H3 methylated at lysine 79 (H3-K79Me) and subsequent recruitment and activation of Rad53 [13,32].
Unless a DSB is repaired by NHEJ, it is eventually processed by 5′ to 3′ nucleolytic resection, which commits the cell to HR. In yeast, resection is a two-step process catalyzed by numerous partially redundant nucleases including Mre11, Sae2, Dna2 and Exo1 [18-25,33]. First, resection is initiated by Sae2 and the MRX complex. Sae2 and MRX are particularly important for removal of hairpins, bulky adducts and other irregular end structures [23,34-36]. Second, a more processive resection is catalyzed by Exo1 and/or Dna2 in collaboration with the Sgs1 helicase [18,19]. The efficiency of resection depends both on cell cycle phase and the type of DSB. While endonuclease induced DSBs are resected only in S and G2 phase, DSB ends induced by ionizing radiation are resected throughout the cell cycle [37,38]. Resection is accompanied by the dissociation of MRX, Sae2 and Tel1 from the DSB and binding of replication protein A (RPA) to the 3′ single-stranded overhangs (Fig. 4A) [13,37]. RPA is necessary for recruiting a number of checkpoint and HR proteins including the Mec1-Ddc2, Rad24-RFC and 9-1-1 (Ddc1-Mec3-Rad17) complexes. Notably, Tel1 and Mec1 have many of the same phosphorylation targets including histone H2A. As a consequence, Tel1-dependent checkpoint signaling is likely replaced by Mec1-dependent signaling upon resection of DSB ends. Interestingly, recent experiments have shown that tethering of Ddc1 and Ddc2 to DNA is sufficient for checkpoint activation . Notably, efficient Rad53 phosphorylation is observed only if 80 or more copies of Ddc1 and Ddc2 are tethered to DNA, which is analogous to the accumulation of these proteins to foci during the DNA damage response . Similar activation of the DNA damage response is observed in mammalian cells upon tethering of NBS1, MRE11, ATM or MDC1 to chromatin, but not after tethering the downstream effector kinases Chk1 and Chk1, which underscores the functional significance of this observation . Consistent with a functional crosstalk between the Ddc2-Mec1 and 9-1-1 complexes during checkpoint signaling, there is a reported requirement for the 9-1-1 complex to stabilize Ddc2 foci in irradiated G1 cells . Further, in S and G2 phase the 9-1-1 complex and the Cdc28 kinase both contribute to the stabilization of DNA damage-induced Ddc2 foci , demonstrating how checkpoint signaling is coordinated with cell cycle phase.
In S and G2 phase, RPA facilitates the recruitment of Rad52 to DSBs likely via a direct physical interaction [13,41]. Rad52 interacts with the Rad51 recombinase and Rad59 to recruit these proteins to foci [13,42,43]. In addition, Rad59 also requires Rad52 for its nuclear localization . Rdh54 is a DNA-dependent ATPase belonging to the SWI/SNF family of chromatin remodeling complexes . Interestingly, Rdh54 is recruited both to DSBs and to the kinetochore, although the functional significance of this dual localization is unknown. The recruitment of Rdh54 to DSBs is Rad52- and Rad51-dependent, while its localization to the kinetochore is independent of the recombination machinery. Rdh54 is a homologue of Rad54, which itself is recruited to DSBs but not to the kinetochore in wild-type cells. Interestingly, Rad54 does localize to the kinetochore in a rdh54Δ mutant , likely explaining some of the functional redundancy between these two proteins . While Rdh54 appears to be recruited to DSBs immediately downstream of Rad51, Rad54 also requires the Rad51 paralogues, Rad55-Rad57, suggesting that Rad54 focus formation requires Rad51 nucleoprotein filament formation . These requirements for focus formation indicate that Rdh54 acts upstream of Rad54 during homologous recombination. Together with Rad52, Rad55-Rad57 mediate the formation of Rad51 filaments and themselves require Rad51 for focus formation [13,46]. Importantly, Rad55 foci are transient suggesting that Rad55-Rad57 is not incorporated into the Rad51 filament (unpublished data) (Fig. 4A). At present, it remains to be determined which proteins are localized to DSBs during the late stages of homologous recombination, when recombination intermediates are resolved, and how chromatin is restored to its initial state.
Many of the proteins involved in the DNA damage response are evolutionarily conserved between yeast and mammals (Fig. 3). Accordingly, the cellular response to DSBs in mammalian cells shows similarities to yeast, but also exhibits some significant differences. Similar to yeast, the MRE11-RAD50-NBS1 (MRN) and KU70-KU80 (Ku) complexes are recruited to DSB ends independently [47-50]. Ku recruits the DNA-PKcs kinase to form the DNA-PK holoenzyme, which facilitates the juxtaposition of ends prior to recruitment of additional end-joining factors [51,52]. A notable difference between yeast and higher eukaryotes in DSB recognition is the absence of a DNA-PKcs homologue in yeast and the requirement for the yeast MRX in both NHEJ and HR, while human MRN is required only for HR [53-55]. Perhaps DNA-PK is responsible for the higher efficiency of NHEJ in mammalian cells compared to yeast.
During HR, MRN recruits the ATM kinase, which phosphorylates histone H2AX at serine 139 to generate a chromatin mark (γ-H2AX) similar to that made by Tel1 in budding yeast [56-58]. Mice lacking H2AX are radiation sensitive, immune deficient, and males are infertile . In mammalian cells, γ-H2AX is bound by the master regulator of recombinational DNA repair, MDC1 [60-63], which itself interacts with ATM via its FHA domain and with NBS1 [61,64-66]. The interactions between γ-H2AX, MDC1, ATM and NBS1 have been proposed to generate a positive feedback loop to amplify the γ-H2AX signal and promote additional recruitment of MDC1, ATM and NBS1  (Fig. 4B). MDC1 itself is phosphorylated in an ATM-dependent manner [67,68], and this modification stimulates binding by the E3 ubiquitin ligase, RNF8, which in turn specifies the ubiquitylation of histone H2A and H2AX [69-71]. At the cell biological level, the recruitment of MRN and ATM, phosphorylation of H2AX, and recruitment of MDC1 and RNF8 occur very quickly and with similar kinetics, suggesting that each of these events are part of an early, coordinated initiation of the response to DSBs .
Ubiquitylated histones H2A and H2AX are bound by RAP80 in a second wave of recruitment of proteins to a DSB [69,72,73]. During this wave, RAP80 interacts with Abraxas (CCDC98) , which associates with the deubiquitylating enzyme BRCC36 and with the BRCT domain of BRCA1 [72,75-77]. The complex also contains BARD1 (BRCA1-associated RING domain protein 1), which interacts with BRCA1 through their respective RING domain-containing N-termini to form an active E3 ubiquitin ligase [78,79]. In addition to the Abraxas interaction, BRCA1 can also interact with the BACH1 helicase and the CtIP nuclease [80,81]. Interactions of Abraxas, BACH1 and CtIP with the BRCT domain of BRCA1 are mutually exclusive, suggesting that the three proteins compete for the same binding site to form distinct BRCA1-containing complexes, of which only the Abraxas-BRCC36-BRCA1-BARD1 complex is recruited to sites of DNA damage by RAP80 . In support of this notion, BACH1 knockdown does not disrupt BRCA1 focus formation . Moreover, sedimentation velocity analysis of IR-induced BRCA1-BARD1 complexes identified distinct complexes containing either MRN and CtIP or BACH1 and TopBP1 . Further, IR-induced Abraxas and RAP80 foci are BRCA1 independent , indicating that BRCA1 is acting at DSBs downstream of RAP80. Similarly, BACH1 and CtIP foci require BRCA1 [80,82,83], suggesting that BACH1 and CtIP are downstream of BRCA1. At the moment, it is unclear how the multiple interactions of the BRCT domain of BRCA1 with these proteins are coordinated in vivo (reviewed in).
Ubiquitylation of histones H2A and H2AX also facilitates binding of 53BP1 to H4-K20Me [69,85,86]. The alternative binding of 53BP1 to H3-K79Me is apparently three orders of magnitude lower although the literature is not in full agreement on this point [85,87]. The similarities of these observations to the recruitment of the 53BP1 homologues to foci in S. cerevisiae and S. pombe (Rad9 and Crb2, respectively) makes it tempting to speculate that ubiquitylation plays a hitherto uncharacterized role in recruitment of these proteins to sites of DNA damage in yeast. However, deletion of the two putative RNF8 homologues in S. cerevisiae, DMA1 and DMA2, does not significantly affect Rad9 focus formation in response to DNA damage (unpublished results). The two BRCT domains of 53BP1 interact with the DNA-binding surface of the p53 tumor suppressor , which possibly allows 53BP1 to recruit p53 to activated Chk2, where p53 is phosphorylated at serine 20 in response to DNA damage . Phosphorylation of p53 leads to its stabilization and activation as a transcription factor [90,91]. Both 53BP1 and BRCA1 foci are dependent on MDC1 and RNF8, but appear to be mutually independent [69,92], although some studies report an interdependency of 53BP1 and BRCA1 . However, the independent recruitment of 53BP1 and BRCA1 is consistent with the notion that these proteins represent distinct protein complexes being recruited to chromatin by H4-K20Me and RAP80, respectively.
A prerequisite for recombinational repair of a DSB is the 5′ to 3′ resection of the two ends to yield 3′ single-stranded overhangs. In mammalian cells, a number of nucleases contribute to the resection of ends including MRE11, EXO1 and CtIP [94-96]. MRE11 and CtIP form a complex, but the two proteins form foci independently of each other and independently of BRCA1 . In budding yeast, Mre11 and Sae2 are also recruited to foci independently . In contrast, the fission yeast CtIP is recruited to DSBs in an Mre11-dependent manner . End resection is accompanied by a third wave of recruitment of proteins to the DNA lesion and a simultaneous loss of binding of NHEJ factors . However, in contrast to yeast, MRN is retained at DSBs after resection by an interaction of NBS1 to γ-H2AX and/or MDC1 [64,98]. As in all organisms studied, RPA binds directly to single-stranded DNA formed by resection of DSBs  and the resulting RPA foci are remarkably alike across species (Fig. 1). RPA recruits the ATR-ATRIP checkpoint kinase [69,100,101]. In addition, RPA also recruits the alternative DNA clamp loader RAD17-RFC, which loads the 9-1-1 complex (RAD9-HUS1-RAD1) onto DNA [102-105]. Both ATR-ATRIP and ATM contribute to the phosphorylation of H2AX and CHK2 to activate the DNA damage checkpoint . Similar to the yeast homologs, phosphorylated CHK2 exhibits very weak retention at sites of DNA damage and its dispersal throughout the nucleoplasm is crucial for checkpoint signaling .
Initiation of recombination requires the recruitment of the Rad51 recombinase. In contrast to yeast wherein Rad52 recruits Rad51 to foci [13,108], the RAD51 recombinase in mammalian cells is recruited to IR-induced DSBs by BRCA2 . This difference could explain the reduced requirement for RAD52 during HR in mammalian cells compared to yeast. Nevertheless, RAD52 is recruited to RAD51 foci in IR-treated cells . BRCA2 (FancD1) is part of the Fanconi anemia (FA) pathway (reviewed in ) and it remains to be fully established how BRCA2 is recruited to foci. However, at least three observations seem to be connected to this question. First, BRCA2 binds directly to single-stranded DNA as suggested by the reported in vitro complex of BRCA2 and DSS1 with single-stranded DNA . Second, RPA directly interacts with the acidic N-terminus of BRCA2, which may contribute to recruiting BRCA2 to RPA-coated single-stranded DNA . Third, BRCA2-PALB2 foci colocalize with FancD2 and BRCA1 foci [114-116] and their interdependencies suggest that BRCA2-PALB2 recruitment is initiated by the binding of BRCA1 to chromatin in the vicinity of a DSB. BRCA1 facilitates the recruitment of FancD2 to foci [117,118], which also requires FancD2 monoubiquitylation at lysine 561 by the FA core complex . Yeast two-hybrid analysis indicates that FancD2 interacts directly with BRCA2 , raising the possibility that FancD2 directly contributes to recruiting BRCA2 to foci. Fittingly, FancD2 foci are independent of BRCA2, supporting the notion that FancD2 is upstream of BRCA2 . Taken together, the data suggest that BRCA2 is recruited to foci by multiple protein and DNA interactions likely depending on the context of the DNA lesion.
In addition to BRCA2, IR-induced RAD51 focus formation also requires RAD54, and the five RAD51 paralogues RAD51B, RAD51C, RAD51D, XRCC2 and XRCC3 [109,120-126]. The RAD51 paralogues form two distinct complexes: a RAD51B-RAD51C-RAD51D-XRCC2 complex and a RAD51C-XRCC3 complex  with RAD51 interacting directly with XRCC3 in the smaller complex [128,129]. Unlike in mammalian cells, the Rad51 paralogues in S. cerevisiae, Rad55 and Rad57, are not required for Rad51 focus formation in yeast . The formation of XRCC3 foci is early and independent of RAD51 foci . During S phase, RAD51 can form BRCA2-independent foci , which are likely to represent replication-coupled HR, where RAD51 is recruited by a direct interaction with RPA [131,132]. Human cells have two homologues of yeast Rad54, RAD54 and RAD54B, both of which are recruited to DNA damage-induced RAD51 foci . These foci also contain the WRN helicase, which interacts with both RAD51 and RAD54B . IR-induced WRN focus formation is defective in Nijmegen breakage syndrome (NBS) cells, whereas Werner syndrome (WS) cells retain the ability to form NBS1 foci consistent with the notion that WRN acts downstream of NBS1 in HR . In addition to its RAD51 and RAD54B interactions, WRN physically and functionally interacts with RAD52 in vitro and in vivo at foci after mitomycin C treatment . Interestingly, simultaneous deletion of RAD52 and XRCC3 in chicken DT40 cells is lethal, suggesting that RAD52 and the XRCC3-RAD51C complex have redundant functions . Consistent with this idea, RAD52 foci in mammalian cells form independently of BRCA2 and the RAD51 paralogues XRCC2, XRCC3 and RAD51C . Most likely RAD52 is recruited to foci by a direct physical interaction with RPA, although this remains to be formally demonstrated .
DNA double-strand break repair by homologous recombination is organized into foci, which can be viewed as repair factories, where a high local concentration of the proteins facilitates efficient repair. However, the function of repair foci likely extends beyond this simplistic biochemical point of view. In fact, most DSBs consist of two ends, which likely require tethering to ensure correct rejoining and perhaps prevent telomere addition . In budding yeast, the tethering of broken ends requires the MRX complex, Sae2 and the Tel1 kinase [26,27,139], but it is likely that other factors are also needed. There is in vitro evidence to indicate that MRN and ATM play a similar role in metazoans [47,140].
Yeast cells that experience multiple DSBs have the ability to recruit these lesions to the same repair focus . In contrast, DSBs in mammalian cells induced by γ-irradiation or the I-SceI endonuclease are largely immobile [142,143] with the interesting exception of DSBs induced by α-irradiation , and deprotected telomeres, which are mobilized in a 53BP1-dependent manner . Possibly, the aggregation of multiple DSBs results from the attempt to tether DNA ends in general. In mammalian KU70-/- cells, translocation between DSBs on different chromosomes are increased . Thus, the Ku complex is important for proper tethering of ends, allowing the cell “to know” which ends belong together in order to avoid translocations.
Although HR is important for maintaining genome integrity, the process must be tightly regulated to avoid untimely and toxic recombination. An example of such regulation at the cell biological level is the exclusion of most HR proteins from the nucleolar compartment in budding yeast despite that fact that these same proteins are required to maintain rDNA homeostasis . Interestingly, DSBs in the rDNA must relocalize to a position outside the nucleolus before association with a Rad52 focus. Sumoylation of Rad52, MRX and the Smc5-Smc6 complex jointly contribute to this process and a failure to suppress the formation of Rad52 foci inside the nucleolus results in rDNA hyperrecombination and formation of extra-chromosomal rDNA circles .
Recently, it was demonstrated that persistent DSBs relocalize to the nuclear pore complex (NPC) in an Slx5-Slx8- and Mec1/Tel1-dependent manner to stimulate gene conversion . Interestingly, “old” NPCs and associated DNA circles are retained in the mother cell during cell division by a lateral diffusion barrier at the septum . Thus, these findings leads to the intriguing possibility that irreparable DSBs are retained in the mother cell during adaptation and that this mechanism may contribute to the mother-daughter senescence asymmetry (reviewed in ).
DNA damage-induced foci of repair proteins are frequently used as markers of DNA damage and ongoing DNA repair. However, at present we know very little about the molecular architecture, regulation and even significance of these foci. The evidence suggests that the nucleation of foci is guided by a few proteins binding directly to the DNA lesion or to chromatin in its vicinity e.g. MRX(N) binding to DNA ends, RPA to single-stranded DNA, and MDC1 to γ-H2AX. These DNA damage sensors initiate a recruitment cascade to attract additional factors via a network of protein-protein interactions which, in some cases, are regulated by post-translational modifications such as phosphorylation, ubiquitylation and sumoylation that act as molecular switches (Fig. 5).
A recent genome-wide screen in S. cerevisiae identified several novel factors that impinge on Rad52 focus formation , indicating that repair factories respond to a variety of genome stress conditions. Among the gene disruptions that cause an increase in the percentage of cells exhibiting spontaneous Rad52 foci were many involved in DNA metabolism and cell cycle regulation as well as 22 uncharacterized open reading frames, IRC2–11, 13–16, 18–25 (Increased Recombination Centers). Many of these ORFs are conserved throughout evolution. Similar screens in other organisms are expected to reveal additional DNA repair factors.
We thank members of the Lisby and Rothstein laboratories for helpful discussions concerning this work. We especially acknowledge the insightful comments and suggestions by Rebecca Burgess, Peter Thorpe and Simon Bekker-Jensen. This work was supported by The Danish Agency for Science, Technology and Innovation (ML), the Villum Kann Rasmussen Foundation (ML), and the NIH grants GM50237 and GM67055 (RR).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.