Detection and repair of ionizing radiation induced DNA double strand breaks: new developments in non-homologous end joining

Summary

DNA double-strand breaks (DSBs) are considered the most cytotoxic form of DNA damage. In human cells, the major pathway for the repair of ionizing radiation (IR)-induced DSBs is non-homologous end joining (NHEJ). Here we discuss recent developments in our understanding of the mechanism of NHEJ, the proteins involved and its regulation.

Introduction

Non-homologous end joining (NHEJ) is the major pathway for the repair of ionizing radiation (IR)-induced DNA double strand breaks (DSBs) in human cells and is also essential for V(D)J recombination and the development of mature T and B cells [1]. Cells compromised for NHEJ therefore have major defects in DSB repair, are radiation sensitive and, at the organismal level, are immune deficient. Given that the majority of cancer patients are treated with IR in the form of radiation therapy and/or with chemotherapeutic drugs that produce DSBs, a thorough understanding of the molecular mechanisms by which cells detect and repair DSBs is of critical importance to improving the effectiveness of existing cancer therapies as well as the development of new approaches to target human malignancies. We recently reviewed the mechanism of NHEJ in mammalian cells [2]. Here, we provide an update to that review, highlighting many of the significant developments in the field since that time.

A model for NHEJ

For convenience NHEJ can be divided into three stages, 1) end-detection and tethering, 2) processing and 3) ligation (Fig. 1). In step one, the Ku70/80 heterodimer detects and binds to the extreme termini of the DSB, acting as the signal for the recruitment of subsequent NHEJ factors, including the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), a serine threonine protein kinase of the phosphatidyl inositol 3-kinase (PIKK) family. Together, DNA-PKcs and Ku tether the DNA ends in a synaptic complex [3]. The interaction with DSB-bound Ku enhances DNA-PKcs’ protein kinase activity [4], which is required for NHEJ [5,6]. The second step in NHEJ involves enzymatic processing of DNA ends. IR frequently produces DNA termini containing non-ligatable 3′-phosphate groups, 3′-phosphoglycolates or 5′-hydroxyl groups that must be converted to 3′-hydroxyls and 5′-phosphates prior to ligation. Removal of these non-ligatable end groups may involve polynucleotide kinase/phosphatase (PNKP), a DNA 3′-phosphatase/5′-kinase [7], and possibly tyrosyl-DNA phosphodiesterase 1 (TDP1), which can convert 3′-phosphoglycolates to 3′-phosphates that are subsequently removed by PNKP [8,9]. Similarly, aprataxin (APTX) may have a role in removal of abortive ligation products at DSB ends [10]. Nucleases such as Exonuclease 1 (Exo1) [11], Mre11 [12], Artemis [13] and the Werner syndrome helicase/exonuclease, WRN [14], may also be involved in processing damaged DNA termini, to reveal regions of DNA sequence micro-homology that help position ends for ligation, while nucleotide gaps or deletions are filled in by DNA polymerases μ and λ [15] (Fig. 1). Processing of DSB ends is predicted to lead to loss or modification of nucleotides from either side of the break, making NHEJ inherently error prone. In the final step of NHEJ, the DNA ends are re-ligated by DNA ligase IV (LIG4) which exists in complex with X-ray cross complementing gene 4 (XRCC4) and XRCC4-like factor (XLF, also known as Cernunnos) [1,2] (Fig. 1).

Figure 1.

A model for NHEJ, showing the three basic steps (end detection, processing and ligation) along with the proteins implicated in each stage. Interactions are shown by black lines with solid arrowheads. Putative interactions are in dashed lines. Blue lines indicate phosphorylation events. Blue, red and black Ps enclosed in circles indicate DNA-PK, ATM and CK2-mediated phosphorylation events, respectively. The AP lyase activity of Ku is indicated by the red arrow. For simplicity, the interaction between Artemis and DNA ligase IV [49,50] is not shown. See text for details.

It is important to note that apart from initial binding of Ku and ultimate ligation by LIG4, there is considerable uncertainty regarding the timing and order of the intervening steps during NHEJ. Indeed, rather that proceed in a stepwise fashion, NHEJ may take place within a dynamic, multi-component protein-DNA complex. Moreover, after DSB detection by Ku, subsequent recruitment of specific end processing factors may depend upon the nature of the lesion to be repaired and/or cell cycle stage. For example, DNA-PKcs may only be required for a subset of more slowly repaired, complex lesions [16], while ATM (Ataxia Telangiectasia mutated), DNA-PKcs and Artemis are required for repair of DSBs in heterochromatic regions [17,18].

Ku

The Ku70/80 heterodimer is regarded as the cornerstone of NHEJ and has been described as a “tool-belt” protein onto which other proteins are recruited [19]. It detects DSBs through its high affinity for double stranded (ds) DNA ends and is required for recruitment of DNA-PKcs to DNA in vitro [20] and to sites of DNA damage in vivo [21]. Small angle X ray scattering (SAXS) revealed that the C-terminal region (CTR) of Ku80 forms a long “arm” that extends from the DNA binding core, suitable for interacting with other proteins [22]. Since the extreme C-terminus of Ku80 interacts with DNA-PKcs [23,24], it is tempting to speculate that it may serve to recruit DNA-PKcs to DSBs, however deletion of the Ku80 CTR did not abolish recruitment of DNA-PKcs to DSBs, DNA-PKcs kinase activity in vitro or DNA-PKcs autophosphorylation on Ser2056 in vivo [13]. Therefore, precisely how DNA-PKcs is recruited to Ku at DSBs and how this leads to activation of DNA-PKcs kinase activity remains unclear.

Ku also interacts with WRN [25], and with XRCC4, which enhances the binding of XRCC4/LIG4 to DNA in vitro, and is required for recruitment of XRCC4 to DSBs in vivo [26,27]. Ku also interacts with XLF, and is also required for XLF recruitment to DSBs in vivo [28,29]. Moreover, Ku is required for recruitment of APLF (APTX and PNK-like factor) to DSBs, which facilitates stability of the repair complex and promotes NHEJ [30] (Fig. 1).

In addition to these critical roles in detection of damage and recruitment of other NHEJ proteins, Ku has 5′-deoxyribose-5-phosphate (5′-dRP)/AP lyase activity, suggesting that it also plays an enzymatic role in processing DNA ends during NHEJ [31]. Ku excises abasic sites near DSBs in vitro, and this activity was highest when the abasic site was within a short 5′ overhang at the DSB end, suggesting that Ku promotes NHEJ fidelity [32]. This novel activity of Ku may be important in repairing IR-induced base damage in the vicinity of the strand breaks, contributing to genomic stability.

DNA-PKcs

DNA-PKcs is composed of a large N-terminal domain composed largely of helical elements and HEAT (Huntingtin, Elongation Factor 3, PP2A and TOR1) repeats, a conserved region termed the FAT (FRAP, ATM, TRRAP) domain, followed by the kinase domain and a short C-terminal region, termed the FATC domain (Fig. 2A). Structural features of DNA-PKcs inferred from cryo-electron microscopy (cryo-EM) [33,34] and SAXS [22] suggested that the protein is composed of a head or crown domain and a central palm domain with a region of low electron density in the centre, proposed to bind dsDNA [35]. These features were confirmed by the elucidation of the X-ray structure of DNA-PKcs [36]. Although still relatively low-resolution (6.6 Å), this structure significantly extends our understanding of DNA-PKcs structure, revealing that the FAT-kinase-FATC domains are located at the head/crown region while the N-terminal α-helical domain forms the arms of a pincer-shaped structure surrounding an open, central channel that is suggested to bind dsDNA (Fig. 2B). A smaller channel in the head domain may accommodate single stranded (ss) DNA [22,33], possibly reflecting unwinding of dsDNA termini within the DNA-PKcs molecule [3]. Future studies aimed at determining precisely how DNA-PKcs interacts with DNA will be key to understanding its mechanism of action.

Figure 2.

A.Schematic of DNA-PKcs showing major in vivo phosphorylation sites discussed in text. Residues in red are equivalent to residues mutated in the 3A mice [66]. Domain boundaries are shown by vertical numbers. See also [37,62,67].

B. Model for structure of DNA-PKcs, adapted from [36].

C. Model for putative effects of phosphorylation on DNA-PKcs structure, adapted from [37].

Significantly, regions at the apex of the arms of DNA-PKcs were predicted to be highly flexible and the region of DNA-PKcs at the base of the arms contained a gap, suggesting that the arms may open and close around the central DNA binding channel [36]. These findings led us to suggest that phosphorylation of DNA-PKcs (discussed below) might mediate opening and closing of the clamp shaped arms of DNA-PKcs to regulate its interaction with Ku and DNA [37] (Fig. 2C). Since the overall features of the PIKK family of proteins are conserved [38], it seems likely that the overall pincer shaped structure revealed for DNA-PKcs may be conserved in other family members.

XRCC4 and XLF

Although it has no direct catalytic activity, XRCC4 plays an essential role in NHEJ due to its ability to interact with and stabilize LIG4. XRCC4 is constitutively phosphorylated by CK2 on Thr233, which interacts with the fork-head associated (FHA) domains of PNKP, APTX and APLF, suggesting that XRCC4 recruits these factors to sites of DNA damage (Fig. 1). Like XRCC4, XLF has no catalytic activity but is able to facilitate XRCC4-LIG4-mediated end joining, particularly on non-cohesive ends [39-41] and promotes re-adenylation of LIG4, recharging it for the next catalytic cycle [42]. How this occurs is unclear but is proposed to involve XRCC4 and XLF-induced conformational changes in LIG4 [43]. XRCC4 and XLF dimers interact via their head domains to form long, helical filaments that may serve to bridge or align DNA ends for ligation [44-47], which could protect DNA ends from nucleases, regulating resection and possibly pathway choice [48].

DNA ligase IV (LIG4)

Perhaps the most critical step in NHEJ is ligation of the DNA ends to repair the break. LIG4 is composed of N-terminal DNA binding and catalytic domains followed by a C-terminal tandem BRCT domain. The N-terminal DNA binding domain of LIG4 interacts with Artemis, to regulate NHEJ in V(D)J recombination [49,50], while the C-terminal BRCT domains interact with the coiled-coil stalk of XRCC4 [51,52], which is required for LIG4 stability. LIG4 destabilizes XRCC4-XLF filaments, suggesting that it regulates their formation [43,53]. The recently reported high-resolution structures of the N-terminal nucleotidyltransferase [43], DNA binding [50] and BRCT domains [52] of LIG4 in combination with results from SAXS [43,44,53] and cryo-electron microscopy [54] shed light on the overall structures of XRCC4-LIG4 and the XLF-XRCC4-LIG4 complex, however, a more complete understanding of the mechanism of DNA ligation will require higher resolution structures of the complex in association with DSBs.

Removal of Ku from DNA

Since the first demonstration that the Ku70/80 heterodimer forms a basket shaped structure that encircles dsDNA [55], the question of how Ku is removed from the DNA ends prior to ligation has remained enigmatic. Studies in Xenopus laevis egg extracts suggest that polyubiquitylation of Ku80 at Lys48 can mediate its degradation by the proteasome and dissociation from DNA [56,57]. In keeping with this, the E3 ubiquitin ligase RNF8 interacts with Ku80 in human cells, and overexpression of RNF8 decreased Ku80 levels in vivo. Moreover, depletion of RNF8 resulted in prolonged retention of Ku80 at laser-induced DSBs in vivo and reduction of NHEJ repair efficiency, suggesting that ubiquitylation of Ku80 by RNF8 is essential for Ku degradation and possibly its release from DSBs [58]. Other studies suggest that PARP (poly (ADP-ribose) polymerase) is required for retention of Ku since in Dictyostelium discoideum, PARP mutation results in reduced NHEJ efficiency possibly through promoting removal of Ku at DSBs [59]. Intriguingly, in the fission yeast Schizosaccharomyces pombe, the Mre11 nuclease is proposed to nick one strand of the DNA, allowing Exo1-mediated DNA resection and removal of Ku [60]. Whether Mre11 and Exo1 carry out a similar function in mammalian cells remains to be determined.

The role of phosphorylation in NHEJ

Although DNA-PKcs is not conserved in lower eukaryotes, the protein kinase activity of DNA-PK is required for NHEJ in mammalian cells [5] and small molecule inhibitors of DNA-PK kinase activity inhibit DSB repair and induce radiosensitivity [6]. DNA-PKcs undergoes autophosphorylation in vitro [61] and is highly phosphorylated in cells in response to DNA damage (reviewed in [37,62]). In vitro, autophosphorylation results in loss of DNA-PK kinase activity and dissociation of DNA-PKcs from the DNA-Ku complex [63]. Moreover, autophosphorylation induces a dramatic conformational change [22] proposed to facilitate its dissociation from dsDNA [37] (discussed above).

The most well-studied DNA-PKcs phosphorylation sites are located between residues 2023-2056 and 2609-2647, and have been termed the PQR/Ser2056 and ABCDE/Thr2609 clusters, respectively [62,64] (Fig. 2A). Cells expressing DNA-PKcs in which the ABCDE sites have been ablated are extremely sensitive to IR, more so than cells lacking DNA-PKcs [65]. Moreover, mutation of phosphorylation sites in the PQR and ABCDE clusters regulates processing of coding joints during V(D)J recombination in a reciprocal manner, consistent with DNA-PK phosphorylation at these sites regulating end access and processing of DNA ends (reviewed in [62,64]). Significantly, transgenic mice expressing DNA-PKcs in which residues equivalent to Thr2609, Thr2638, and Thr2647 in humans have been mutated to alanine, termed 3A mice, have severe defects in hematopoietic stem cell development and die shortly after birth. These mutations also increased spontaneous DNA lesions in hematopoietic stem cells, as well as hypersensitivity to replication stress inducers [66].

Other phosphorylation sites of DNA-PKcs that are important for function include Thr976/Ser1104 and Ser51/Ser72 in the N-terminal domain (termed JK and N clusters, respectively) [67], and Thr3950 in the catalytic domain [68]. In contrast, Ser3205 in the FAT domain is phosphorylated in an ATM-dependent manner in response to IR but is not required for NHEJ [67]. Fully understanding the effects of DNA-PKcs phosphorylation on its function represents a considerable challenge, as DNA-PKcs is phosphorylated at over 40 sites in vitro (Yu and Lees-Miller, unpublished data) and phosphoproteomics studies have identified over 37 in vivo phosphorylation sites and 17 acetylation sites (reviewed in [37]).

DNA-PKcs phosphorylation is also regulated by protein phosphatases [61,69]. Indeed, DNA-PKcs interacts with protein phosphatases PP5 [70] and PP6 [71,72] and possibly PP1 [73] and PP2A [71], suggesting that it may act as a scaffold to recruit protein phosphatases to sites of DNA damage [74]. Moreover, PP4 and PP2A have been implicated in NHEJ regulation [75,76].

Given that DNA-PK protein kinase activity is required for DSB repair, we and others have long sought to identify physiological substrates of DNA-PKcs. Although DNA-PK phosphorylates Ku70 and Ku80 in vitro, these sites are not required for survival after IR [77]. Similarly although the CTRs of XRCC4 and XLF are phosphorylated in vitro and in vivo [78,79], and although the extreme CTR of XLF is required for interaction with Ku and for recruitment to DSBs in vivo [29], the CTRs of XRCC4 and XLF are not required for cellular survival after X-ray irradiation or for V(D)J recombination [80,81]. However, functional redundancy of the CTRs of may have masked effects of mutation of XLF phosphorylation sites on NHEJ [82], and, in light of this, the effects of XRCC4 and XLF phosphorylation on NHEJ and DSB repair need to be re-evaluated.

ATM and DNA-PK phosphorylate PNKP on Ser114 and Ser126 in vitro, and PNKP phosphorylation in response to DNA damage is largely ATM dependent [83,84]. Phosphorylation of these sites is required for efficient DSB repair, retention of PNKP at sites of DNA damage and cell survival after DNA damage [83,84]. The fact that in vivo phosphorylation of Artemis [85,86], PNKP [83,84], XLF [78] and DNA-PKcs [87] can be ATM-dependent may reflect cross talk between DNA damage signaling and repair pathways or, possibly, ATM-dependent repair of heterochromatic DSBs [17,88].

New substrates and roles of DNA-PKcs

Phosphoproteomics studies have identified hundreds of new DNA-damage induced phosphorylation events in cells [89-92], many of which have been attributed to the activity of ATM [89,92]. However, others are likely phosphorylated by DNA-PK and/or ATR. Moreover, DNA-PK phosphorylates several ATM target proteins in ATM-defective cells, revealing functional redundancy between PIKKs [71,93,94].

Other newly identified in vivo substrates for DNA-PKcs include nuclear receptor 4A (NR4A) [95] (Table 1), which was shown to co-localize with γ-H2AX, a marker of DSBs, and cells in which NR4A was depleted have defects in DSB repair. The heat shock protein Hsp90α [96,97] and the scaffold attachment factor A (SAF-A, also known as heterogenous nuclear ribonucleoprotein, hnRNPU) are also phosphorylated by DNA-PK in vivo in response to DNA damage, [98,99], however whether these phosphorylation events directly impact NHEJ is not known. Moreover, it is becoming clear that DNA-PKcs functions in cellular processes other than NHEJ, such as transcription and mitosis (Table 2), and it is possible that it may target proteins in other pathways in vivo.

Table 1. New in vivo substrates of DNA-PKcs.

Protein Name	Phosphorylation site	Induced by	References
Akt1/PKBα	Thr-308 Ser-473	IR	[117,118]
Hsp90α	Thr-5 Thr-7	IR or apoptosis	[96,97]
NR4A	Ser-337	IR or camptothecin	[95]
SAF-A/hnRNPU	Ser-59	Calicheamicin or etoposide	[98,99]
UPF1	Unknown	Replication stress	[140]
USF1	Ser-262	Insulin	[141]

Table 2. New functions of DNA-PKcs.

Function	Action by DNA-PKcs	References
Mitotic regulation and maintenance of genome stability	May stabilize centrosome and spindle formation and alignment of chromosomes	[142-144]
Transcriptional regulation of estrogen responsive genes	Localizes with Ku, PARP1, and Topo II β at the promoter region of this gene	[145]
Regulation during replication stress and control of histone mRNA stability	Phosphorylation of UPF1 and regulation of histone mRNA stability	[140]
Regulation in response to insulin signaling	Regulation of FAS gene in response to insulin signaling through phosphorylating USF1	[141]
Cellular adaptation to hypoxia stress response	DNA-PKcs phosphorylation on S2056 initiated by histone acetylation in response to hypoxia	[146]

Regulation of NHEJ

APLF has emerged as a major player in regulation of NHEJ. It interacts with CK2 phosphorylated XRCC4 [100] and with Ku [30,101] and stimulates NHEJ in vivo by promoting retention of the XRCC4/LIG4 complex at DSBs [30,102]. APLF also interacts with PARP3 [102] and histones [103,104], providing a potential mechanism to interface NHEJ with chromatin. The primate specific DNA methylase, metnase also regulates NHEJ in the context of chromatin, by regulating histone methylation surrounding DSBs [105,106]. Moreover, histone deacetylases HDAC1 and HDAC2 are required for cellular survival after IR, promote NHEJ, regulate retention of Ku, XRCC4 and Artemis at sites of DNA damage, and DNA damage induced phosphorylation of DNA-PKcs [107], linking NHEJ with chromatin structure.

Another important modulator of NHEJ is 53BP1. 53BP1 is recruited to IR-induced foci (IRIF) through RNF8-dependent histone ubiquitylation [108-110] and/or MDC1-dependent histone methylation [111,112], and promotes NHEJ through its ability to bridge DSBs over long ranges of chromatin [113,114]. In addition, 53BP1 promotes NHEJ and suppresses DNA end resection, possibly by forming a chromatin barrier around breaks [115], or by inhibiting Exo1-mediated DNA resection activity [116].

Akt1 (also known as PKBα) has also been shown to regulate NHEJ. Akt1 interacts with DNA-PKcs, promoting DNA-PK kinase activity, and enhances cellular radioresistance possibly by facilitating DNA-PKcs recruitment to DSBs and improving NHEJ repair efficiency (Table 1) [117,118]. DNA-PKcs also interacts with epidermal growth factor receptor (EGFR), and EGFR stabilizes phosphorylation of DNA-PKcs on T2609 and may promote Artemis-mediated end processing at DSBs [119-121].

DSB pathway choice

One of the critical questions in DSB repair is what determines whether a DSB is repaired via the classical NHEJ pathway (described here), an alternative NHEJ pathway (Alt-NHEJ, which is characterized by larger deletions and translocations) [122], or the more accurate homologous recombination (HR) pathway [123] (Fig. 3). One major factor is clearly cell cycle stage, as NHEJ occurs in all stages of the cell cycle whereas HR is active only after DNA replication when an undamaged sister chromatid is present as a template. Another major factor in pathway choice is resection of the DNA ends. Current models suggest that blunt or modestly processed DNA ends are targeted by Ku70/80 for NHEJ whereas CtIP-mediated resection by Mre11, Exo 1 and/or other nucleases in late S and early G2, primes cells for loading of Rad51 and consequently initiation of HR [124,125]. In the fission yeast Schizosaccharomyces pombe, Mre11 promotes Exo-1 mediated DNA resection, which promotes HR [60].

Figure 3.

A model for the DNA damage response in mammalian cells indicating competition of DSB ends by classical NHEJ, Alt-NHEJ and HR. Adapted from [139] with permission.

In mammalian cells, DNA-PKcs phosphorylation may also regulate pathway choice. The protein kinase activity of DNA-PKcs suppresses HR and phosphorylation of DNA-PKcs on the JK cluster or Thr3950 inhibits NHEJ and promotes HR [67]. Moreover, cells from 3A mice lacking three of the ABCDE phosphorylation sites (discussed above), showed lower HR activity than those from DNA-PKcs-null mice, and cells expressing phosphoablation mutations in the ABCDE cluster have slower rates of Rad51 foci formation [105,126]. Together, these studies are consistent with phosphorylation of DNA-PKcs regulating access of the DNA ends to resection factors and initiation of HR.

Therapeutic potential of NHEJ

Given the importance of NHEJ in the repair of IR-induced DSBs, NHEJ has been considered a potential therapeutic target for cancer treatment [127]. Indeed, inhibitors of DNA-PKcs kinase activity have been shown to have efficacy in human cells, two are in phase I clinical trials (Table 3) and others are in development [128-131]. A small molecule inhibitor of PNKP that sensitizes cells to camptothecin has also been developed [132,133]. However, the therapeutic implications of targeting NHEJ will depend in part on NHEJ proficiency in tumour versus normal cells. Recent studies have also suggested that NHEJ can modulate the cellular response to other therapeutic strategies. For example, PARP inhibitors have shown considerable potential in the treatment of HR deficient tumors [134] and this effect is dependent on functional NHEJ [135,136]. Moreover, inhibition of NHEJ suppresses sensitivity to interstrand crosslinking agents in Fanconi Anemia [137,138].

Table 3. NHEJ inhibitors.

PI3K, phosphatidyl inositol 3 kinase.

Inhibitor	Mechanism/Comments	References
A12B4C3	PNKP inhibitor, sensitizes cells to camptothecin	[132,133]
BTW3	A small peptide DNA-PK inhibitor, proposed to compete for DNA-PKcs autophosphorylation site.	[147]
KU0060648	DNA-PK and PI3K inhibitor.	[148]
NU7441/KU57788	DNA-PK inhibitor, competitive with ATP	[6,148]
ScFv 18-2	An antibody-derived DNA-PK inhibitor that can bind to an epitope unique to DNA-PKcs.	[149]
ZSTK474	DNA-PK and PI3K inhibitor, competitive with ATP; in phase 1 clinical trials (NCT01280487)	[150-152]
CC-115	Dual inhibitor of DNA-PKcs and mTOR, in phase 1 clinical trials.	NCT01353625
CC-122	DNA-PK inhibitor, in phase 1 clinical trials.	NCT01421524

Conclusions

In this review we summarized the many recent significant findings related to NHEJ in human cells. Despite these advances, many critical questions remain. One is how the different NHEJ proteins are coordinated to access and dissociate from DSBs in vivo and the role of phosphorylation and other post-translational modifications in this process. Other critical questions include what determines access of Ku and MRN to DSB ends to initiate NHEJ, ATM-mediated signaling and HR, how pathway choice is regulated, and how repair occurs in the context of chromatin. Further understanding of the mechanism of NHEJ is critical to understanding the effects of IR on human cells, which has the potential to improve existing cancer therapies, as well as developing new cancer biomarkers and novel therapeutics.