Sharpening the ends for repair: mechanisms and regulation of DNA resection

Abstract

DNA end resection is a key process in the cellular response to DNA double-strand break damage that is essential for genome maintenance and cell survival. Resection involves selective processing of 5′ ends of broken DNA to generate ssDNA overhangs, which in turn control both DNA repair and checkpoint signaling. DNA resection is the first step in homologous recombination-mediated repair and a prerequisite for the activation of the ataxia telangiectasia mutated and Rad3-related (ATR)-dependent checkpoint that coordinates repair with cell cycle progression and other cellular processes. Resection occurs in a cell cycle-dependent manner and is regulated by multiple factors to ensure an optimal amount of ssDNA required for proper repair and genome stability. Here, we review the latest findings on the molecular mechanisms and regulation of the DNA end resection process and their implications for cancer formation and treatment.

Introduction

DNA double-strand breaks (DSBs) are arguably the most toxic form of DNA damage, which, if unrepaired or improperly repaired, could cause genomic instability and a wide range of human diseases such as cancer, premature aging, immunodeficiency, neurodegeneration, and developmental disorders [1–4]. Eukaryotic cells are equipped with a conserved mechanism called DNA damage response (DDR) to detect, signal, and repair the damage by activating multiple repair and checkpoint pathways [5–7]. DSBs are repaired mainly by non-homologous end joining (NHEJ) or homologous recombination (HR). NHEJ repairs the break through direct re-ligation of the broken DNA ends with no or limited end processing and thus is error-prone. By comparison, HR repairs the break in an error-free manner, and is initiated by nucleolytic processing of the 5′ ends of a DSB through a process called DNA end resection [8–16]. Resection occurs in 5′→3′ direction to generate 3′ ssDNA overhangs, which are initially bound by ssDNA-binding protein replication protein A (RPA) and then replaced by Rad51 during HR. The Rad51-ssDNA filament mediates homology search and strand invasion, followed by DNA synthesis, Holliday junction resolution, and DNA ligation to restore the integrity of the DNA structure [9,10,15,16]. The RPA-bound ssDNA structure also serves as the signal to activate the ATR checkpoint pathway that coordinates DNA repair with other cellular processes [17–19]. The generation of ssDNA by resection also indirectly inhibits NHEJ and attenuates the activation of the ataxia telangiectasia mutated (ATM) checkpoint pathway [8,9,11,20,21]. Thus, resection is considered to be the major event in the DDR that dictates the pathway choice of both DNA repair and checkpoint signaling (Fig. 1). While DSB repair by NHEJ can occur at any time during the cell cycle, HR occurs primarily in S and G2 phases when sister chromatids are available [9–11,22–25]. This cell cycle control of HR is in part mediated by the regulation of DNA end resection by cyclin dependent kinase (CDK) activity [26–29]. Resection is apparently also regulated by the checkpoint response to prevent deleterious consequences resulted from excessive resection [30–35]. Besides its role in HR, resection also plays a role in the maintenance of 3′ overhangs at telomeres and repair of uncapped telomeres at the end of chromosomes [36–39]. Likewise, resection also occurs at ssDNA-dsDNA junctions of stalled replication forks and at dsDNA ends of reversed forks, and is important for fork repair and restart [40–45]. However, the detailed mechanisms of end resection in these contexts are much less understood. Genetic mutations in resection factors are associated with multiple genetic disorders and predisposition to cancer and premature aging [1,2,4]. On the other hand, DNA end resection could also be a suitable target for cancer therapy because rapidly dividing cancer cells rely heavily on HR and the ATR checkpoint for growth and survival. In this review, we discuss our current understanding of the mechanisms and regulation of the DNA end resection process and their potential implications for cancer formation and treatment, focusing mainly on vertebrate systems.

Figure 1.

DNA end resection dictates the pathway choice for both DNA repair and checkpoint response DSBs can be repaired by NHEJ or by HR that requires 3′ ssDNA generated by end resection. Checkpoint kinase ATM is activated on the dsDNA flanking the break, whereas ATR is activated on the ssDNA structure generated by resection. Thus, DNA end resection promotes HR and ATR activation and attenuates NHEJ and ATM activation.

Key Steps and Core Factors of DNA End Resection

Initiation of resection by MRN and CtIP

Studies in yeast, human cells, and in vitro reconstituted systems with purified proteins suggest that DSB end resection is initiated by a concerted action of MRN (Mre11-Rad50-NBS1) [MRX (Mre11-Rad50-Xrs2) in budding yeast] together with CtIP (Sae2 in budding yeast and Ctp1 in fission yeast) [46–51] (Fig. 2). MRN complex, which is among the first set of proteins to localize to sites of DNA damage, has a high affinity for DSB ends and plays a central role in sensing breaks in chromatin [52–54]. MRN promotes the damage recruitment of the ATM checkpoint kinase and its subsequent activation [55]. They also promote the recruitment of CtIP to sites of damage [51]. The NBS1 subunit of the MRN complex plays a key role in coupling these events through its direct interactions with Mre11, CtIP, and ATM [51,55–59]. Rad50 is an ATPase that maintains the conformation of MRN complex and promotes DNA binding of the complex, as well as DNA resection and ATM activation by the complex [60–63]. The Mre11 subunit possesses the catalytic function of MRN complex in resection and has both 5′ flap endonuclease activity and 3′→5′ exonuclease activity. Its endonuclease function is believed to initiate resection by internal cleavage of the 5′ strand to generate oligonucleotides that will be released, while the exonuclease activity processes the resulting 3′ ends on the DNA [64–71]. While MRN is necessary, the complex by itself is not sufficient to initiate resection. CtIP is also required for the initiation of DNA end resection by MRN complex [50,51,72–74]. In vitro studies with purified yeast MRX and Sae2 proteins suggest that MRN-CtIP is a minimal system for resection initiation [46]. CtIP interacts directly with NBS1 and promotes the endonuclease activity of Mre11 at the DSB ends [50,51]. Both CtIP and Sae2 have also been shown to contain an endonuclease activity [75,76]. While there is no direct evidence for the nuclease activity of CtIP or Sae2 in the resection initiation at ‘clean’ DSBs, they have been suggested to function to remove secondary DNA structures on the 5′ strand DNA at DSB ends [75–77]. Resection initiation by MRN-CtIP is especially important when the ends are bound by chemical or protein adducts that prevent exonucleases from binding and processing them [69,70,78,79]. A prominent example of such breaks is the DSBs generated during meiotic recombination, which are covalently linked to the Spo11 protein. Resection of these Spo11-blocked ends in yeast is initiated by the endonucleolytic activity of MRX-Sae2 to initiate the resection before further processing [69,70]. For DSBs that are free of chemical or protein adducts, MRX-Sae2 function is dispensable for end resection [47,71,80]. In addition to end cleavage, studies in yeast suggest that MRX-Sae2 or MRN-Ctp1 plays a role in removing the NHEJ factor Ku from the DNA ends to promote the binding of nucleases Exo1 and Dna2 that mediate resection extension [81–83]. Moreover, MRN-CtIP in human cells also provide structural and catalytic support to recruit Exo1 and Dna2 to the damage site to extend the resection [82–87].

Figure 2.

Key steps and core factors of DNA end resection Resection is initiated on the 5′ strand of the DNA by the endocleavage activity of MRN-CtIP and extended by Exo1 and Dna2 in two parallel pathways. The underlying mechanism for the 5′ strand selectivity of MRN-CtIP in resection initiation is still not completely understood. The ssDNA generated from resection is bound and protected by RPA which promotes ATR activation and HR when replaced by Rad51. Precisely how resection is terminated remains unclear, but the ATR checkpoint pathway may help terminate resection via a feedback loop mechanism.

Extension of resection by Exo1 and Dna2

Limited resection by MRN-CtIP alone could lead to DNA repair by a less common and highly error-prone mechanism called microhomology-mediated end joining (MMEJ), which involves the alignment of short ssDNA overhangs before ligation [9,88–90]. While limited resection by MRX-Sae2 in yeast has been shown to be sufficient for HR repair, extended resection appears to be required to avoid MMEJ and promote HR [9,71,90–92]. Resection extension is carried out by Exo1 and Dna2 in two parallel pathways, which can produce ssDNA of several kilobases long [47,71,93,94] (Fig. 2). Exo1 is a member of the RAD2 family of nucleases that possesses 5′→3′ dsDNA exonuclease and 5′-flap endonuclease activities, and plays a role in a plethora of biological processes including DNA replication, recombination, repair, checkpoint activation, and telomere maintenance [95–103]. The resection function of Exo1 is positively regulated by MRN, Bloom syndrome RecQ-like helicase (BLM), RPA, proliferating cell nuclear antigen (PCNA), and 9-1-1 [87,103–106]. MRN, PCNA, and likely 9-1-1 complex act to promote the processivity of Exo1 [87,104–106]. It has been reported that CtIP promotes the loading of Exo1 to the damage site but negatively regulates Exo1 nuclease activity [107]. However, CtIP has also been shown to be required for extensive resection and checkpoint maintenance, although the detailed mechanism remains to be determined [108].

Another major resection extension factor is the helicase/endonuclease Dna2, which is well known for its role in Okazaki fragment maturation and G-quadruplex DNA processing during DNA replication [109–113]. During DNA end resection, Dna2 works together with Sgs1-Top3-Rmi1 complex in yeast and Sgs1 ortholog BLM in cultured human cells [71,85–87,114]. Studies in Xenopus egg extracts as well as human cells show that another RecQ family of helicase Werner syndrome RecQ-like helicase (WRN) also promotes resection by unwinding the DNA ends and making it accessible for Dna2 [115–119]. Although Dna2 functions as both a helicase and a nuclease, only the nuclease activity is essential for the extension of DNA end resection [87,120,121]. The long stretch of ssDNA generated by Exo1 and Dna2 serves as the substrate for HR, and in the meantime prevents repair by NHEJ or MMEJ [9,92]. The ssDNA-binding protein RPA promotes resection extension by enhancing the nuclease activity of Dna2 on the 5′ strand and by suppressing the inhibitory effects of the 3′ ssDNA resection product on Exo1 [85–87,92,117,118,122–124].

Termination of resection

Although ssDNA generated by resection is essential for ATR checkpoint and HR, uncontrolled excessive resection could be deleterious to genome integrity, as ssDNA is more prone to degradation that causes loss of genetic information. Excessive ssDNA generated by resection may also exhaust the RPA pool in cells, leading to unprotected ssDNA and genomic instability [123,125]. Therefore, it is expected that when the length of ssDNA reaches a certain threshold, resection activities would stop processing the DNA ends. However, the control of the timing and the mechanism of resection termination are still unclear. Studies in yeast and human cells both suggest that the nuclease activity of major resection factor Exo1 is inhibited in a checkpoint-dependent manner. In yeast, Exo1 nuclease activity is inhibited by both Mec1 and Rad53 at uncapped telomeres [30]. Phosphorylation of Exo1 by Rad53 in yeast appears to inhibit its activity in processing DSB ends, unprotected telomeres and stalled replication forks. In human cells, direct phosphorylation of Exo1 by ATR leads to Exo1 degradation during replication stress [31–33,126]. It is possible that Exo1 is also negatively regulated by the ATR checkpoint response during DNA end resection. Another study shows that ATM phosphorylates Exo1 and limits its activity after RPA is bound to ssDNA [34]. Together, these observations suggest that checkpoint-mediated phosphorylation of Exo1 inhibits its activity to terminate the resection. Interestingly, Exo1 interacts with phospho-peptide binding proteins 14-3-3s, and this interaction inhibits its damage recruitment and subsequent DNA resection [127–129]. Thus, it is plausible that phosphorylation of Exo1 by ATM, ATR, or their downstream kinases promotes the interaction of Exo1 with 14-3-3s, preventing its association with DNA damage, thereby promoting resection termination (Fig. 2). Interestingly, Durocher and colleagues have recently proposed another negative feedback mechanism for resection termination in which the recruitment of DNA helicase HELB by RPA to ssDNA inhibits the nuclease activities of Exo1 and BLM-Dna2, although the detailed biochemical mechanism of this inhibition remains to be defined [130]. Another possible mechanism for resection termination is the second end capture during HR, which may prevent further resection by annealing the complimentary strands and formation of double Holliday junction [131–134]. Recent studies have also shown that Dna2 is inhibited by fanconi anemia complementation group 2 (FANCD2) in human cells and Pxd1 in fission yeast [135,136], which could be the mechanisms to terminate Dna2-mediated DNA end resection.

RPA selects 5′ strand for resection and protects 3′ strand resection product

DSB resection occurs in the 5′→3′ direction, but what determines this directionality for the cleavage of the 5′ ends during resection initiation is still a mystery. An in vitro study by Petr Cejka and colleagues shows that MRX together with Sae2 selectively cleave the 5′ strand of a linear dsDNA substrate to initiate the resection, although the detailed mechanism for this strand selectivity in this ‘minimal’ resection initiation system remains to be determined [46]. Nevertheless, the mechanism for the strand selectivity during resection extension is better understood. Exo1 acts as a 5′→3′ exonuclease and thus has intrinsic polarity [95–98]. RPA plays a key role in selecting the 5′ strand for processing by Dna2, which functions as a flap endonuclease in resection [85–87,120]. Initial unwinding of the broken DNA ends by helicase BLM or WRN generates both 5′ and 3′ ssDNA strands. In vitro studies using purified proteins show that RPA binds to both strands but allows resection to occur only on the 5′ strand of the DNA [85–87,120]. A recent structural study of Dna2-ssDNA-RPA complex and in vitro nuclease assays using mouse Dna2 shows that Dna2 physically interacts with RPA bound to both strands but can only displace RPA from the 5′ strand and hence the resection occurs only on the 5′ strand [120]. Studies in Xenopus egg extracts also show that RPA interacts with both WRN and Dna2 to promote 3′→5′ helicase activity of WRN and 5′→3′ nuclease activity of Dna2 [117,118]. In addition to its role in directing 5′→3′ resection, RPA binds promptly to the newly generated 3′ ssDNA and protects the resection product [85–87,92,118,120,123,124]. Functional disruption of RPA in yeast not only abrogates resection extension, but also causes formation of hairpin structures on the short 3′ ssDNA generated by MRX-Sae2, which can be further processed, resulting in genomic instability [123]. Binding RPA to 3′ ssDNA overhangs also suppresses DNA repair by MMEJ [92].

Regulation of DNA End Resection

Either insufficient or excessive resection could compromise genome stability and cellular viability. While insufficient resection impairs the process of HR and ATR activation, over-resection could cause persistent checkpoint activation, loss of genetic information, and even cell death [123,137]. In fact, accumulation of ssDNA is a major source of mutational load and genomic rearrangements in different forms of cancer [138–140]. Hence, resection must be properly controlled to prevent under- or over-resection. To avoid HR in G1 phase of the cell cycle, DNA resection is also regulated by the cell cycle [9,11,13,23,25,74]. While the key steps of resection and core factors have been widely studied, many questions remain open as to precisely how the overall extent of resection is controlled. Below we will discuss the regulation of the resection process by the cell cycle, checkpoint response, and other factors.

Cell cycle regulation of resection

In G1 phase of the cell cycle, DSBs are repaired mainly by NHEJ or MMEJ, two pathways that require no or little end resection [8,9,74,89,90,141]. DNA end resection in G1 phase in general is suppressed by low activity of CDKs and higher activity of NHEJ factors [22,28]. Ku70-Ku80 protein heterodimer, a major NHEJ factor, loads onto DSB ends during G1 to promote repair by NHEJ while indirectly inhibiting DNA end resection [142–146]. Nevertheless, limited end processing is still possible during G1 due to the activities of MRN and CtIP [147,148]. However, this limited resection by MRN-CtIP during G1 could be mechanistically different from their resection function during S and G2 phases. Suppression of DNA end resection in G1 phase is important as it prevents HR between homologous chromosomes that can lead to loss of heterozygosity.

During S and G2 phases of the cell cycle, DSBs can be repaired by HR that requires more extensive resection to generate a significant length of ssDNA for Rad51 binding and homology search on a sister chromatid [9,14–16]. This increased resection results from the high level of CDK activity, which promotes the functions of the core resection factors including MRN, CtIP, Exo1, and Dna2 [26–29,149–156]. The NBS1 subunit of the MRN complex is phosphorylated by CDKs at S432 in S, G2, and M phases (but not in G1), which is important for DNA end resection [149,150]. Mre11 interacts directly with CDK2 and promotes phosphorylation of CtIP/Sae2 by CDK2, which is also crucial for resection in S and G2 phases [147,151–153]. In budding yeast, Sae2 is phosphorylated at S267 by Cdc28 (CDK1) and mutation of this residue to alanine inhibits DNA resection in vivo [152]. In human CtIP, the CDK phosphorylation sites S327 and T847 have been reported to be important for resection and subsequent HR in S and G2 phases of the cell cycle [153,154]. Phosphorylation of CtIP by CDK2 promotes resection in part through increased damage recruitment of CtIP and association with MRN complex [72,153]. CDKs also regulate resection extension by direct phosphorylation of Exo1 and Dna2, which also promotes their damage recruitment [155,156].

Compared to S and G2 phases, less is understood about DSB resection during M phase. The highly condensed nature of chromosomes in M phase may preclude the accessibility of repair factors to DSBs in M phase, thus cells could just exit mitosis with the DSBs that will be repaired by NHEJ in the next G1 phase [157,158]. A recent study in Xenopus egg extracts and cultured human cells shows that limited resection still occurs at DSBs during M phase by the activity of MRN and CtIP [159]. However, the high CDK1 activity also prevents the loading of ATR and Rad51 to the RPA-coated ssDNA. As a result, ATR checkpoint and HR are not activated in M phase [157,159].

Checkpoint regulation of resection

DNA end resection is also regulated by checkpoint kinases. Mass spectrometric studies have shown that hundreds of DDR proteins including major resection factors are phosphorylated by ATM/ATR after DNA damage [160,161]. ATM promotes the damage recruitment of CtIP in human cells and Xenopus egg extracts [51]. CtIP phosphorylation by ATR on T818 in Xenopus egg extracts also promotes its damage recruitment and resection activity [162]. In yeast, Sae2 is phosphorylated by both Mec1 (ATR) and Tel1 (ATM), which promotes its function in DNA end processing [163,164].

Interestingly, checkpoint kinases not only promote resection but also prevent unscheduled and over-resection by nucleases. Consistently, Mec1 deletion in yeast causes accelerated rate of DSB resection [21]. A study in Xenopus egg extracts showed that phosphorylation of Mre11 at SQ/TQ sites facilitates MRN complex dissociation from the damage site [165], which could be dependent on ATM/ATR to down-regulate Mre11 activity after initiation of resection. In human cells, ATM phosphorylates Mre11 on S676 and S678, which promotes Exo1 phosphorylation by ATM that attenuates its activity [34,35]. Mec1 and its downstream kinase Rad53 in budding yeast inhibit Exo1 activity at unprotected telomeres and prevent the accumulation of ssDNA [30]. Rad53-mediated phosphorylation of Exo1 attenuates its nuclease function and prevents uncontrolled resection at DSBs, telomeres, and stalled replication forks [32,33]. In human cells, ATR phosphorylates Exo1 in response to replication stress, which promotes Exo1 degradation to prevent the aberrant processing of replication forks [31]. Yeast Dna2 has also been suggested to be phosphorylated by checkpoint kinase Mec1, although whether this phosphorylation suppresses Dna2 resection activity remains to be determined [155]. Overall, it appears that the checkpoint kinases play a positive role in an early stage of resection and a negative role in a late stage of resection.

Regulation of resection by 53BP1 and BRCA1

The tumor suppressor BRCA1 promotes DNA end resection and is important for HR [72,166–169]. The HR defects of BRCA1-deficient cells are synthetic lethal with inhibitors of PARP1 that is involved in base excision repair [170,171]. Interestingly, the HR defects and cellular hyper-sensitivity to PARP1 inhibitors of BRCA1-deficient cells can be rescued by inactivation of 53BP1, which is important for NHEJ [168,172–177]. While the detailed mechanisms of their respective functions in HR and NHEJ are still incompletely understood, BRCA1 and 53BP1 act antagonistically to regulate DNA end resection. 53BP1 inhibits DNA end resection through its associated factors Rap1 interacting factor 1 homolog (RIF1) and pax transactivation domain interacting protein (PTIP) [178–183]. RIF1 inhibits BRCA1 damage recruitment during G1, inhibits resection, and hence promotes repair by NHEJ [179,180,183]. During S and G2 phases of the cell cycle, BRCA1 together with CtIP inhibits the damage recruitment of RIF1, allowing for resection and repair by HR [147,167,174,184]. BRCA1 also recruits the E3 ubiquitin ligase UHRF1 to the damage site where it ubiquitinates and removes RIF1 from the damage site, thereby promoting resection and HR [185]. While the mechanism of how BRCA1 inhibits PTIP is unclear, it may involve the disruption of its interaction with 53BP1 and damage association [174,178,182]. The striking functional relationship between BRCA1 and 53BP1 underscores the delicate balance between the HR and NHEJ pathways and the importance of proper regulation of the DNA resection process.

Mechanisms that prevent over-resection

DNA end resection must be properly controlled to prevent over-resection, as excessive ssDNA could cause cell death or genomic instability. Over-resection may result from unscheduled initiation, uncontrolled extension, or untimely termination. The function of Exo1 in resection is restrained by 14-3-3 proteins, which limit the damage recruitment of Exo1 by suppressing the binding of Exo1 to poly(ADP-ribose) (PAR) and PCNA, both of which promote Exo1’s damage association [104,127,186] (Fig. 3). Disruption of the Exo1-14-3-3 interaction causes over-resection and increased sensitivity to DNA damage [127]. Exo1 activity in DSB resection may also be regulated by post-translational modifications such as phosphorylation, SUMOylation, and ubiquitination. In budding yeast, Exo1 is phosphorylated by Rad53 in response to DSBs, telomere uncapping, and replication stress, which inhibits its nuclease activity [32,33]. In human cells, Exo1 is phosphorylated by ATR and SUMOylated by UBC9-PIAS1/PIAS4 in response to stalled replication, which induces its ubiquitination and degradation in a proteasome-dependent manner [31,126,187]. It is possible that similar mechanisms exist to limit Exo1 activity during resection of DSBs. Recent studies have shown that Dna2 resection activity is restrained by FANCD2 in human cells and Pxd1 in fission yeast, although the detailed mechanisms remain to be defined [135,136]. Studies in Xenopus egg extracts and human cells suggest that ATM-mediated phosphorylation of Mre11 inhibits its damage association as well as Exo1 nuclease activity preventing over-resection [35,165]. The function of CtIP in resection is negatively regulated by phosphorylation-specific prolyl-isomerase PIN1, which binds to CtIP, and promotes its isomerization and subsequent ubiquitination and degradation [188]. It is expected that these regulatory mechanisms function collectively to prevent uncontrolled excessive resection.

Figure 3.

Human Exo1 is regulated by PCNA, poly(ADP-ribosyl)ation, and 14-3-3s The damage recruitment and resection activity of Exo1 are controlled by three sets of factors that directly interact with different domains in Exo1. PARylated proteins bind to the N-terminus of Exo1 and promote the initial damage recruitment of Exo1. PCNA binds to the PCNA-Interacting Protein box in the C-terminus of Exo1 and promotes the damage retention and processivity of Exo1. 14-3-3 Proteins interact with the central domain of Exo1 and inhibit its damage recruitment by suppressing the interactions of Exo1 with PARylated proteins and PCNA. These coordinated regulations of Exo1 by multiple factors with opposing activities ensure a highly orchestrated resection process and a proper level of ssDNA at DSBs.

Other regulatory factors

In addition to the core factors described above, recent studies in multiple organisms have identified many other factors such as EXD2, PCNA, 9-1-1, PAR, lysine deacetylase SIRT6, chromatin-binding protein LEDGF/p75, chromatin remodelers SMARCAD1/Fun30 and SRCAP, ssDNA-binding protein SOSS1, and RNA-binding hnRNPU-like proteins in DNA end resection [104–106,186,189–197]. These factors promote resection by promoting the damage recruitment of core resection factors, remodeling the chromatin structure at damage sites or enhancing activities of resection nucleases. The existence of these many regulatory factors further demonstrates that DNA resection is a highly orchestrated process that involves sophisticated coordination of nuclease and helicase activities.

DNA Resection at Telomeres, Stalled Replication Forks, and Heterochromatin

The 3′ ssDNA overhangs at telomeres are essential for telomerase binding and telomere maintenance [198–200]. Replication of lagging strand DNA naturally generates ssDNA overhangs at telomeres in a sister chromatid. However, the leading strand is replicated completely, generating a blunt end that requires resection to produce a 3′ ssDNA overhang [198–200]. This resection is carried out by an exonuclease called Apollo, which acts immediately after the completion of replication [201,202]. Extensive resection by Apollo is inhibited by the binding of Pot1b to the ssDNA [201]. In yeast, Dna2 has also been suggested to be involved in limited processing of the 5′ strand to maintain the telomere length and telomerase binding [203,204]. While 5′ strand resection is important for telomere maintenance, over-resection could lead to telomere shortening, senescence, and other deleterious consequences [198]. Indeed, studies in yeast have shown that when the telomeric ends are not protected by capping proteins, Exo1 together with the Pif1 helicase could resect the ends of replicated DNA of both leading and lagging strands and initiates a protracted checkpoint response [205,206]. To avoid this, it has been shown in human cells that resection of telomere ends by Exo1 is inhibited by Pot1b and RIF1 [176,181,201,202].

DNA resection is also highly regulated in DNA replication. Studies in yeast and human cells suggest that Exo1 degrades stalled forks and that checkpoint-dependent phosphorylation of Exo1 inhibits this activity [31–33]. Mre11 has been suggested to play a major role in the processing and restart of stalled replication forks [43,207]. However, uncontrolled resection by Mre11 could also lead to degradation of stalled forks, leading to fork collapse. Indeed, it has been shown that BRCA2 and PARP1 inhibit Mre11 nuclease activity to prevent fork degradation [43,208]. WRN helicase also plays a major role in coordinating fork processing and restart by preventing unscheduled nascent DNA degradation by Mre11 and Exo1 [209,210]. In Xenopus egg extracts, Rad51 binds to the newly synthesized DNA during replication and protects it from degradation by Mre11 [211]. In yeast and human cells, Dna2-mediated end processing is important for the restart of stalled or reversed replication forks [44,45,212]. However, upon replication fork stalling caused by interstrand crosslink, Dna2 activity is restrained by FANCD2 to prevent uncontrolled resection [135].

To date, little is known about the mechanisms and regulation of DNA end resection in heterochromatin. Recent studies in Drosophila melanogaster suggest that resection occurs efficiently in heterochromatin; however, DSBs are not repaired by HR until the DNA ends relocate outside of the heterochromatin domain [213,214]. The relocalization of DSB is facilitated by resection and ATR [213]. Counter-intuitively, resection of DSBs in heterochromatin and subsequent loading of ATR interacting protein (ATRIP) and TopBP1 required for ATR activation appear to occur in a faster kinetics than in euchromatin, suggesting that resection is regulated differently in these two types of chromatin domains [213–215].

Relevance of DNA End Resection for Cancer Formation and Therapy

DNA end resection is essential for ATR checkpoint activation and HR, both of which play a critical role in genome maintenance and tumor suppression [1–4,7,10,18]. Genetic knockout of the major resection factors Mre11, Rad50, NBS1, CtIP, and Dna2 in mice is embryonically lethal, and their deficiencies cause hypersensitive to DNA damaging agents [64,112,216–218]. Exo1 knockout in mice leads to meiotic defects and cancer susceptibility [219]. Mutations in Mre11, NBS1, Rad50, BLM, and WRN are causes of genetic diseases AT-like Disorder, Nijmegen Breakage Syndrome, NBS-like Disorder, Bloom Syndrome, and Werner Syndrome, respectively, all of which are associated with cancer predispositions [3,220–224]. These findings further highlight the importance of DNA end resection in genome protection and tumor suppression. Paradoxically, DNA end resection may also be targeted for cancer therapy. The synthetic lethal relationship between PARP inhibition and HR deficiency suggests that combining inhibitors of PARPs with that of resection activities may be effective in cancer treatment [170,171]. Moreover, over-resection (e.g. by disrupting Exo1-14-3-3 interaction) increases cellular sensitivity to DNA damage, and thus may also be exploited for cancer treatment [127].

Conclusions and Perspective

DNA end resection is a key process in the DDR that controls both DNA repair and checkpoint response. Resection is initiated by an endocleavage step that is carried out by the MRN complex in collaboration with CtIP. Extended resection is mediated by two parallel pathways involving the Exo1 and Dna2 nucleases, respectively. The resection process is tightly regulated by multiple mechanisms and accessory factors to ensure proper repair at DSBs, telomere, replication forks, and heterochromatin. The DNA resection process is highly relevant to tumorigenesis and may be targeted for cancer therapy. Understanding the detailed mechanisms and regulation of DNA resection is the key to designing more efficient cancer therapeutics. Future work is needed to address many outstanding questions in the field, e.g. what determines the strand specificity during resection initiation? How do MRN and CtIP cooperate to initiate DNA resection? Are there any mechanistic differences in the initiation of resection of DNA ends with different structures? Do Exo1 and Dna2 pathways function redundantly at telomeres and stalled/collapsed replication forks? How is resection terminated? How is over-resection avoided? What does control the extent of DNA resection? How is resection regulated in heterochromatin? Can DNA end resection be exploited for cancer therapy, and if so, which resection activities can be targeted? The next few years will see major advances in addressing these important questions.

Funding

The work was supported by the grants from the National Institutes of Health (No. R01GM098535) and the American Cancer Society Research Scholar Grant (No. RSG-13-212-01-DMC).