Radiation-induced clustered DNA lesions: Repair and mutagenesis

Author’s Accepted Manuscript Radiation-induced clustered DNA lesions: repair and mutagenesis Evelyne Sage, Naoya Shikazono

www.elsevier.com

PII: DOI: Reference:

S0891-5849(16)31095-4 http://dx.doi.org/10.1016/j.freeradbiomed.2016.12.008 FRB13114

To appear in: Free Radical Biology and Medicine Received date: 2 October 2016 Revised date: 5 December 2016 Accepted date: 7 December 2016 Cite this article as: Evelyne Sage and Naoya Shikazono, Radiation-induced clustered DNA lesions: repair and mutagenesis, Free Radical Biology and Medicine, http://dx.doi.org/10.1016/j.freeradbiomed.2016.12.008 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 galley 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.

Radiation-induced clustered DNA lesions: repair and mutagenesis Evelyne Sage1*, Naoya Shikazono2 1Institut

Curie, PSL Research University, CNRS, UMR3347, F-91405 Orsay, France

2Quantum

Beam Science Research Directorate, National Institutes of Quantum and

Radiological Science and Technology, Kansai Photon Science Institute, 8-1-7 Umemidai, Kizugawa-Shi, Kyoto 619-0215, Japan. [email protected] [email protected] *Corresponding

author : Institut Curie, Centre Universitaire, F-91405 Orsay, France

Abstract Clustered DNA lesions, also called Multiply Damaged Sites, is the hallmark of ionizing radiation. It is defined as the combination of two or more lesions, comprising strand breaks, oxidatively generated base damage, abasic sites within one or two DNA helix turns, created by the passage of a single radiation track. DSB clustered lesions associate DSB and several base damage and abasic sites in close vicinity, and are assimilated to complex DSB. Non-DSB clustered lesions comprise single strand break, base damage and abasic sites. At radiation with low Linear Energy Transfer (LET), such as X-rays or J-rays clustered DNA lesions are 3-4 times more abundant than DSB. Their proportion and their complexity increase with increasing LET; they may represent a large part of the damage to DNA. Studies in vitro using engineered clustered DNA lesions of increasing complexity have greatly enhanced our understanding on how non-DSB clustered lesions are processed. Base excision repair is compromised, the observed hierarchy in the processing of the lesions within a cluster leads to the formation of SSB or DSB as repair intermediates and increases the lifetime of the lesions. As a consequence, the chances of mutation drastically increase. Complex DSB, either formed directly by irradiation or by the processing of non-DSB clustered lesions, are repaired by slow kinetics or left unrepaired and cause cell death or pass mitosis. In surviving cells, large deletions, translocations, and chromosomal aberrations are observed. This review details the most recent data on the processing of non-DSB clustered lesions and complex DSB and tends to demonstrate the high significance of these specific DNA damage in terms of genomic instability induction. Abbreviations: LET, linear energy transfer; MDS, multiply damaged site; SSB single strand break; DSB, double strand break; AP sites, abasic sites; 8-oxoG, 8-oxo-7,8-dihydroguanine; 5-ohU,

5-hydroxyuracil; Tg, 5,6-dihydroxy-5,6-dihydrothymine; U, uracil; BER, base excision repair; NHEJ, non-homologous end joining; HR, homologous recombination; bp, base pair; IRIF, ionizing radiation-induced foci. Keywords: Clustered DNA lesions; multiply damaged sites; non-DSB clustered lesions; complex DSB; oxidatively generated DNA damage; base excision repair; DNA repair; ionizing radiation

Introduction About two third of the DNA damage induced by low linear energy transfer (LET) radiation (e.g., X-rays and J-rays) are formed by reactive oxygen species, including the highly damaging hydroxyl radical, produced by water radiolysis in the vicinity of DNA [1, 2]. Thus, ionizing radiation induces random (isolated) single strand breaks (SSB), oxidation reactions to the sugar moieties which partly lead to the formation of SSB, oxidized/reduced base damage, base loss, many of which do not chemically differ from that formed during endogenous oxidative metabolism or by oxidizing agents. Meanwhile, at equal number of oxidatively generated DNA damage, ionizing radiation is far more toxic than hydrogen peroxide [3, 4]. Modeling of the radiation track structures ([1, 5] evidenced that the energy deposition of low LET radiation leads to two or more ionizations within a radius of 1-4 nm, like the diameter of the DNA double helix and its water layers. Multiple ionizations result in multiply damaged sites, e.g. clustered DNA lesions, which consists in >2 SSB, abasic (AP) sites, oxidized purine or pyrimidine bases, or double strand breaks (DSB), formed within one or two helix turns from the same energy deposition event. Overall, ionizing radiation, like oxidative metabolism, produces isolated SSB, AP sites, oxidatively generated base damage, but also produces DSB, DNA protein crosslinks and clustered DNA lesions (Table 1) which are rarely formed by endogenous oxidative stress but recognized as largely responsible for the toxicity of ionizing radiation. High LET radiation (e,g., D-particles, protons and carbon-ions such as used in hadrontherapy) has an elevated propensity to form clustered DNA lesions of higher complexity in comparison with low LET radiation. This is well schemed in a recent review by Georgakilas et al ([6]; see Figure 2 of the cited review). In fact, high LET radiation shows a greater relative effectiveness than low LET radiation regarding many cellular end points and they are more harmful. Clustered DNA lesions, as soon as they were predicted by chemists and physicists, were suspected to have a low ability to be repaired, due to the peculiar spatial distribution of the lesions [1, 5]. In addition, for the same reason, they cannot be measured by the analytical methods used for isolated oxidatively generated lesions, and until now there

is no reliable methods for their precise quantification. The first observation was made in irradiated plasmid DNA [7], extended by Sutherland findings in phage DNA and mammalian cells [8]. Nevertheless, a good estimation was first provided by Sutherland and coworkers, who demonstrated that non-DSB clustered lesions (Figure 1), carrying oxidized purine or pyrimidine bases, AP sites or SSB, comprise the majority (70-80%) of clustered DNA lesions in mammalian cells and are 3-4 times more frequent than prompt DSB by low LET radiation [8-12]. Isolated oxidatively generated DNA lesions (SSB and oxidized base) are known to be efficiently and accurately repaired by base excision repair (BER) which is largely reviewed in this issue. Radiation-induced simple prompt DSB, although carrying 3’-ends phosphate and phosphoglycolate that require processing (in contrast to DSB with ligatable ends produced by restriction enzymes) can also be efficiently and accurately repaired within 1-2 hours by non-homologous end joining (NHEJ), or in G2 phase by homologous recombination (HR), in mammalian cells. A small fraction of radiation-induced DSB, and in particular those induced by high LET radiation, often called complex DSB, are known to have a slow rate of repair and to require DNA damage response via the ATM pathway [13]. Sutherland and collaborators have reported that the non-DSB clustered lesions produced in mammalian cells by radiation show a greatly longer lifetime than that of isolated DNA lesions [8, 10, 14]. The last two decades the strategy that consists in building synthetic non-DSB clusters of various complexity, carried on oligonucleotides, and in using them to study DNA repair in test tubes or in bacterial, yeast or mammalian cells, has greatly helped understanding the processing of clustered DNA lesions at the molecular level (reviewed in [4, 15-18]). Indeed, studies of survival, mutation induction and repair of/in plasmid carrying non-DSB clusters of various nature and complexity, in Escherichia coli, yeast and mammalian cells have been of utmost importance. An attempted simultaneous repair of opposed base damage or AP sites at non-DSB clusters may result in the formation of DSB [4]. Measuring DSB post-irradiation in mammalian cells and bacteria has revealed an increase in DSB when cells were allowed time for repair [19-21]. Such a DSB produced as repair intermediate may be surrounded by 1-2 unrepaired base damage and become an unrepairable, complex DSB. Alternatively, a non-DSB cluster may be resistant to glycosylases and endonucleases and the persistent lesions within the cluster may lead to mutations. Clustered DNA lesions, including DSB and non-DSB clusters, challenge the DNA repair and the outcome of a cell. They are considered as responsible for the genomic damage and instability inflicted by ionizing radiation. This mini-review will focus on the most recent aspects of the formation, repair and biological consequences of clustered DNA lesions induced by ionizing radiation.

Formation, detection and structural features of clustered DNA lesions Clustered DNA lesions are considered as a hallmark of ionizing radiation exposure, as their endogenous production in cells is very low [22, 23] and due to their mode of formation. The simplest clustered DNA lesion is the DSB (2 opposed SSB in a close vicinity). Complex DSB and non-DSB clusters can carry as many as 10 lesions per damaged site for low LET radiation, and more for high LET radiation, according to theoretical calculation [24]. Indeed, the average number of lesions per cluster tends to increase with increasing LET. For example, 1MeV Protons (LET 25,4 keV/μm) are predicted to produce DNA lesions with a ratio of 1 five-lesion cluster : 20 two-lesion clusters : 60 isolated lesions, whereas 4 MeV D

μm) are

predicted to produce DNA lesions with a ratio of 1 five-lesion cluster : 4 two-lesion clusters : 8 isolated lesions [25]. Monte Carlo track structure simulations of ionizing particles, also indicated that about 30% of DSB are complex DSB with at least one lesion (including base damage) in close proximity at low LET, whereas this is 90% at high LET [25]. It was also predicted 3-4 times more non-DSB clusters than DSB at low LET, and up to 8 times more at high LET [25]. This was confirmed by experimental determinations for low LET radiation [8, 12]. However, Hada & Sutherland [26] reported that the relative frequencies of DSBs compared to bistranded abasic and oxidized base clusters were higher for the charged particles than for X-rays, in (linear) T7 phage DNA. Extended and refined Monte Carlo track structure simulations recently calculated that complex SSB associated with base damage represent 49% of all SSB for low LET radiation, and 87% for the high LET radiation tested. In addition, calculations showed that nearly all DSB induced by high LET radiation are complex, e.g. 97% versus 84% for low LET radiation [27]. Noteworthy, with increasing LET, the cluster complexity (average number of lesions per cluster) increases, while the overall number of DSB and non-DSB clustered lesions does not increase, and even decreases, as well as single isolated DNA lesions [6, 26, 28-31]. Altogether, this demonstrates that quantitatively, clustered DNA lesions induced by ionizing radiation are not anecdotal. Clustered DNA lesions, in fact non-DSB clustered lesions, were first detected in irradiated plasmid DNA [7, 32], and soon after in mammalian cells [8-10, 12]. The methodology makes use of purified BER proteins as enzymatic probes that convert clustered DNA lesions, carrying opposed SSB, base damage or AP sites, into DSB which are then quantified using constant or pulsed field gel electrophoresis [33, 34]. Single-molecule Laser fluorescence sizing, based on microfluidic technology was also

proposed as being as or more sensitive than gel electrophoresis [35]. Radiation-induced complex DSB, which have associated base damage and abasic sites in proximity to the break termini, were also revealed by the use of BER proteins as enzymatic probes [36, 37]. Nevertheless, the efficiency of the BER proteins depends on the inter-lesions distance (should be a few bp apart) and on the nature of the lesions, thus, some clustered DNA lesions may be refractory to cleavage. Although it is not fully appropriate and that it does not reveal the complexity of the clustered DNA lesions, such methodology has been successful in estimating the yield and the repair rates of radio-induced bistranded clustered DNA lesions in mammalian cells, human fibroblast cells, human cancer cells and irradiated mouse skin tissue [11, 38, 39]. It remains the most reliable method at the moment. To overcome the lack of detection of very closely opposed AP sites (< 5bp apart) by repair proteins, putrescine has been successfully employed and reveal 1.7 to 2 times more AP clusters than revealed by bacterial Nfo protein [11]. A novel method which makes use of fluorescent markers that react with carbonyls at AP sites (and 5-formyluracil) wherever they are, followed by fluorescence resonance energy transfer (FRET) analysis, allowed to reveal the AP clusters induced in solid state DNA by J-rays and carbon particles [40, 41]. Hopefully, it should possibly be applied to the detection of such clustered DNA lesions in mammalian cells. To date, the best estimate for clustered DNA lesions induced in human cells by X-rays gives the ratio of 1 DSB : 1.2 abasic clusters : 1 oxidized purine (oxyPur) cluster : 0.9 oxidized pyrimidine (oxyPyr) cluster, with AP clusters comprising 30% of the complex damage and DSB about 25 % [10, 11]. Plasmid assay in conjunction with enzymatic treatment continues to be employed to better characterize DNA damage and estimate the biological effects of ionizing radiation of various energy, including charged particles used for radiotherapy [30, 42-48]. Some of these studies point towards a major role of clustered DNA lesions in plasmid inactivation, compared to DSB. Interestingly, low energy electrons of energies less than 20 eV have been shown to induce SSB and DSB in plasmid DNA, are now demonstrated to produce non-DSB clustered lesions, while infrequently, and to lead to loss of transformation efficiency [49]. The authors suspect that the type of clustered DNA lesions, created directly on DNA by 10eV electrons, may be more difficult to repair than those produced by other species from radiolysis. The last twenty years, immunofluorescent staining and microscopy or direct live cell imaging methodologies have used DNA repair and response proteins to localize radiation-induced DSB and explore their repair. Phosphorylation of histone H2AX, resulting in JH2AX, occurs relatively rapidly at DSB sites and propagates over

megabases through chromatin on both sides. Indeed, JH2AX appears as distinct ionizing radiation-induced foci (IRIF) around the DSB. Attempts have been made to correlate the number of JH2AX foci to the number of induced DSB [50]. Phosphorylated H2AX allows the recruitment of other repair and signaling proteins which also form foci at DSB sites. For example, JH2AX, 53BP1, DNA-PK foci are used as DSB markers for damage localization and repair rate analysis (for review see [51]. Similarly, Xrcc1 and Ogg1 are recruited at sites of SSB and 8-oxo-7,8-dihydroguanine (8-oxoG) induced by laser irradiation [52] and reveal iron particles tracks in cells within minutes (< 10 min) after irradiation [51]. Moreover, Chen and coworkers [53] observed co-localized 53BP1, Xrcc1, hOgg1 foci in cells exposed to high LET charged particles, claiming a direct visualization of clustered DNA lesions at the single-cell level. In spite of the fact that sensitive image acquisition and precise determination of spatial distribution of clustered DNA lesions were performed, and the fact that cell exposed to H2O2 did not reveal overlapping 53BP1, Xrcc1, hOgg1 foci, whether or not the described method have resolved clustered DNA lesions is questionable. In contrast, sensitive and quantitative transmission electron microscopy method allowed Lorat et al. to detect for the first time at a nanoscale level clustered DNA lesions induced in cells by high LET radiation [54]. Evidently, imaging is not appropriate for precise quantitation of clustered DNA lesions, but is of utmost importance to better understand the spatio-temporal characterization of the cellular response to radiation. Could clustered DNA lesions be formed by other agents, in particular by oxidizing agents, such as hydrogen peroxide and UV radiation? There were some early reports that UVC, UVB, UVA, UVA plus metal or dyes and H2O2 plus metal induce DSB in plasmid DNA or cells (reviewed in [1]). Knowing that UVA-induced DNA damage are not produced via a photosensitizer which would sit in a close proximity with DNA, the formation of clusters of radical oxygen species next to DNA is extremely unlikely [55]. In fact the main reactive oxygen species induced by UVA radiation in mammalian cells is singlet oxygen through type II photosensitization [56], that specifically induces 8-oxoG at the exclusion of DNA nicks [57, 58]. The recent report, in which the authors used indirect methods such as neutral comet assay and stretched chromatin fibers to reveal oxidatively generated clustered DNA lesions by UVA [59] is not convincing and is opportunistic. A solid piece of work showed that low levels of H2O2 cause significant elevation of replication-independent DSB resulting from oxidatively generated bistranded clustered DNA lesions, and lead to NHEJ-mediated mutations [60]. Oxidatively generated clustered DNA lesions were clearly observed while their measurement may be taken with caution, due to the mode of quantitation. Ravanat and coworkers [61] isolated and identified a complex DNA lesion resulting from a single

radical hit, in cells treated with the radiomimetic bleomycin or with ionizing radiation. It comprises 2’-deoxycytidine adducts involved in an interstrand crosslink, and a single strand break. Its yield of formation is similar to that of DSB and is about 1% of that of 8-oxoG after irradiation. It was also detected at low rate in untreated cells, emphasizing that it may also be formed by endogenous oxidative stress. In fact, Georgakilas and coworkers were also able to detect endogenous oxidatively generated bistranded clustered DNA lesions in various human cell lines, but using a modified method to that originally used [15]. The indirect methods (specially the process of cells) that have been developed so far for measuring oxidatively generated clustered DNA lesions likely provide overestimated values of lesions. It is clear that further work is needed to develop accurate assays for measuring oxidatively generated complex DNA lesions including clustered DNA lesions (a highly challenging issue) and one radical hit-mediated tandem lesions [62]. Defining the structural determinants of some non-DSB clustered lesions should bring clues in their recognition by BER proteins and help understanding their reparability. NMR experiments, molecular dynamics (MD) simulations studies have been performed on short (11 to 40 bp) double stranded oligonucleotides carrying either 2 bistranded AP sites or 1 AP site and 1 opposed 8-oxoG), separated by 0, 1, 3 or 5 bp, in the two orientations [63-67]. The studies demonstrate that none of these bistranded clusters disturb drastically the B-form structure of DNA double helix which experiences local rearrangements and restore a maximum of non-covalent interactions. The lesions site may be more or less bent and the thermal stability smaller by few degrees. Indeed, the interaction of 8-oxoG with bases on the complementary strand is weakened. In contrast, depending on the sequence, the orientation and the inter-lesion distance, AP site may be extruded from the double helix towards the minor or the major groove. The extrusion of the AP site towards the major groove is favorable for an efficient repair by the AP endonuclease, e.g. hAPE1. Altogether, these studies corroborate very well the biochemical data obtained with bacterial, yeast and human BER proteins, as well as mammalian cells extracts [68]. From these studies, it appears clearly that the structural characteristics of single lesions cannot be applied to clustered DNA lesions, likely due to the flexibility and the dynamical aspect of the DNA helix. The differences observed between the analyzed clustered DNA lesions largely depend on the lesions (8-oxoG vs AP site), the relative orientation and the distance between the lesions in the cluster. The extrahelical behavior of a lesion within a cluster is an important feature for the reparability of a cluster and for its outcome. The nature of the lesions within a cluster may also be important. The structural aspects of DNA at more complex clustered DNA lesions could be advantageously studied by molecular dynamics. However, the kinetics

characteristics of the BER enzyme involved are also important as discussed previously [69, 70].

Repair of complex DSB and non-DSB clustered lesions in irradiated cells A good biological approach to get insight into the repair capacity and mechanism of clustered DNA lesions, including complex DSB and non-DSB clustered lesions, is to analyze DNA repair in cells exposed to high LET radiation which induces essentially complex DNA damage (see [25, 27]). Repair of complex DSB by pulsed field gel electrophoresis analysis It is generally considered that DSBs induced by high-LET heavy ions are rejoined less efficiently than those by low-LET radiations, leaving higher rate of non-rejoined DNA breaks after exposing to high-LET radiations (for review see [71]. In the early nineties, the complexity and the severity of clustered DNA lesions was inferred from the low repair rate of DSB produced by D-particles, in comparison with J-rays, in rodent cells or by cell extracts in irradiated plasmid DNA (in cells, within 3hrs, 30-50% rejoining vs >90%, respectively) [72-74]. In these early studies, only DSB and rejoining were examined. However, unrepaired or slowly-repaired DSB could be either complex DSB directly produced by the particles or DSB produced during the processing of non-DSB clustered lesions. Since, analyzing the repair rate of DSB produced by radiation of various energies and LET has brought some clues on the complexity of the DNA damage formed and on their role in the radiation biological effects. For example, a significant increase in DSB induction by ultrasoft X-rays at 350 eV, above the carbon-K shell ionization threshold, compared to 250 eV, below this threshold. Moreover, DSB produced by 350 eV are less repaired within 2 hrs post-irradiation [75]. Indeed a role for inner shells ionizations in DNA constituent atoms (including C, N and O atoms) has been proposed for cellular inactivation upon heavy ion- and J-ray- irradiation [76]. Using an in vitro non-homologous end joining assay, Pastwa et al. [77] found that end joining of J-ray or

125I

decay-induced DSB was strongly inhibited compared to that of

restriction enzyme-induced DSB. Furthermore, Datta et al. demonstrated that DSB end structures generated by radiation (i.e., 3’-phosphates and 3’-phosphoglycolates, and base damage and AP sites in proximity to the DSB ends) retard the joining process and that the damage that occurs proximal to the DSB ends is a greater inhibitor of end joining than 3’-phosphates or 3’-phospoglycolates [78]. Heat or alkali labile sites are non-DSB DNA damage known to be produced by oxidizing agents and ionizing radiation, and may be converted into DSB. Using controlled temperature-protocols for cell lysis

and electrophoresis and a heat treatment to reveal such lesions, a 30-40% increase in DSB has been reported in cells exposed to high LET radiation. Interestingly, the heat-released DSB are extremely rapidly repaired (< 1-2 hrs) post-irradiation for both low and high LET radiation [79, 80]. Repair of complex DSB by imaging analysis The last twenty years, the DNA repair field, in particular DSB repair, has largely gained from the constant developments in the domain of fluorescent proteins engineering and high resolution microscopy. Immunofluorescent staining and microscopy or direct live cell imaging methodologies are used to examine the recruitment and retention of DNA repair and response proteins at DSB sites induced by ionizing radiation of low and high LET, enabling a spatiotemporal analysis of DNA repair process at single-cell level. Various aspects of the advances in the field have been reviewed in detail [13, 51, 81, 82]. Briefly, radiation-induced DSB and subsequent chromatin alterations serve as signal for ATM protein kinase to induce the phosphorylation of histone variant H2AX at DSB sites and expand it nearby, as a result the signal is amplified and promotes the recruitment and accumulation of chromatin-binding or modifying factors and repair proteins (MDC1, RNF8, 53BP1, RAD51, Mre11/RAD50/Nbs1 complex, PARP-1, DNA-PK…), thus forming discrete radiation-induced foci or IRIF. Therefore, using immunofluorescence microscopy, DSB can be rapidly visualized by JH2AX foci, 53BP1 or DNA-PK foci, and their repair rate analyzed. However, what is underneath foci and how their decrease with time reflects DNA repair rate is uncertain. Indeed, an increase in the size of IRIF with time after irradiation has often been observed and could be due to aggregation of smaller foci. The spatio-temporal formation and dynamics of foci, related to the cell fate, have been well reviewed by Belyaev and Costes et al [83, 84]. Yet, recent evidence suggests that chromatin organization mediates the response to DNA damage. For instance, densely compacted heterochromatin is relatively refractory to JH2AX foci formation and DSB repair, while repair protein are fastly recruited and H2AX rapidly phosphorylated at DSB sites, but the damaged sites are then relocated at the periphery of heterochromatin at the border with euchromatin [83, 85-87]. The resolution of foci in heterochromatin is slower than that in euchromatin and ATM signalling is specifically required for DSB repair within heterochromatin, indicating that the < 25% of DSB that require ATM signalling for repair may correspond to heterochromatic DSB rather than to complex DSB [9, 80]. Furthermore, Neumaier et al. [88] have provided evidence for the formation of DNA repair centers after irradiation with X-rays or Fe ions, emphasizing that slower DSB repair kinetics may not only be related to complex damage, but also to the presence of multiple breaks within a repair center. The proximity of multiple DSB in repair center or clustered foci may favor misjoining of the

ends during repair, resulting in chromosomal aberrations [85, 88, 89]. Moreover, using imaging with deconvolution, Nakajima et al [90] have also observed a clustering of 10 or more individual foci into large JH2AX foci formed within the radiation track, 8 hours after exposure to Fe particles. Those foci were slowly repaired and caused a prolonged checkpoint arrest, as compared to X-rays. Notably, mitotic entry of cells containing as many as 10 clustered foci has been observed in this study. Using high resolution deconvolution imaging of Si and Fe ion-irradiated human cells, Chen & col. [53] have observed 53BP1, Xrcc1 and Ogg1 foci which co-localize, emphasizing that those overlapping foci are uniquely formed at clustered DNA lesions. A few large, overlapping foci persisted 24 hours post-irradiation, in particular for Fe ion exposure, showing that clustered DNA lesions can be refractory to repair. Importantly, unrepaired clustered DNA lesions resulted in high level of gross-chromosomal aberrations in Fe ion-irradiated cells, due to checkpoint release before the complete repair of these lesions. The most comprehensive and quantitative analysis of DSB and clustered DNA lesions after low and high LET irradiation has recently been reported using immunogold-labeling of DNA repair factors and electron microscopy [54]. DSB were numbered and localized as pKu70 dimers beads, either isolated for isolated DSB or clustered at the nanoscale level, at various time post-irradiation, along with 53BP1 beads. Low LET irradiation produced predominantly single Ku70 dimers randomly distributed through the nucleus. High LET irradiation lead to clustered DSB and the size of the clusters depended on the chromatin packing density. The clustering of DSB was most frequent at 5hrs post-irradiation, and a large proportion of the clustered DSB remained unrepaired 24 hours after irradiation, those containing up to 30 closely spaced pKu70 beads surrounded by many 53BP1 beads and located exclusively in the highly compacted heterochromatin. Altogether, these in situ analysis have provided coherent data and have demonstrated that the complex DSB induced by high LET radiation (formed directly or by the processing of non-DSB clustered lesions) are far less repaired than those induced by radiation of smaller LET, in particular those located in heterochromatin. They accumulate in cells, most of which may die due to their inability to complete mitosis, though some other survive with genome instability, e.g. mutations and chromosomal aberrations. Pathways and proteins involved in the repair of complex DSB In mammalian cells DSB are repaired via two major pathways, non-homologous end joining (NHEJ) and homologous recombination (HR). NHEJ which does not require undamaged DNA template, operates primarily in G1 and early S phase of the cell cycle,

while HR pathway utilizes the sister chromatid as template, and consequently, it operates in S phase, likely in an error-free manner. DSB that are present in late S or G2 phase can be repaired by either pathway. The use of real-time imaging has revealed that simple DSBs, induced by Ultra Soft X-rays or Near Infra-Red microbeam irradiation, are repaired rapidly involving NHEJ proteins, Ku70/80 and XRCC4/Ligase IV/XLF. In contrast, complex DSB are repaired slowly, involving the same proteins but also DNA-PKcs, and the absence of functional ATM retards the repair of those complex DSB [91]. Although NHEJ appears to be the predominant repair pathway for DSB induced by low LET radiation, it does not seem to be the case for the more complex lesions induced by high LET radiation [92, 93] reviewed in [51, 82]. Evidence indicates that the NHEJ-mediated repair of DSB induced by high LET radiation is delayed, in comparison with low LET radiation, due to the complexity and damage clustering. In addition, cells deficient in NHEJ (Ku80-/-, DNA-PKcs-/-) are less sensitive to high LET radiation than to low LET radiation. One likely explanation is that in NHEJ-proficient cells Ku70/Ku80 or DNA-PK are inhibited, trapped by the small DNA fragments (about 40 bp) generated by high LET radiation [94], while PARP-dependent alternative NHEJ or Mre11-dependent HR are not impacted. As a result, the repair of the complex DNA damage induced by high LET radiation relies more on homology-directed repair pathways and in particular HR [82]. However, DNA-PK foci can be formed and co-localize with JH2AX and 53BP1 foci in iron tracks [51]. Artemis, a protein essential for (V(D)J) recombination in humans, has been shown to play a role in the repair of a subset (10-20%) of low LET radiation-induced DSB, particularly those located within heterochromatin and that require ATM signaling. In fact, Artemis mutated cell lines are sensitive to low and high LET radiation and exhibit a significantly higher defect in the repair of DSB induced by high LET than by low LET radiation [51, 82] and references therein). In addition, ATM and Artemis promote homologous recombination in G2 phase of irradiated cells that requires Artemis endonuclease function. Interestingly, ATM and Artemis may promote the repair of radiation-induced DSB by NHEJ during G1 and by HR during G2, and the role of Artemis may be to remove DNA lesions or secondary structure, enabling completion of end resection and then repair by either pathway [95]. More recently, it has been reported that end resection is essential for repair of complex DSB in all phase of the cell cycle, as seen by the replication protein A (RPA), ATR, Mre11, Exo1 foci in S/G2 and G1 cells exposed to a series of low and high LET radiation [96]. Notably the proportion of the resection-positive cells correlates with the complexity of the induced damage. In conclusion, DNA resection is required to resolve complex DSB and clustered lesions such as those induced by high LET radiation. In S and G2 phases resection is probably used for HR, whereas in G1 phase it is utilized for error-prone, alternative or microhomology-mediated end joining. As a result, HR plays a more important role in

complex DSB repair by high LET than by low LET radiation and Artemis may thus have a central role. Repair of non-DSB clustered lesions by static or pulsed field gel electrophoresis analysis In the 2000’s, the repair rate of specific non-DSB bistranded clustered lesions was examined by Sutherland, Georgakilas and co-workers. It was clearly shown that cells deal poorly with closely spaced abasic clusters, as revealed by Nfo- or putrescine-detected abasic clusters in human cells exposed to J-rays. Indeed, while DSB rejoined efficiently within 24 hrs, abasic clusters decreased very slowly over 14 days, indicating that abasic clusters may become persistent DNA damage [14]. After human cells were exposed to

56Fe

ions, DSB decreased to background levels within 1-2 days,

whereas oxidatively generated clustered DNA lesions required about 4-5 days [29]. Noteworthy, the repair of oxidatively generated clustered DNA lesions is less delayed than for low LET radiation (7-14 days), in particular for the Nfo detected clusters [14], probably related to the increase in apoptosis observed 4-5 days after ions exposure. Moreover, it was reported that mice tissues exposed to X-rays accumulated clustered DNA lesions up to 20 weeks after irradiation [38]. To get insight into the biological impact of heavy ions such as those employed in hadrontherapy, yields and repair process of oxidatively generated clustered DNA lesions were examined in hamster or MEF cells, mutated or not for BER proteins, irradiated by a series of heavy ion beams of increasing LET and with J-rays [31]. It was shown that the yields of DSB, Fpg and Nth clusters, and isolated lesions as well, decreased with increasing LET due to the characteristics of the radiation tracks. Furthermore, DSB increase 1-2 hours after irradiation, but much more for carbon ions than for J-rays, reflecting the higher proportion of non-DSB clustered lesions for high LET radiation. Interestingly, more DSB accumulate in cells deficient for Xrcc1 (a scaffolding BER protein that facilitates SSB repair), and these mutant cells are more sensitive to carbon ions that to J-rays. In addition, cells deficient in Ogg1 or Nth1 are resistant to irradiation, as previously shown [20] and are more resistant to carbon ions than to J-rays in comparison with normal cells. These results show that attempted repair at oxidatively generated bistranded clustered DNA lesions results in the formation of unrepairable DSB that lead to cell killing. Indeed, in the Ogg1 and Nth1 mutant cells, DSB cannot be formed, whereas they accumulate in the Xrcc1 mutant cells. This study is a clear demonstration of the biological relevance of clustered DNA lesions. Other data suggests that DNA-PKcs and BRCA1, involved in NHEJ and HR, respectively, are implicated in the repair of oxidatively generated bistranded non-DSB clustered lesions in human tumor cells since such damage are repaired a bit slower in the deficient cells [97, 98].

Engineered non-DSB clustered lesions: processing and mutation induction Radiation-induced non-DSB clustered lesions comprise SSB, abasic sites, oxidized purines and pyrimidines, and each individual lesion is mainly repaired by base excision repair. The processing of non-DSB clustered lesions and its outcome cannot be assessed by direct irradiation of cells. It cannot be predicted from the repair kinetics of single lesions, either. The last twenty years, the strategy of introducing two or more chosen DNA lesions, site specifically located and closely opposed, in oligonucleotide duplexes to be inserted into DNA vectors for cell transformation, tremendously helped in deciphering the processing of a variety of clustered DNA lesions and their biological outcome in bacteria, yeast and human cells. The biochemical studies using purified DNA repair enzymes or nuclear and whole cell extracts have been recently reviewed in great detail by Eccles et al [17]. Data show that the repair of the lesions within clustered DNA lesions is compromised in comparison with the same isolated lesions, and the repair retardation depends on the nature, the location and the orientation of the lesions relative to each other, and on the inter-lesion distance. Long-lived repair intermediates can also be generated, such as DSB. Oligonucleotides carrying various combinations of lesions, e.g. 2 AP sites or uracils, 8-oxoG plus AP sites, 8-oxoG plus SSB or gap, 8-oxoG plus 5,6-dihydrothymine (DHT), 5,6-dihydroxy-5,6-dihydrothymine, the so-called thymine glycol (Tg) or 5-hydroxyuracil (5-ohU)…., of increasing complexity (two opposed lesions to up to 5), have been built in plasmid DNA for extensive in cellulo studies. The relative transformation efficiency of the clustered DNA lesions-carrying plasmids in bacteria or yeast (plasmid survival) provides an estimate of the extent of the DSB formation during the repair process. The extent of DSB formation can also be estimated in mammalian cells after co-transfection with two different vectors, one damaged and one undamaged [99-101]. Analyzing the mutations produced at the damaged sites after repair and replication in wild type bacterial or yeast cells and in cells mutated in DNA repair genes, allows to model the processing of the studied non-DSB clustered lesions. In mammalian cells, the extent of repair has been directly assessed by targeting the constructed clusters into a reporter gene, e.g. luciferase gene [99-101]. In our laboratory, we have designed a beta-galactosidase-based white/blue colony assay to roughly estimate DSB formation, and mutations are analyzed by sequencing of the repaired and replicated plasmid (Sage, and Sedletska, unpublished). Most published data obtained with bacteria have been extensively reviewed in [17]. Based on these accumulated data on the repair of a variety of non-DSB clustered lesions, we know that the type of lesions comprised in the clusters largely influence the repair

efficiency of the lesions, the manner the clusters are processed and finally the outcome of the clusters and their respective biological consequences. Bistranded AP sites and bistranded uracils without or with a vicinal 8-oxoG and irrespective of its position, are rapidly and near-simultaneously incised, result in rapid DSB formation as abortive repair and can no longer be repaired by BER, leading to cytotoxicity (plasmid loss). In the rare cases of low incision at the clusters, the surviving plasmids carry predominantly small deletions (1-4 bp to 40-50 bp) at uracil positions ([17] and references therein). In yeast, bistranded uracil residues or AP sites separated by 6 bp are also readily converted into DSB and in cells deficient in AP endonucleases (apn1 and apn2 cells), unrepaired AP sites in these clusters are toxic (plasmid loss) [102]. A recent report shows that DSB are also readily formed at bistranded GAP and AP site [103]. Whether or not non-DSB clusters containing closely opposed base lesions lead to DSB formation depends in fact on the type and the distribution of the damaged bases on the two strands. The question is which DNA N-glycosylase will bind and cleave first at twoor three-lesion clusters carrying for example 8-oxoG, oxidized/reduced pyrimidine and uracil/AP site, situated 5 – 10 bp apart on both strands. Does incision at such bistranded clusters lead to DSB formation? The early studies focused on clusters where 8-oxoG is opposed to oxidized/reduced bases or uracil residue, and it was concluded that non-DSB clusters containing closely opposed base lesions do not lead to DSB formation, but rather to increased mutation induction at the sites of the lesions which are still present at replication [17]. Indeed, non-DSB clusters containing 8-oxoG opposing another 8-oxoG, DHT, Tg, 5-ohU or uracil are not converted into DSB, either in E coli or yeast, even when separated by 9 bp like for 8-oxoG and 5-ohU in the clusters studied in yeast [17, 102, 104, 105]. In all these cases, the mutation frequencies are highly increased and GC to TA transversion at 8-oxoG largely predominates, in both E coli and yeast ([17] and references therein; [18]; Sage and Kozmin, unpublished). Some differences in the mutation frequencies at the clusters are observed and depend on the relative orientation of the lesions and on inter-lesion distance. Interestingly, when 2-deoxyribonolactone (dL), an oxidized AP site repaired by long patch BER, is placed in close proximity to 8-oxoG on the opposing strand, the mutation frequency is also enhanced, GC to TA transversions also predominates, but additional single deletions and base substitutions affecting both dL and 8-oxoG containing strands are observed in the vicinity [106]. Collectively, these observations demonstrate that excision/incision at 8-oxoG is retarded in non-DSB clusters and plasmid replication occurs before 8-oxoG removal. However, it has recently been reported that up to 80% of the non-DSB clusters composed of uracil and 8-oxoG in one strand and 5-ohU in the opposite strand (U-8-oxoG/5-ohU) are converted into DSB in E coli (plasmid loss) and that nearly all

surviving clones carry mutation at 5-ohU [107]. In addition, it was demonstrated that the extremely high mutation frequency at 5-ohU (84-97%) in these clones is subjected to the initial excision of opposed uracil by Ung protein, followed by incision of the phosphodiester backbone and strand loss at replication due to replication fork collapse. This does not occur in ung cells (that are unable to excise uracil), in which mutation frequency largely decreases (to 8-14 %) and some mutations at 8-oxoG are observed. In contrast to what observed with the other non-DSB clusters, the mutation frequency at the tandem AP-8-oxoG in fpg and mutY cells is lower than that at single 8-oxoG. Interestingly, at tandem damaged sites carrying 8-oxoG and AP site 1 to 5 bp apart, the incision at AP site is rapid whereas the ligation is inefficient due to the presence of 8-oxoG, so that the repair gap is still open at replication, leading to loss of the 8-oxoG containing strand [108]. The authors concluded that an AP site can protect from the mutagenic potential of 8-oxoG when present in a tandem clustered site, at least in E. coli. Addition of 8-oxoG opposite tandem AP-8-oxoG (three-lesion clusters) retards the in vitro repair of the AP site and increases the mutations at 8-oxoG [105]. To further explore the role of a repair gap when formed at non-DSB clusters, Shikazono and col. [104] built a series of bistranded GAP/8-oxoG clusters, tandem 8-oxoG-GAP clusters and three-lesion clusters with two opposed 8-oxoG and a GAP. Although carrying a pre-existing SSB, none of the clusters form deleterious DSB in E. coli. Analysis of mutation frequency in fpg mutY cells and wt, fpg and mutY cells demonstrated that a GAP on the opposing strand enhances the mutagenic potential of 8-oxoG, while a gap on the same strand does not, due to loss of the damaged strand. Adding an 8-oxoG to the bistranded GAP/8-oxoG (three-lesion clusters) does not increase the mutation frequency, and mutations originate from 8-oxoG located opposite the tandem 8-oxoG-GAP. Data suggest that lesions may be sequentially processed in the three-lesion cluster, with excision of 8-oxoG when located at more than 3 bp from the GAP, resulting in an enlargement of the gap. In all cases the retardation of the closing of the gap leads to strand loss. Noteworthy, the mutation frequency of bistranded 8-oxoG-GAP is very similar to that of bistranded 8-oxoG. In yeast, the presence of a 1nt gap in four- or five-lesion

clusters

containing

8-oxoG,

5-ohU,

5-formyluracil

(5-foU)

and

8-oxo-7,8-dihydroadenine (8-oxoA) did not favor the formation of DSB either [102], but prevented the excision of 8-oxoG located at 3 bp on the opposite strand [69, 109]. As a result, the mutation frequency for these complex non-DSB clusters is highly increased (4 to 7 times) in comparison with single 8-oxoG or 5-ohU [18]. Mutations at 8-oxoG predominate

in

the

bistranded

8-oxoG/5-ohU

and

in

the

clusters

8-oxoG-8-oxoA/5-ohU-GAP-5-foU, though mutations at 5-ohU and 5-foU are also recovered (Sage and Kozmin, unpublished).

The involvement of DNA polymerases in the processing of non-DSB clustered lesions has also been explored. In particular, the role of the DNA polymerase I (Pol I), the most abundant DNA polymerase, responsible for repair synthesis during BER of base lesions and AP sites in E. coli, has been investigated using bi-stranded GAP and AP site or 8-oxoG clusters [103]. In the presence of Fpg protein, the deficiency in Pol I increases the mutation frequency roughly by a factor 2 and 7 when GAP and 8-oxoG are separated by 1 and 10 bp, respectively, independently on the location of 8-oxoG in the leading or lagging strand. Mutations are mainly GC to TG transversions at 8-oxoG. Pol I is thus involved in reducing mutagenesis at 8-oxoG/GAP clusters but also in limiting the formation of DSB at AP/GAP clusters, particularly when the lesions are more than 10 bp apart. Based on their data using purified human DNA polymerases, Lavrik and col.[110] propose that after incision of the AP site at bistranded 5-foU and AP site clusters, Pol E

O perform repair synthesis across 5-foU by a

short or a long patch BER, depending on the relative location and orientation of the lesions. The translesional human DNA polymerase iota is also able to bypass 5-foU at these clusters [111]. The results from all the above studies in bacteria and yeast converge towards a hierarchy in the lesion processing, that determines repair, DSB formation and mutability of the different non-DSB clustered lesions. Within a non-DSB clustered lesions, uracil and AP sites are the first lesions to be cleaved, then come oxidized pyrimidines; a BER-resulting SSB does not prevent incision at oxidized pyrimidine, while incision at 8-oxoG is inhibited [69, 70, 102, 105, 107, 109, 112]. In spite of the hierarchy in the incision rates at lesions within non-DSB clusters containing oxidized bases, full repair of the first incised lesion does not occur before the second incision, except in the case of 8-oxoG whose incision is always retarded or inhibited. The consequence of retarded or impaired repair is an extension of the lifetime of the lesion in the clusters and mutations occur during replication. The mutation frequency depends on the relative orientation and spacing of the lesions within clusters and increase with increasing complexity of the non-DSB clusters [18, 104, 105, 107, 112]. In mammalian cells, a similar non-DSB clusters-containing plasmid approach has been undertaken, however, the experiments are a true challenge and the results are not as clear-cut as for E. coli and yeast. Yet, isogenic DNA repair deficient cells are not easily available, though siRNA can be used to partially inactivate DNA repair genes. To analyze mutations, plasmids DNA carrying the non-DSB clustered lesions of interest, once processed and replicated in mammalian cells, have to be extracted from a population of cells and used to transform E. coli for further analysis (sequencing of the targeted region

of the plasmid). In contrast, for bacteria and yeast, each colony originates from a single cell transformed with a single plasmid. Another major difference is that a processed plasmid harboring a DSB as repair intermediate is lost in bacteria and yeast, whereas it can be repaired by NHEJ in mammalian cells. Therefore, the deletions that can be observed are interpreted as originating from DSB repaired by error-prone NHEJ. Malyarchuk et al. [99] have inserted two opposing furans, surrogates of AP sites, located 2, 5 and 12 bp apart, in a firefly luciferase reporter plasmid, and transfected mouse wild type, Ku-/- or DNA-PKcs-/- cells. The decrease in luciferase activity and plasmid survival, and the observation of a high deletion (< 1 bp) frequency demontrate that the bistranded lesions are converted into DSB by APE1 (APEX1 protein, the major AP endonuclease in mammalian cells), which are repaired by Ku-dependent and Ku-independent pathways. The maximal effect is observed at 5 bp apart. Next, the authors have examined the processing of two-, three-, four-lesions clusters containing furans plus one 8-oxoG next or opposite to a furan residue, using the same assays [100]. In mammalian cells, like in bacteria and yeast, a clustered DNA lesion consisting of an 8-oxoG and an opposing furan does not lead to DSB formation, while two opposed furans can be converted to DSB, even in the presence of a base damage (8-oxoG) or other furans in the cluster. A third furan in the cluster slightly reduces the DSB formation but does not abolish it and in vitro data show that 8-oxoG situated 5’ in tandem with a furan can also modestly reduce the cleavage at this furan. The in vitro studies did not reveal any 8-oxoG removal from the tested clusters. In fact, the data suggest that these specific three- and four-lesions clusters can be converted into complex DSB, that can be either degraded or inaccurately repaired and result in deletions or insertions. Artemis is an endonuclease that removes blocking residues at DSB ends during repair by NHEJ. The role of Artemis has thus been investigated using the same two- or three-lesions clusters containing furans and 8-oxoG [101]. It is shown that loss of Artemis does not decrease plasmid survival, but leads to a more mutagenic repair, involving larger deletions, and the repair is independent of Mre11. This study suggests that Artemis is not implicated in the repair of complex DSB per se, but rather in the processing of DSB with blocked termini. Artemis has been suspected to play a role in the processing of clustered DNA lesions and complex DSB [51] and this study pinpoints it. We have analyzed the processing of two- or three-lesions clusters containing uracil, 5-ohU and 8-oxoG, in human cells (similar non-DSB clustered lesions as in [107]). Our data (Sage and Sedletska, unpublished) indicates that DSB can be formed and repaired in an inaccurate manner by NHEJ. The mutation frequency increases with the complexity of the clusters. Mutations comprise base substitutions at damaged bases, small and large deletions (1-2 bp to 200 bp), and the ratio of the mutation types depends on the spacing between lesions. Bistranded uracil cluster leads to higher level of DSB formation and deletions

than the three-lesion clusters consisting of U-8-oxoG/5-ohU, while bistranded 8-oxoG/5-ohU cluster results mainly in point mutation and very low DSB formation. Although more investigations are needed in mammalian cells, it appears that most of the processing models proposed for the variety of non-DSB clustered lesions studied to date in E. coli or yeast, also apply to mammalian cells, with subtle differences. This implies that a hierarchy in repair and DSB formation also occurs at non-DSB clustered lesions in mammalian cells. The work with mammalian cells pinpoints the deletions that are formed at non-DSB bistranded clusters carrying one or more U/AP sites, probably originating from DSB formation and repair by error-prone NHEJ, sometimes with microhomologies at the junctions [99, 100]. The nucleosomal organization of the chromatin can protect DNA from damaging but also generate a steric impediment to DNA repair enzymes. Indeed, chromatin condensation likely helps to exclude water molecules, limiting the number of ROS created by radiation that damage DNA, whereas chromatin decondensation sensitizes DNA to radiation [113, 114]. The condensation level may also impact the repair efficiency, as described above for DSB repair in human cells. Moreover, it has been hypothesized that chromatin protects DNA from the formation of DSB during BER of non-DSB clustered lesions. The in vitro repair of non-DSB clusters, e.g. bistranded Tg/Tg, AP/AP and AP/8-oxoG, 8-oxoG/8-oxoG, by BER N-glycosylases or by mammalian nuclear extract has been investigated in nucleosomal environment. The cleavage at single base damage or AP sites by purified BER proteins is usually less efficient in nucleosome than in naked DNA. As expected, the efficiency of cleavage at Tg, 8-oxoG, AP by hNTH1, hOGG1, and APE1, respectively, is markedly reduced at the bistranded lesions in nucleosomes, as compared to naked DNA [115-117]. In comparison with APE1, nuclear extract is relatively more efficient at cleaving at AP sites clusters in nucleosomal template; it is fully proficient at incising DNA at AP site in AP/8-oxoG, while excision of 8-oxoG is inhibited [117]. Overall, DSB formation at bistranded clusters is substantially reduced in nucleosomal environment. Meanwhile, the extent of this reduction depends on the enzyme tested, the local sequence context, the orientation of the damaged base regarding the histone octamers and the distance between the opposing lesions. As suggested by Cannan et al. [115], in cells, BER-generated DSB should occur preferentially in linker DNA and in regions associated with high rate of nucleosome turnover or remodeling.

Concluding remarks : biological relevance of clustered DNA lesions Quantitatively, clustered DNA lesions, including non-DSB clustered lesions and complex

DSB are not rare radiolesions, even at low LET. The in vitro studies using engineered clustered DNA lesions of increasing complexity described above allow to summarize the processing of non-DSB clustered lesions as shown in Figure 2. Indeed, most of the processes observed in E. coli or yeast have also been found in mammalian cells. From the comparison of DSB repair mechanism and kinetics in cells exposed to low or high LET radiation, the repair processes of complex DSB and their biological consequences can be estimated as shown in Figure 2. Notably, part of the complex DSB arise from the processing of non-DSB clustered lesions. Briefly, the observed hierarchy in the processing of the lesions within a non-DSB cluster leads to the formation of SSB, as repair intermediates, which retard the excision/incision at other lesions within the clusters. As a consequence of the reduced reparability of the clustered DNA lesions, the lifetime of the lesions within the clusters increases in comparison with isolated lesions. The lesions in the clusters still unrepaired at S-phase may encounter the replication machinery and, thus, mutation rate is increased. DSB can also arise very rapidly at damaged sites comprising opposing AP sites (or U) or AP/oxyPy. Evidence has been provided that in mammalian cells, such complex DSB are slowly and inaccurately repaired by alternative or microhomology-mediated end joining, leading to deletions of few to several hundred bp [99-101]. Altogether, the mutation spectrum of non-DSB clustered lesions comprises deletions, base substitutions, 1-2 bp deletion or insertion (indels) largely targeted at the lesions within the clusters. Multiple mutations within the clusters, targeted or not at lesions, may also occur at least in mammalian cells. In addition, the mutation frequency largely increases with the complexity of the clustered DNA lesions. While a majority of DSB induced by low LET radiation are repaired with fast kinetics, largely through canonical NHEJ, DSB induced by high LET radiation, mostly complex DSB, are resolved with slow kinetics and persist for longer periods. Despite some of the (complex) DSB are repaired, though inaccurately, during G1-phase, by NHEJ with the requirement of Artemis function, more of them seem to be repaired at late S-phase and G2-phase by homologous recombination [92, 93, 118]. Though unrepaired complex DSB may predominantly cause cell death due to the inability of the cells to complete mitosis, a substantial fraction outlives arrest from cell cycle checkpoint and pass through mitosis [53, 82, 90]. Unrepaired clustered DNA lesions result in an elevated level of chromatid and chromosomal aberrations, including chromatid and chromosome breaks, as seen in Fe ion-irradiated cells [53]. Noteworthy, the observed dynamics of the DSB repair foci, their relocation at less dense chromatin, their clustering at probable repair centers, may favor misrejoining of the ends, in particular in late S- and G2- phase. Indeed, translocations, sister chromatid exchanges and gross chromosomal aberrations are

drastically increased in surviving cell after exposure to high LET relative to low LET radiation and chromosomal damage is more severe and complex [53, 119, 120]. A high frequency of complex and often large deletion events (over Mbp size) and complex rearrangements at deletion junctions have been observed with high LET radiation and have not been reported for low LET radiation [121-123]. More recently, Fe ions at low fluence have been reported to induce large deletions and complex mutational patterns, despite the apparent lack of increased mutation frequency in the system used [124]. A genome-wide analysis of mutation induction by low LET radiation in the mammalian germline revealed interesting features [125]. Mutation induction at post-meiotic and pre-meiotic stages of spermatogenesis was analyzed by comparative genome hybridization and whole-genome sequencing, in the offspring of male mice irradiated with 3 Gy of X-rays. The induction rates of de novo CNV (copy number variant; primarily deletions, many of them >1,000 kb) and indels were significantly increased in the offspring issued of irradiated male. Though SNVs (single-nucleotide variants) were not significantly increased, the spectrum of mutation was markedly different, with an elevated frequency of clustered de novo mutations (clusters of 1-4 SNVs or 1-2 SNVs and indels within a few bp). Those clustered mutations and large deletions are regarded as originating from radio-induced clustered DNA lesions. This study is an excellent example of the outcomes of non-DSB clustered lesions as illustrated in figure 2. Isolated oxidatively generated base damage, abasic sites and prompt DSB are efficiently repaired with fast kinetics, which is in large contrast for non-DSB clusters and complex DSB. It may thus be concluded, based on the above discussion, that most of, if not all, mutations and chromosomal rearrangements induced by ionizing radiation, whatever the LET, result from unrepaired or mis-repaired non-DSB clustered lesions and complex DSB. These clustered DNA lesions are undoubtly the only damage with biological significance and consequences at low dose of radiation.

Acknowledgements Stanislav Kozmin and Yuliya Sedletska are thanked for their excellent work and Anton Granzhan, Institut Curie, CNRS UMR9187, for the graphical abstract. Funding: This work was supported by Electricité de France (RB2008-05; RB2011-27; RB2014-15), Agence Nationale de la Recherche ANR-09-PIRI-022, Centre National de la Recherche Scientifique, Institut Curie and short term JSPS (Japan Society for the

Promotion of Science) fellowship (FY2015, ID No. S15080) to ES, by Grants-in-Aid for Scientific Research (No. 22310038; 26550034; 16H02959) to NS, and by REIMEI joint research program from Japan Atomic Energy Agency in 2013 and 2014.

References [1] J.F. Ward, The complexity of DNA damage: relevance to biological consequences., Int J Radiat Biol., 66 (1994) 427-432. [2] R. Roots, S. Okada, Estimation of life times and diffusion distances of radicals involved in x-ray-induced DNA strand breaks of killing of mammalian cells, Radiat Res, 64 (1975) 306-320. [3] K.M. Prise, S. Davies, B.D. Michael, Cell killing and DNA damage in Chinese hamster V79 cells treated with hydrogen peroxide, Int J Radiat Biol, 55 (1989) 583-592. [4] J.O. Blaisdell, L. Harrison, S.S. Wallace, Base excision repair processing of radiation-induced clustered DNA lesions, Radiat Prot Dosimetry, 97 (2001) 25-31. [5] D.T. Goodhead, Initial events in the cellular effects of ionizing radiations: clustered damage in DNA., Int J Radiat Biol., 65 (1994) 7-17. [6] A.G. Georgakilas, P. O'Neill, R.D. Stewart, Induction and repair of clustered DNA lesions: what do we know so far?, Radiat Res, 180 (2013) 100-109. [7] K.M. Prise, C.H. Pullar, B.D. Michael, A study of endonuclease III-sensitive sites in irradiated DNA: detection of alpha-particle-induced oxidative damage, Carcinogenesis, 20 (1999) 905-909. [8] B.M. Sutherland, P.V. Bennett, O. Sidorkina, J. Laval, Clustered DNA damages induced in isolated DNA and in human cells by low doses of ionizing radiation., Proc Natl Acad Sci U S A., 97 (2000) 103-108. [9] B.M. Sutherland, P.V. Bennett, O. Sidorkina, J. Laval, Clustered damages and total lesions induced in DNA by ionizing radiation: oxidized bases and strand breaks, Biochemistry, 39 (2000) 8026-8031. [10] B.M. Sutherland, P.V. Bennett, J.C. Sutherland, J. Laval, Clustered DNA damages induced by X-rays in human cells., Radiat Res., 157 (2002) 611-616. [11] A.G. Georgakilas, P.V. Bennett, B.M. Sutherland, High efficiency detection of bi-stranded abasic clusters in gamma-irradiated DNA by putrescine, Nucleic Acids Res, 30 (2002) 2800-2808. [12] M. Gulston, J. Fulford, T. Jenner, C. de Lara, P. O'Neill, Clustered DNA damage induced by gamma radiation in human fibroblasts (HF19), hamster (V79-4) cells and plasmid DNA is revealed as Fpg and Nth sensitive sites, Nucleic Acids Res, 30 (2002) 3464-3472.

[13] A.A. Goodarzi, P. Jeggo, M. Lobrich, The influence of heterochromatin on DNA double strand break repair: Getting the strong, silent type to relax, DNA Repair (Amst), 9 (2010) 1273-1282. [14] A.G. Georgakilas, P.V. Bennett, D.M. Wilson, 3rd, B.M. Sutherland, Processing of bistranded abasic DNA clusters in gamma-irradiated human hematopoietic cells, Nucleic Acids Res, 32 (2004) 5609-5620. [15] A.G. Georgakilas, Processing of DNA damage clusters in human cells: current status of knowledge, Mol Biosyst, 4 (2008) 30-35. [16] N. Shikazono, M. Noguchi, K. Fujii, A. Urushibara, A. Yokoya, The yield, processing, and biological consequences of clustered DNA damage induced by ionizing radiation, J Radiat Res, 50 (2009) 27-36. [17] L.J. Eccles, P. O'Neill, M.E. Lomax, Delayed repair of radiation induced clustered DNA damage: friend or foe?, Mutat Res, 711 (2011) 134-141. [18] E. Sage, L. Harrison, Clustered DNA lesion repair in eukaryotes: relevance to mutagenesis and cell survival, Mutat Res, 711 (2011) 123-133. [19] J.O. Blaisdell, S.S. Wallace, Abortive base-excision repair of radiation-induced clustered DNA lesions in Escherichia coli, Proc Natl Acad Sci U S A, 98 (2001) 7426-7430. [20] N. Yang, M.A. Chaudhry, S.S. Wallace, Base excision repair by hNTH1 and hOGG1: a two edged sword in the processing of DNA damage in gamma-irradiated human cells, DNA Repair (Amst), 5 (2006) 43-51. [21] M. Gulston, C. de Lara, T. Jenner, E. Davis, P. O'Neill, Processing of clustered DNA damage generates additional double-strand breaks in mammalian cells post-irradiation, Nucleic Acids Res, 32 (2004) 1602-1609. [22] B.M. Sutherland, P.V. Bennett, N.S. Cintron, P. Guida, J. Laval, Low levels of endogenous oxidative damage cluster levels in unirradiated viral and human DNAs, Free Radic Biol Med, 35 (2003) 495-503. [23] P.V. Bennett, N.S. Cintron, L. Gros, J. Laval, B.M. Sutherland, Are endogenous clustered DNA damages induced in human cells?, Free Radic Biol Med, 37 (2004) 488-499. [24] V.A. Semenenko, R.D. Stewart, A fast Monte Carlo algorithm to simulate the spectrum of DNA damages formed by ionizing radiation, Radiat Res, 161 (2004) 451-457. [25] H. Nikjoo, P. O'Neill, W.E. Wilson, D.T. Goodhead, Computational approach for determining the spectrum of DNA damage induced by ionizing radiation., Radiat Res, 156 (2001) 577-583. [26] M. Hada, B.M. Sutherland, Spectrum of complex DNA damages depends on the incident radiation, Radiat Res, 165 (2006) 223-230.

[27] R. Watanabe, S. Rahmanian, H. Nikjoo, Spectrum of Radiation-Induced Clustered Non-DSB Damage - A Monte Carlo track structure modeling and calculations, Radiat Res, 183 (2015) 525-540. [28] J.P. Pouget, S. Frelon, J.L. Ravanat, I. Testard, F. Odin, J. Cadet, Formation of modified DNA bases in cells exposed either to gamma radiation or to high-LET particles, Radiat Res, 157 (2002) 589-595. [29] D. Tsao, P. Kalogerinis, I. Tabrizi, M. Dingfelder, R.D. Stewart, A.G. Georgakilas, Induction and processing of oxidative clustered DNA lesions in 56Fe-ion-irradiated human monocytes, Radiat Res, 168 (2007) 87-97. [30] H. Terato, R. Tanaka, Y. Nakaarai, T. Nohara, Y. Doi, S. Iwai, R. Hirayama, Y. Furusawa, H. Ide, Quantitative analysis of isolated and clustered DNA damage induced by gamma-rays, carbon ion beams, and iron ion beams, J Radiat Res, 49 (2008) 133-146. [31] Y. Tokuyama, Y. Furusawa, H. Ide, A. Yasui, H. Terato, Role of isolated and clustered DNA damage and the post-irradiating repair process in the effects of heavy ion beam irradiation, J Radiat Res, 56 (2015) 446-455. [32] E. Sage, S. Rousset, D. Perdiz, C. Martin, E. Balanzat, Detection of DNA damage induced by low and high TEL ionizing radiations, J Chim Phys Pcb, 94 (1997) 331-336. [33] J.C. Sutherland, D.C. Monteleone, J.G. Trunk, P.V. Bennett, B.M. Sutherland, Quantifying

DNA

damage

by

gel

electrophoresis,

electronic

imaging

and

number-average length analysis, Electrophoresis, 22 (2001) 843-854. [34] B.M. Sutherland, A.G. Georgakilas, P.V. Bennett, J. Laval, J.C. Sutherland, Quantifying clustered DNA damage induction and repair by gel electrophoresis, electronic imaging and number average length analysis, Mutat Res, 531 (2003) 93-107. [35] E.M. Filippova, D.C. Monteleone, J.G. Trunk, B.M. Sutherland, S.R. Quake, J.C. Sutherland, Quantifying double-strand breaks and clustered damages in DNA by single-molecule laser fluorescence sizing, Biophys J, 84 (2003) 1281-1290. [36]

K.

Datta,

R.D.

Neumann,

T.A.

Winters,

Characterization

of

complex

apurinic/apyrimidinic-site clustering associated with an authentic site-specific radiation-induced DNA double-strand break, Proc Natl Acad Sci U S A, 102 (2005) 10569-10574. [37] K. Datta, P. Jaruga, M. Dizdaroglu, R.D. Neumann, T.A. Winters, Molecular analysis of base damage clustering associated with a site-specific radiation-induced DNA double-strand break, Radiat Res, 166 (2006) 767-781. [38] E. Gollapalle, R. Wang, R. Adetolu, D. Tsao, D. Francisco, G. Sigounas, A.G. Georgakilas, Detection of oxidative clustered DNA lesions in X-irradiated mouse skin tissues and human MCF-7 breast cancer cells, Radiat Res, 167 (2007) 207-216. [39] K. Magnander, R. Hultborn, K. Claesson, K. Elmroth, Clustered DNA damage in irradiated human diploid fibroblasts: influence of chromatin organization, Radiat Res,

173 (2010) 272-282. [40] K. Akamatsu, N. Shikazono, A methodology for estimating localization of apurinic/apyrimidinic sites in DNA using fluorescence resonance energy transfer, Anal Biochem, 433 (2013) 171-180. [41] K. Akamatsu, N. Shikazono, T. Saito, Localization estimation of ionizing radiation-induced abasic sites in DNA in the solid state using fluorescence resonance energy transfer, Radiat Res, 183 (2015) 105-113. [42] A. Eschenbrenner, M.A.H. Du Penhoat, A. Boissiere, G. Eot-Houllier, F. Abel, M.F. Politis, A. Touati, E. Sage, A. Chetioui, Strand breaks induced in plasmid DNA by ultrasoft X-rays: Influence of hydration and packing, Int J of Radiat Biol, 83 (2007) 687-697. [43] P.K. Agrawala, A. Eschenbrenner, M.A.H. Du Penhoat, A. Boissiere, M.F. Politis, A. Touati, E. Sage, A. Chetioui, Induction and repairability of DNA damage caused by ultrasoft X-rays: Role of core events, Int J of Radiat Biol, 84 (2008) 1093-1103. [44] T. Shiina, R. Watanabe, I. Shiraishi, M. Suzuki, Y. Sugaya, K. Fujii, A. Yokoya, Induction of DNA damage, including abasic sites, in plasmid DNA by carbon ion and X-ray irradiation, Radiat Environ Bioph, 52 (2013) 99-112. [45] K.S. Saloua, G. Sonia, C. Pierre, S. Leon, H.J. Darel, The Relative Contributions of DNA Strand Breaks, Base Damage and Clustered Lesions to the Loss of DNA Functionality Induced by Ionizing Radiation, Radiat Res, 181 (2014) 99-110. [46] S.K. Sahbani, M. Rezaee, P. Cloutier, L. Sanche, D.J. Hunting, Non-DSB clustered DNA lesions induced by ionizing radiation are largely responsible for the loss of plasmid DNA functionality in the presence of cisplatin, Chem-Biol Interact, 217 (2014) 9-18. [47] L. Vysin, K.P. Brabcova, V. Stepan, P. Moretto-Capelle, B. Bugler, G. Legube, P. Cafarelli, R. Casta, J.P. Champeaux, M. Sence, M. Vlk, R. Wagner, J. Stursa, V. Zach, S. Incerti, L. Juha, M. Davidkova, Proton-induced direct and indirect damage of plasmid DNA, Radiat Environ Bioph, 54 (2015) 343-352. [48] D. Pang, S. Chasovskikh, J.E. Rodgers, A. Dritschilo, Short DNA fragments are a hallmark of heavy charged-particle irradiation and may underlie their greater therapeutic efficacy, Front Oncol, 6 (2016) 130. [49] S. Kouass Sahbani, L. Sanche, P. Cloutier, A.D. Bass, D.J. Hunting, Loss of cellular transformation efficiency induced by DNA irradiation with low-energy (10 eV) electrons, J Phys Chem B, 118 (2014) 13123-13131. [50] K. Rothkamm, M. Lobrich, Evidence for a lack of DNA double-strand break repair in human cells exposed to very low x-ray doses, Proc Natl Acad Sci USA, 100 (2003) 5057-5062. [51] A. Asaithamby, D.J. Chen, Mechanism of cluster DNA damage repair in response to high-atomic number and energy particles radiation, Mutat Res, 711 (2011) 87-99. [52] L. Lan, S. Nakajima, Y. Oohata, M. Takao, S. Okano, M. Masutani, S.H. Wilson, A. Yasui,

In situ analysis of repair processes for oxidative DNA damage in mammalian cells, Proc Natl Acad Sci USA, 101 (2004) 13738-13743. [53] A. Asaithamby, B.R. Hu, D.J. Chen, Unrepaired clustered DNA lesions induce chromosome breakage in human cells, Proc Natl Acad Sci USA, 108 (2011) 8293-8298. [54] Y. Lorat, C.U. Brunner, S. Schanz, B. Jakob, G. Taucher-Scholz, C.E. Rube, Nanoscale analysis of clustered DNA damage after high-LET irradiation by quantitative electron microscopy - The heavy burden to repair, DNA Repair (Amst), 28 (2015) 93-106. [55] E. Sage, P.M. Girard, S. Francesconi, Unravelling UVA-induced mutagenesis, Photochem Photobiol Sci, 11 (2012) 74-80. [56] D. Graindorge, S. Martineau, C. Machon, P. Arnoux, J. Guitton, S. Francesconi, C. Frochot, E. Sage, P.M. Girard, Singlet Oxygen-Mediated Oxidation during UVA Radiation Alters the Dynamic of Genomic DNA Replication, PLoS One, 10 (2015) e0140645. [57] J.P. Pouget, T. Douki, M.J. Richard, J. Cadet, DNA damage induced in cells by gamma and UVA radiation as measured by HPLC/GC-MS and HPLC-EC and Comet assay, Chem Res Toxicol, 13 (2000) 541-549. [58] J. Cadet, T. Douki, Oxidatively generated damage to DNA by UVA radiation in cells and human skin, J Invest Dermatol, 131 (2011) 1005-1007. [59] R. Greinert, B. Volkmer, S. Henning, E.W. Breitbart, K.O. Greulich, M.C. Cardoso, A. Rapp, UVA-induced DNA double-strand breaks result from the repair of clustered oxidative DNA damages, Nucleic Acids Res, 40 (2012) 10263-10273. [60] V. Sharma, L.B. Collins, T.H. Chen, N. Herr, S. Takeda, W. Sun, J.A. Swenberg, J. Nakamura, Oxidative stress at low levels can induce clustered DNA lesions leading to NHEJ mediated mutations, Oncotarget, 7 (2016) 25377-25390. [61] P. Regulus, B. Duroux, P.A. Baylet, A. Favier, J. Cadet, J.L. Ravanat, Oxidation of the sugar radiation or bleomycin of a cluster DNA lesion moiety of DNA by ionizing could induce the formation, Proc Natl Acad Sci USA, 104 (2007) 14032-14037. [62] J. Cadet, J.R. Wagner, Oxidatively generated base damage to cellular DNA by hydroxyl radical and one-electron oxidants: similarities and differences, Arch Biochem Biophys, 557 (2014) 47-54. [63] Z. Lin, C. de los Santos, NMR characterization of clustered bistrand abasic site lesions: effect of orientation on their solution structure, J Mol Biol, 308 (2001) 341-352. [64] R.D. Hazel, K. Tian, C. de Los Santos, NMR solution structures of bistranded abasic site lesions in DNA, Biochemistry, 47 (2008) 11909-11919. [65] H. Fujimoto, M. Pinak, T. Nemoto, P. O'Neill, E. Kume, K. Saito, H. Maekawa, Molecular dynamics simulation of clustered DNA damage sites containing 8-oxoguanine and abasic site, J Comput Chem, 26 (2005) 788-798. [66] J. Zalesak, M. Lourdin, L. Krejci, J.F. Constant, M. Jourdan, Structure and Dynamics of DNA Duplexes Containing a Cluster of Mutagenic 8-Oxoguanine and Abasic Site Lesions, J

Mol Biol, 426 (2014) 1524-1538. [67] E. Bignon, H. Gattuso, C. Morell, F. Dehez, A.G. Georgakilas, A. Monari, E. Dumont, Correlation of bistranded clustered abasic DNA lesion processing with structural and dynamic DNA helix distortion, Nucleic Acids Res, 44 (2016) 8588-8599. [68] M.E. Lomax, S. Cunniffe, P. O'Neill, Efficiency of repair of an abasic site within DNA clustered damage sites by mammalian cell nuclear extracts, Biochemistry, 43 (2004) 11017-11026. [69] G. Eot-Houllier, M. Gonera, D. Gasparutto, C. Giustranti, E. Sage, Interplay between DNA N-glycosylases/AP lyases at multiply damaged sites and biological consequences, Nucleic Acids Res, 35 (2007) 3355-3366. [70] N. Shikazono, C. Pearson, P. O'Neill, J. Thacker, The roles of specific glycosylases in determining the mutagenic consequences of clustered DNA base damage, Nucleic Acids Res, 34 (2006) 3722-3730. [71] R. Okayasu, Repair of DNA damage induced by accelerated heavy ions--a mini review, Int J Cancer, 130 (2012) 991-1000. [72] T.J. Jenner, C.M. deLara, P. O'Neill, D.L. Stevens, Induction and rejoining of DNA double-strand

breaks

in

V79-4

mammalian

cells

following

gamma-

and

alpha-irradiation, Int J Radiat Biol, 64 (1993) 265-273. [73] P.S. Hodgkins, M.P. Fairman, P. O'Neill, Rejoining of gamma-radiation-induced single-strand breaks in plasmid DNA by human cell extracts: dependence on the concentration of the hydroxyl radical scavenger, Tris, Radiat Res, 145 (1996) 24-30. [74] S. Cunniffe, P. O'Neill, The complexity of radiation-induced DNA damage as revealed by exposure to cell extracts, Radiat Res, 152 (1999) 421-427. [75] M.A. du Penhoat, A. Eschenbrenner, F. Abel, A. Boissiere, J.M. Guigner, A. Chetioui, M.F. Politis, A. Touati, E. Sage, T.J. Jenner, D.L. Stevens, M.A. Hill, Double-strand break induction and repair in V79-4 hamster cells: the role of core ionisations, as probed by ultrasoft X-rays, Int J Radiat Biol, 86 (2010) 205-219. [76] A. Boissiere, C. Champion, A. Touati, M.A. Herve du Penhoat, L. Sabatier, A. Chatterjee, A. Chetioui, DNA core ionization and cell inactivation, Radiat Res, 167 (2007) 493-500. [77] E. Pastwa, R.D. Neumann, K. Mezhevaya, T.A. Winters, Repair of radiation-induced DNA double-strand breaks is dependent upon radiation quality and the structural complexity of double-strand breaks, Radiat Res, 159 (2003) 251-261. [78] K. Datta, S. Purkayastha, R.D. Neumann, E. Pastwa, T.A. Winters, Base damage immediately upstream from double-strand break ends is a more severe impediment to nonhomologous end joining than blocked 3'-termini, Radiat Res, 175 (2011) 97-112. [79] B. Stenerlow, K.H. Karlsson, B. Cooper, B. Rydberg, Measurement of prompt DNA double-strand breaks in mammalian cells without including heat-labile sites: Results for

cells deficient in nonhomologous end joining, Radiat Res, 159 (2003) 502-510. [80] A.S. Gustafsson, T. Hartman, B. Stenerlow, Formation and repair of clustered damaged DNA sites in high LET irradiated cells, Int J Radiat Biol, 91 (2015) 820-826. [81] F. Tobias, M. Durante, G. Taucher-Scholz, B. Jakob, Spatiotemporal analysis of DNA repair using charged particle radiation, Mutat Res, 704 (2010) 54-60. [82] S. Moore, F.K.T. Stanley, A.A. Goodarzi, The repair of environmentally relevant DNA double strand breaks caused by high linear energy transfer irradiation--no simple task, DNA Repair (Amst), 17 (2014) 64-73. [83] S.V. Costes, I. Chiolo, J.M. Pluth, M.H. Barcellos-Hoff, B. Jakob, Spatiotemporal characterization of ionizing radiation induced DNA damage foci and their relation to chromatin organization, Mutat Res, 704 (2010) 78-87. [84] I.Y. Belyaev, Radiation-induced DNA repair foci: spatio-temporal aspects of formation, application for assessment of radiosensitivity and biological dosimetry, Mutat Res, 704 (2010) 132-141. [85] M. Falk, E. Lukasova, B. Gabrielova, V. Ondrej, S. Kozubek, Chromatin dynamics during DSB repair, Biochim Biophys Acta, 1773 (2007) 1534-1545. [86] A.A. Goodarzi, A.T. Noon, D. Deckbar, Y. Ziv, Y. Shiloh, M. Lobrich, P.A. Jeggo, ATM signaling

facilitates

repair

of

DNA

double-strand

breaks

associated

with

heterochromatin, Mol Cell, 31 (2008) 167-177. [87] B. Jakob, J. Splinter, S. Conrad, K.O. Voss, D. Zink, M. Durante, M. Lobrich, G. Taucher-Scholz, DNA double-strand breaks in heterochromatin elicit fast repair protein recruitment, histone H2AX phosphorylation and relocation to euchromatin, Nucleic Acids Res, 39 (2011) 6489-6499. [88] T. Neumaier, J. Swenson, C. Pham, A. Polyzos, A.T. Lo, P. Yang, J. Dyball, A. Asaithamby, D.J. Chen, M.J. Bissell, S. Thalhammer, S.V. Costes, Evidence for formation of DNA repair centers and dose-response nonlinearity in human cells, Proc Natl Acad Sci U S A, 109 (2012) 443-448. [89] I. Chiolo, A. Minoda, S.U. Colmenares, A. Polyzos, S.V. Costes, G.H. Karpen, Double-strand breaks in heterochromatin move outside of a dynamic HP1a domain to complete recombinational repair, Cell, 144 (2011) 732-744. [90] N.I. Nakajima, H. Brunton, R. Watanabe, A. Shrikhande, R. Hirayama, N. Matsufuji, A. Fujimori, T. Murakami, R. Okayasu, P. Jeggo, A. Shibata, Visualisation of gammaH2AX foci caused by heavy ion particle traversal; distinction between core track versus non-track damage, PLoS One, 8 (2013) e70107. [91] P. Reynolds, J.A. Anderson, J.V. Harper, M.A. Hill, S.W. Botchway, A.W. Parker, P. O'Neill, The dynamics of Ku70/80 and DNA-PKcs at DSBs induced by ionizing radiation is dependent on the complexity of damage, Nucleic Acids Res, 40 (2012) 10821-10831. [92] R. Okayasu, M. Okada, A. Okabe, M. Noguchi, K. Takakura, S. Takahashi, Repair of

DNA damage induced by accelerated heavy ions in mammalian cells proficient and deficient in the non-homologous end-joining pathway, Radiat Res, 165 (2006) 59-67. [93] A. Gerelchuluun, E. Manabe, T. Ishikawa, L. Sun, K. Itoh, T. Sakae, K. Suzuki, R. Hirayama, A. Asaithamby, D.J. Chen, K. Tsuboi, The major DNA repair pathway after both proton and carbon-ion radiation is NHEJ, but the HR pathway is more relevant in carbon ions, Radiat Res, 183 (2015) 345-356. [94] D. Pang, T.A. Winters, M. Jung, S. Purkayastha, L.R. Cavalli, S. Chasovkikh, B.R. Haddad, A. Dritschilo, Radiation-generated short DNA fragments may perturb non-homologous end-joining and induce genomic instability, J Radiat Res, 52 (2011) 309-319. [95] A. Beucher, J. Birraux, L. Tchouandong, O. Barton, A. Shibata, S. Conrad, A.A. Goodarzi, A. Krempler, P.A. Jeggo, M. Lobrich, ATM and Artemis promote homologous recombination of radiation-induced DNA double-strand breaks in G2, EMBO J, 28 (2009) 3413-3427. [96] N.B. Averbeck, O. Ringel, M. Herrlitz, B. Jakob, M. Durante, G. Taucher-Scholz, DNA end resection is needed for the repair of complex lesions in G1-phase human cells, Cell Cycle, 13 (2014) 2509-2516. [97] P. Peddi, C.W. Loftin, J.S. Dickey, J.M. Hair, K.J. Burns, K. Aziz, D.C. Francisco, M.I. Panayiotidis, O.A. Sedelnikova, W.M. Bonner, T.A. Winters, A.G. Georgakilas, DNA-PKcs deficiency leads to persistence of oxidatively induced clustered DNA lesions in human tumor cells, Free radical biology & medicine, 48 (2010) 1435-1443. [98] J.M. Hair, G.I. Terzoudi, V.I. Hatzi, K.A. Lehockey, D. Srivastava, W.X. Wang, G.E. Pantelias, A.G. Georgakilas, BRCA1 role in the mitigation of radiotoxicity and chromosomal instability through repair of clustered DNA lesions, Chem-Biol Interact, 188 (2010) 350-358. [99] S. Malyarchuk, R. Castore, L. Harrison, DNA repair of clustered lesions in mammalian cells: involvement of non-homologous end-joining, Nucleic Acids Res, 36 (2008) 4872-4882. [100] S. Malyarchuk, R. Castore, L. Harrison, Apex1 can cleave complex clustered DNA lesions in cells, DNA Repair (Amst), 8 (2009) 1343-1354. [101] S. Malyarchuk, R. Castore, R. Shi, L. Harrison, Artemis is required to improve the accuracy of repair of double-strand breaks with 5'-blocked termini generated from non-DSB-clustered lesions, Mutagenesis, 28 (2013) 357-366. [102] S.G. Kozmin, Y. Sedletska, A. Reynaud-Angelin, D. Gasparutto, E. Sage, The formation of double-strand breaks at multiply damaged sites is driven by the kinetics of excision/incision at base damage in eukaryotic cells, Nucleic Acids Res, 37 (2009) 1767-1777. [103] N. Shikazono, K. Akamatsu, M. Takahashi, M. Noguchi, A. Urushibara, P. O'Neill, A.

Yokoya, Significance of DNA polymerase I in in vivo processing of clustered DNA damage, Mutat Res, 749 (2013) 9-15. [104] M. Noguchi, A. Urushibara, A. Yokoya, P. O'Neill, N. Shikazono, The mutagenic potential of 8-oxoG/single strand break-containing clusters depends on their relative positions, Mutat Res, 732 (2012) 34-42. [105] S. Cunniffe, A. Walker, R. Stabler, P. O'Neill, M.E. Lomax, Increased mutability and decreased repairability of a three-lesion clustered DNA-damaged site comprised of an AP site and bi-stranded 8-oxoG lesions, Int J Radiat Biol, 90 (2014) 468-479. [106] S. Cunniffe, P. O'Neill, M.M. Greenberg, M.E. Lomax, Reduced repair capacity of a DNA clustered damage site comprised of 8-oxo-7,8-dihydro-2'-deoxyguanosine and 2-deoxyribonolactone results in an increased mutagenic potential of these lesions, Mutat Res, 762 (2014) 32-39. [107] Y. Sedletska, J.P. Radicella, E. Sage, Replication fork collapse is a major cause of the high mutation frequency at three-base lesion clusters, Nucleic Acids Res, 41 (2013) 9339-9348. [108] S.M. Cunniffe, M.E. Lomax, P. O'Neill, An AP site can protect against the mutagenic potential of 8-oxoG when present within a tandem clustered site in E. coli, DNA Repair (Amst), 6 (2007) 1839-1849. [109] G. Eot-Houllier, S. Eon-Marchais, D. Gasparutto, E. Sage, Processing of a complex multiply damaged DNA site by human cell extracts and purified repair proteins, Nucleic Acids Res, 33 (2005) 260-271. [110] E.A. Belousova, I.A. Vasil'eva, N.A. Moor, T.S. Zatsepin, T.S. Oretskaya, O.I. Lavrik, Clustered DNA lesions containing 5-formyluracil and AP site: repair via the BER system, PLoS One, 8 (2013) e68576. [111] E.A. Belousova, O.I. Lavrik, Repair of Clustered Damage and DNA Polymerase Iota, Biochemistry (Mosc), 80 (2015) 1010-1018. [112] L.J. Eccles, M.E. Lomax, P. O'Neill, Hierarchy of lesion processing governs the repair, double-strand break formation and mutability of three-lesion clustered DNA damage, Nucleic Acids Res, 38 (2010) 1123-1134. [113] M. Falk, E. Lukasova, S. Kozubek, Chromatin structure influences the sensitivity of DNA to gamma-radiation, Biochim Biophys Acta, 1783 (2008) 2398-2414. [114] H. Takata, T. Hanafusa, T. Mori, M. Shimura, Y. Iida, K. Ishikawa, K. Yoshikawa, Y. Yoshikawa, K. Maeshima, Chromatin compaction protects genomic DNA from radiation damage, PLoS One, 8 (2013) e75622. [115] W.J. Cannan, B.P. Tsang, S.S. Wallace, D.S. Pederson, Nucleosomes suppress the formation of double-strand DNA breaks during attempted base excision repair of clustered oxidative damages, J Biol Chem, 289 (2014) 19881-19893. [116] V. Singh, B. Kumari, P. Das, Repair efficiency of clustered abasic sites by APE1 in

nucleosome core particles is sequence and position dependent, Rsc Adv, 5 (2015) 23691-23698. [117] L.J. Eccles, H. Menoni, D. Angelov, M.E. Lomax, P. O'Neill, Efficient cleavage of single and clustered AP site lesions within mono-nucleosome templates by CHO-K1 nuclear extract contrasts with retardation of incision by purified APE1, DNA Repair (Amst), 35 (2015) 27-36. [118] F. Zafar, S.B. Seidler, A. Kronenberg, D. Schild, C. Wiese, Homologous recombination contributes to the repair of DNA double-strand breaks induced by high-energy iron ions, Radiat Res, 173 (2010) 27-39. [119] M. Lobrich, P.A. Jeggo, The impact of a negligent G2/M checkpoint on genomic instability and cancer induction, Nat Rev Cancer, 7 (2007) 861-869. [120] B.D. Loucas, M.N. Cornforth, The LET dependence of unrepaired chromosome damage in human cells: a break too far?, Radiat Res, 179 (2013) 393-405. [121] Y. Kagawa, T. Shimazu, A.J. Gordon, N. Fukunishi, N. Inabe, M. Suzuki, M. Hirano, T. Kato, M. Watanabe, F. Hanaoka, F. Yatagai, Complex hprt deletion events are recovered after exposure of human lymphoblastoid cells to high-LET carbon and neon ion beams, Mutagenesis, 14 (1999) 199-205. [122] K. Masumura, K. Kuniya, T. Kurobe, M. Fukuoka, F. Yatagai, T. Nohmi, Heavy-ion-induced mutations in the gpt delta transgenic mouse: comparison of mutation spectra induced by heavy-ion, X-ray, and gamma-ray radiation, Environ Mol Mutagen, 40 (2002) 207-215. [123] B.K. Singleton, C.S. Griffin, J. Thacker, Clustered DNA damage leads to complex genetic changes in irradiated human cells, Cancer Res, 62 (2002) 6263-6269. [124] D. Grygoryev, S. Gauny, M. Lasarev, A. Ohlrich, A. Kronenberg, M.S. Turker, Charged particle mutagenesis at low dose and fluence in mouse splenic T cells, Mutat Res, 788 (2016) 32-40. [125] A.B. Adewoye, S.J. Lindsay, Y.E. Dubrova, M.E. Hurles, The genome-wide effects of ionizing radiation on mutation induction in the mammalian germline, Nat Commun, 6 (2015) 6684. [126] J.F. Ward, DNA damage produced by ionizing radiation in mammalian cells: identities, mechanisms of formation, and reparability, Prog Nucleic Acid Res Mol Biol, 35 (1988) 95-125. [127] W. Burkart, T. Jung, G. Frasch, Damage pattern as a function of radiation quality and other factors, C R Acad Sci III, 322 (1999) 89-101. [128] M.M. Vilenchik, A.G. Knudson, Endogenous DNA double-strand breaks: production, fidelity of repair, and induction of cancer, Proc Natl Acad Sci U S A, 100 (2003) 12871-12876.

Figure 1. Scheme of oxidatively generated DNA damage induced by ionizing radiation. Non-DSB clustered lesions comprises 2, 3 or more oxidatively generated base damage (noted oxyPu, oxyPy, U), abasic site, SSB within one or two helix turns of DNA, produced by a single radiation track, the complexity of which increases with increasing LET. Complex DSB are DSB associated with several oxidized bases and abasic sites. Figure 2. Summary of the processing of non-DSB clustered lesions and complex DSB, and biological consequences. SSB produced either directly by radiation or excision/incision at oxyPy or U or by incision at AP site inhibits BER of the opposing oG (or some oxyPy), until it is repaired. DSB formation is thus avoided and the remaining base damage can be repaired as single lesion. However, depending on the type of the base damage in the clusters, the orientation and the interlesion distance, the repair of the SSB may be retarded and replication may occur before repair, increasing the likelihood of mutation at the remaining base damage (oG). Clusters containing bistranded U, AP, U/AP and oxyPy are rapidly cleaved and result in DSB. Such DSB carrying unrepaired base damage nearby are slowly and inaccurately rejoined by NHEJ in mammalian cells and leads to deletion. In the case of single incision at three- or more lesion clusters, the repair of the generated SSB is delayed and replication may occur before repair ; the replication fork thus collapses, resulting in mutation induction (eventually also complex DSB in mammalian cells). Complex DSB are repaired with slow kinetics and persist for longer time. A fraction (probably small) of them are repaired in G1 phase by NHEJ, leading to deletion, while more of them seem to be repaired later by error-prone HR which generates translocations and large deletions. Another fraction may outlast the checkpoint arrest and pass through mitosis. Persisting complex DSB result in sister chromatid exchanges (SCE) and chromosomal aberrations (CA) in surviving cells. oxyPy, oxidized pyrimidine ; oG, 8-oxoG ; U, uracil ; AP, abasic site ; GAP, gap. The base damage oG is expected to be the main oxyPu to be present in radio-induced clustered DNA lesions and to be mostly unrepaired.

Table 1. DNA damage induced by ionizing radiation* type of damage single strand breaks

radioinduced damage

endogenous damage

per cell per Gy

per cell per day

1000

> 10OOO

base damage

2000

3200

abasic sites

250

12600

Double strand breaks

40

40-50***

DNA-protein XL

150

?

122**

?

non-DSB clustered lesions complex DSB

?

* from 126-127 ** from 14, 29 *** from 128

Highlights x

Clustered DNA lesions are a hallmark of ionizing radiation

x

Clustered DNA lesions consists of >2 oxidatively generated base damage, abasic sites, SSB or DSB within 10-40 bp

x

Complexity of clustered DNA lesions increases with increasing radiation LET

x

Incomplete repair at non-DSB clusters may result in DSB formation

x

Clustered DNA lesions are slowly and inefficiently repaired, leading to genome instability

10bp = 1 helical turn

Sugar-phosphate hosphat backbone

OxyPy

OxyPu

AP site

strand break

base bas

Isolated damage

Non-DSB Clustered lesions

Simple DSB

Complex DSB

non-DSB clusters

Complex DSB oG

U/AP

oG

oPy

oPy

oG

AP

oG

AP

U/AP

rapid excision/ Incision at oPy/U/AP oG

oG

GAP

no repair

AP

Cleavage at AP; GAP

rapid cleavage leading to DSB oG G

oG AP/ oPy

Gap filling

retarded repair air of oG oG G

Slow, inaccurate rejoining (NHEJ)

Replication: strand loss / replication fork collapse

slow inaccurate rejoining

SCE CA HR H

translocations l

AP/ oPy

oG G repair air

no effect

large deletions

replication tion

n at oG mutation

Graphical Abstract

deletion

mutation in opposite strand