Cell Reports

Article Escape from Telomere-Driven Crisis Is DNA Ligase III Dependent Rhiannon E. Jones,1,3 Sehyun Oh,2,3 Julia W. Grimstead,1,3 Jacob Zimbric,2 Laureline Roger,1 Nicole H. Heppel,1 Kevin E. Ashelford,1 Kate Liddiard,1 Eric A. Hendrickson,2,4 and Duncan M. Baird1,4,* 1Institute

of Cancer and Genetics, School of Medicine, Cardiff University, Heath Park, Cardiff CF14 4XN, UK of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota Medical School, Minneapolis, MN 55455, USA 3Co-first author 4Co-senior author *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2014.07.007 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). 2Department

SUMMARY

Short dysfunctional telomeres are capable of fusion, generating dicentric chromosomes and initiating breakage-fusion-bridge cycles. Cells that escape the ensuing cellular crisis exhibit large-scale genomic rearrangements that drive clonal evolution and malignant progression. We demonstrate that there is an absolute requirement for fully functional DNA ligase III (LIG3), but not ligase IV (LIG4), to facilitate the escape from a telomere-driven crisis. LIG3- and LIG4-dependent alternative (A) and classical (C) nonhomologous end-joining (NHEJ) pathways were capable of mediating the fusion of short dysfunctional telomeres, both displaying characteristic patterns of microhomology and deletion. Cells that failed to escape crisis exhibited increased proportions of C-NHEJ-mediated interchromosomal fusions, whereas those that escaped displayed increased proportions of intrachromosomal fusions. We propose that the balance between inter- and intrachromosomal telomere fusions dictates the ability of human cells to escape crisis and is influenced by the relative activities of Aand C-NHEJ at short dysfunctional telomeres. INTRODUCTION Telomeres are specialized nucleoprotein structures that cap the ends of linear eukaryotic chromosomes and prevent the recognition of the chromosomal terminus by the DNA damage response apparatus (de Lange, 2005). Ongoing cell division results in gradual telomere erosion (Harley et al., 1990), and ultimately, the loss of the end-capping function; in the context of a functional DNA damage response, this leads to the induction of a p53-dependent G1/S cell cycle arrest, known as replicative senescence (d’Adda di Fagagna et al., 2003). This cell-intrinsic limit on replicative lifespan provides a stringent tumor suppressive mechanism. However, in the absence of a fully functional

DNA damage checkpoint response, cells containing short dysfunctional telomeres enter a state of crisis during which telomeres undergo fusion with other telomeres or nontelomeric double-stranded DNA breaks (DSBs), creating dicentric chromosomes and initiating breakage-fusion-bridge cycles (Counter et al., 1992; Murnane, 2012). This leads to the creation of genomic rearrangements, including nonreciprocal translocations that are common in cells from many different tumor types (Artandi et al., 2000; Shih et al., 2001). Therefore, crisis is a key event driving genomic instability and clonal evolution during the progression to malignancy; this is consistent with data obtained from mouse models (Artandi et al., 2000) and observations of telomere dynamics and fusion in a broad range of human tumor types (Lin et al., 2010; Meeker et al., 2004; Roger et al., 2013). Telomere fusion is an important and physiologically relevant, mutational event; however, the molecular mechanisms that result in the recombination of short dysfunctional human telomeres are not fully understood. Key to the function of mammalian telomeres is the shelterin complex that plays a fundamental role in protecting the natural chromosomal termini from aberrant nonhomologous end joining (NHEJ; de Lange, 2005). The abrogation of TRF2, a core component of shelterin, confers a widespread telomere fusion phenotype (van Steensel et al., 1998) that depends on the activities of DNA ligase IV (LIG4), specifically required for the classical NHEJ (C-NHEJ) pathway (Smogorzewska et al., 2002). In contrast, fusion was readily detected in telomerase-deficient mice, with short dysfunctional telomeres, despite the absence of core components of C-NHEJ pathway, including DNA-PKcs, LIG4 or 53BP1 (Maser et al., 2007; Rai et al., 2010). Thus, in the context of short dysfunctional telomeres, which is likely the most biologically relevant form of telomere dysfunction, telomeres may no longer be fully recognized by the shelterin complex and the processing of telomere fusion may be mediated by alternative end-joining processes. This view is consistent with the molecular analysis of telomere fusion events directly from human cells undergoing a telomere-driven crisis in culture, where fusion is accompanied by deletion into the telomere-adjacent DNA and microhomology is observed across the fusion points (Capper et al., 2007; Letsolo et al., 2010). This characteristic profile is also observed at telomere fusion junctions isolated from human malignancies, including

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1063

early stage and premalignant lesions (Lin et al., 2010; Roger et al., 2013), as well as in normal human cells, in which rare stochastic telomeric deletion results in fusion (Capper et al., 2007; Lin et al., 2010; Roger et al., 2013). The profile of mutation that accompanies telomere fusion is indicative of the involvement of error-prone, alternative-NHEJ (A-NHEJ) processes (Boulton and Jackson, 1996; Feldmann et al., 2000). The concept that A-NHEJ processes may play a role in mediating telomere fusion is further supported by observations in mouse cells, where following the removal of the TPP1-Pot1 proteins or the deletion of both TRF1 and TRF2, telomere fusion is independent of Ku70, ATM, LIG4, and 53BP1 (Rai et al., 2010; Sfeir and de Lange, 2012). Furthermore, molecular analysis of fusion events following replicative telomere erosion in human cells carrying hypomorphic meiotic recombination 11 (MRE11) alleles revealed a change in the mutational spectrum with an increase in insertions at the fusion point (Tankimanova et al., 2012). The existence of an A-NHEJ pathway in mammalian cells that is mutagenic and uses microhomology is only apparent in the absence the C-NHEJ components LIG4, XRCC4, and Ku70Ku80. In mouse models, A-NHEJ is required for the formation of complex translocations and amplifications initiated by RAGinduced DNA cleavage (Difilippantonio et al., 2002; Zhu et al., 2002) or at induced site-specific breaks (Guirouilh-Barbat et al., 2004), as well as class-switch recombination (Boboila et al., 2010; Yan et al., 2007). Several proteins have been implicated in A-NHEJ, including MRE11, which provides scaffold functions and resection activity (the latter facilitated by the CtIP-interacting protein), to expose potential microhomologies (Zhang and Jasin, 2011). Components that play roles within the base excision repair pathways have also been implicated in A-NHEJ. These include poly [ADP-ribose] polymerase 1, DNA ligase III (LIG3), and, to a lesser extent, DNA ligase I (Boboila et al., 2012; Simsek et al., 2011a). In this study, we examined the contributions of both the CNHEJ and A-NHEJ pathways in facilitating the fusion of short dysfunctional telomeres in human cells following replicative erosion. Our data are consistent with a role for both pathways in mediating telomere fusion, but importantly they show that the telomere fusion mediated by LIG3 allows cells to escape a telomere-driven crisis. We propose that the balance between intra- and interchromosomal telomere fusion dictates the ability of cells to escape crisis and this may represent a key mechanism driving clonal evolution. RESULTS To study telomere fusion in the absence of C-NHEJ and A-NHEJ, we used HCT116 cell lines in which either one or both alleles of the LIG4 (Khan et al., 2011; Oh et al., 2013) or LIG3 (Oh et al., 2014) genes had been inactivated using rAAV-mediated gene targeting. HCT116, a colorectal carcinoma cell line, is diploid with a stable karyotype, and exhibits functional double-stranded DNA (dsDNA) break repair and cell-cycle checkpoint control (Hendrickson, 2008). LIG4
/
HCT116 cells displayed extreme sensitivity to DNA damaging agents, genomic instability, and an increased use of microhomology during dsDNA repair (Oh et al., 2013), indicating that these cells use A-NHEJ in the

absence of C-NHEJ. The LIG3 gene encodes two mRNA isoforms; nuclear and mitochondrial, the mitochondrial form of which is essential and nonredundant (Simsek and Jasin, 2011). We rescued the lethal phenotype of LIG3
/
cells by complementation with the mitochondrial isoform alone, allowing the creation of a viable cell line that lacked nuclear LIG3 (LIG3
/
:mL3; Oh et al., 2014). Escape from Crisis Is Accompanied by Telomere Fusion and Genomic Rearrangements To examine the fusion of short dysfunctional telomeres in the absence of C- or A-NHEJ activities, we transfected wild-type (WT), LIG3+/
, LIG3
/
:mL3, and LIG4
/
HCT116 cells with a dominant-negative hTERT (DN-hTERT) construct, shown to inhibit telomerase activity and induce telomere erosion (Hahn et al., 1999). We subsequently picked single-cell clones and analyzed telomerase activity, telomere dynamics, and fusion. The expression of DN-hTERT effectively inhibited telomerase activity (Figure 1A; Figure S1A). Telomere dynamics were monitored using single telomere length analysis (STELA); the WT HCT116 culture displayed a mean telomere length of 4.0 kb, whereas LIG3- and LIG4-inactivated cells displayed telomere lengths of approximately half that size (Figure S1B). Gradual telomere erosion was observed in the WT HCT116DN-hTERT at a mean rate of 45 bp/population doublings (PD; Figures 1B and S1C) that was similar to that observed in telomerase-negative fibroblast cultures (p = 0.42; Figure S1C; Baird et al., 2003); no erosion could be detected in empty vector controls (data not shown). Telomere erosion in the WT HCT116DN-hTERT clones continued to a point at which the means of the telomere length distributions were approximately 1.4 kb, with a lower limit of 300 bp; at this point, telomere length stabilized despite ongoing cell division (Figure 1B). Importantly, the telomere dynamics and the lower limit of telomere erosion observed in the WT HCT116DN-hTERT clones were indistinguishable from those observed in normal human fibroblast cultures undergoing a telomere-driven crisis following the expression of HPVE6E7 (Capper et al., 2007). Coincident with the stabilization of telomere length (after 40–45 PD), all 11 HCT116DN-hTERT clonal populations entered a transient crisis-like state that was characterized by slowed cell growth (Figures 1C and 2A) and the appearance of enlarged, vacuolated, multinucleate cells (Figure S1D) and increased apoptosis (Figure S1E). However, all of the HCT116DN-hTERT clonal populations eventually escaped crisis and continued to over 85 PD, at which point the cultures were intentionally terminated (Figure 2A). Following the escape from crisis, telomerase activity was detected and telomere elongation was observed (Figures 1A, S1A, and S1F). Telomere fusion was analyzed using a singlemolecule PCR strategy targeted to multiple chromosome ends (Letsolo et al., 2010). Telomere fusion was not detected until telomere length had reached the lower limit of erosion and the cells had entered a period of crisis (e.g., PD45; Figure 1D). Following the escape from crisis, telomere fusion was no longer detected (e.g., PD64; Figure 1D) and was consistent with telomerasemediated telomere lengthening in these postcrisis cultures, stabilizing telomeres and preventing further fusion (Figure S1F). Telomere fusion during crisis creates dicentric chromosomes that can initiate breakage-fusion-bridge cycles, driving genomic

1064 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

Figure 1. Telomere Erosion, Fusion, and the Induction of and Escape from Crisis in HCT116 Cells (A) Telomerase activity plotted as total product generated (TPG; mean ± SD). (B) STELA of the 17p telomere. The PD from the point of single cell cloning is show above, the mean and SD of the telomere length profiles are show below, together with the rate of erosion prior to crisis. Dotted red line indicates the lower limit of telomere erosion. (C) Growth curve plotting PD from the point of single cloning and days in cultures. PD points analyzed with STELA and for fusions are indicated and the period of crisis is shown in red. The sampling points for aCGH are shown. (D) Single-molecule telomere fusion analysis, using oligonucleotide PCR primers targeted to XpYp, 17p, and the 21q family of related telomeres (Letsolo et al., 2010) at the PD points as indicated. (E) aCGH on samples taken before and after crisis. The data are plotted as a Log2 ratio, whereas amplification of the hTERT locus is indicated in red. The data shown are all derived from the same HCT116 DN-hTERT clone. See also Figure S1.

amplified the hTERT locus. We therefore consider that this system represents a good biological model for a telomereinduced crisis.

instability and clonal evolution. We therefore looked for changes to the genomes of the clones as consequence of the escape from crisis, and to do this we compared the genomes from the same clone before and after crisis with array-CGH (Figure 1C). The escape from crisis was characterized by the appearance of large-scale clonal genomic rearrangements, which in some clones included amplification of the hTERT locus in the terminal regions of chromosome 5p (Figures 1E and S1G). These data are consistent with gradual telomere erosion as a consequence of telomerase inhibition with the DN-hTERT construct in the WT HCT116 cells, resulting in the induction of a telomere-driven crisis, telomere fusion, genomic instability, and the clonal selection for cells that have either lost the DN-hTERT construct or

Escape from Crisis Requires LIG3 Clonal populations of LIG4
/
, LIG3+/
, and LIG3
/
:mL3 HCT116 cells expressing DN-hTERT showed similar levels of telomerase inhibition to that observed in HCT116 WT cells (p = 0.29 ANOVA; Figure S1A), they exhibited telomere erosion at rates indistinguishable from HCT116 WT or telomerase-negative fibroblast strains (Figure S1C) and all entered a crisis-like state at between 12 and 40 PD similar to that observed in HCT116 WT cells (Figures 2B–2D, 3A, 3B, S3A, and S3B); no such crisis state was observed in empty vector controls (Figures S2A and S2B). Consistent with previous observations (Hahn et al., 1999), the number of PDs obtained at the point of crisis correlated with the initial telomere length of the parental cultures; those with longer telomeres underwent more PDs prior to crisis (p = 0.03, Figure S3C). Telomere erosion in the LIG4
/
, LIG3+/
, and LIG3
/
:mL3 deficient cells proceeded to a minimum telomere length (300 bp), similar to that observed in WT HCT116DN-hTERT
/
(Figures 3C and 3D). A total of 13 LIG4 clones were picked and 11 (85%) escaped crisis (Figure 2B and 2H), and of the 14 LIG3+/
clones picked, seven escaped crisis (50%; Figures 2C and 2H). The escape from crisis was associated with an increase in mean telomere length (Figures 3C and S3D) and was

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1065

Figure 2. LIG3 Is Required for Cells to Escape a Telomere-Driven Crisis (A–G) Cell growth curves of multiple, independent single-cell clonal populations. Population doublings are shown from the point of single cell cloning. The period of crisis is indicated on each plot and clones that did not escape crisis are highlighted in red. (H) Summarizing the total number of PD achieved. Each symbol represents a single clone. Clones were stopped in culture once they had achieved greater than 85 PD. See also Figure S2.

accompanied by genomic rearrangements detected with array comparative genomic hybridization (aCGH; Figures 3G and S3E). Surviving clones continued in culture for over 85 PD at which point the cultures were terminated (Figure 2H). In stark contrast to the LIG4
/
or the LIG3+/
clones, none of the LIG3
/
:mL3 clones escaped crisis (n = 11), despite being kept in culture for up to 120 days, after which there were no cells remaining (Figure 2D). Telomere length remained constant during the crisis period, but the number of detectable telomeres decreased in the later stages of culture (Figure 3D). Telomere fusion was not detected in the LIG3- or LIG4-inactivated cells prior to the onset of crisis, indicating that these components do not directly modulate telomere stability. Telomere fusion was readily detected during crisis, irrespective of the presence or absence of LIG3 or LIG4, but was not detectable in cells that had escaped (Figures 3E and 3F). From the point of single

cell cloning to the first cell count at PD 22 (mean), the rate of cell growth of the LIG3
/
:mL3 HCT116 cells (1.7 PD/day) was indistinguishable from that of the HCT116 WT DN-hTERT cells (1.8 PD/day, p = 0.11). These cells therefore do not appear to be adversely affected by the DN-hTERT during the first 22 PDs of growth and thus the inability of these cells to escape a telomere-driven crisis is unlikely to be just a consequence of the additive effect of the DN-hTERT, together with a DNA repair defect. Taken together, these data are consistent with a role for LIG3, but not LIG4, in mediating the escape from crisis. To further characterize the role of LIG3 in the escape from crisis, we complemented LIG3
/
:mL3 cells with a cDNA encoding a nuclear-targeted WT LIG3 (Oh et al., 2014). DN-hTERT was expressed in these cells and this led to telomerase inhibition and gradual telomere erosion (Figures S1A, S1C, and 4A). At approximately PD 19, all of these clones (n = 15) entered a period of

1066 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

Figure 3. Telomere Fusions at the Point of Crisis Are Detected in Both LIG3
/
:mL3 and LIG4
/
Cells

(A and B) Growth curves of single clonal populations of LIG4
/
or LIG3
/
:mL3 cells, depicting the point of crisis and analysis points. (C and D) 17p STELA of the same clonal populations. Dotted red line indicates the lower limit of telomere erosion. (E and F) Fusion analysis using the XpYp, 17p, and 21q family telomere primers (Letsolo et al., 2010), products were detected with a 17p telomere-adjacent hybridization probe. (G) aCGH of DNA from the LIG4
/
clone before and after crisis as indicated in (A). See also Figure S3.

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1067

Figure 4. Telomere Erosion, Fusion, and the Induction of and Escape from Crisis in LIG3
/
:mL3 Complemented with Nuclear LIG3 (A) Cell growth curves of two single cell clonal populations, plotting PD from the point of single cloning. PD points analyzed with STELA and for fusions are indicated and the period of crisis is shown in red. The PD from the point of single cell cloning is shown above, the mean and SD of the telomere length profiles are shown below. (B) Telomere fusion analysis at the PD points as indicated. (C) aCGH on DNA samples taken from LIG3
/
:mL3 complemented with nuclear LIG3 cells before and after crisis. See also Figure S4.

resulted in a slower rate of growth (p < 0.0001, F-test; Figures S2B and S2C). The escape from crisis was accompanied by an increase in telomere length, and the cessation of telomere fusions and the occurrence of genomic rearrangements (Figures 4A–4C, S4B, and S4C). We detected amplification of the hTERT locus in three of 10 clones analyzed with aCGH (4 WT, 4 LIG4
/
, and 2 LIG3
/
:mL3 nuclear LIG3 complemented cells); we therefore consider that amplification of this locus provides a mechanism for the upregulation of telomerase and the escape from crisis. Consistent with this hypothesis is the observation that the amplification of this region is detected in 30% tumors of the lung, cervix, and breast (Zhang et al., 2000).

crisis with slowed cell growth, increased apoptosis, and telomere fusions (Figures 2E, S4A, and 4B); these phenotypes were not observed in controls (Figure S2C). However, in total contrast to the LIG3
/
:mL3 cells, all the LIG3-complemented clones readily escaped crisis, continuing to over 85 PD (Figures 2E and 2H); the escape from crisis took longer, but we attributed this to the dual puromycin and hygromycin selection, which

LIG3 Is Required for the Escape from Crisis in Other Cell Types We tested the cell specificity of these observations in HCT116 cells by undertaking similar experiments in the breast cancer cell line MCF7. We used an shRNA against LIG3 (provided by K. Caldecott) that resulted in a >90% reduction in LIG3 expression (Figure S5A); knockdown of LIG3 did not affect the rate of cell growth (Figure S5B). The expression of DN-hTERT in MCF7 cells resulted in telomere erosion, telomere fusion, and the induction of a crisis state (Figures 5A and 5B), from which all clones (n = 13) without the LIG3 shRNA readily escaped (Figure 5C). No crisis was observed in empty vectors control cultures (Figures S5C and S5D). Clones coexpressing the LIG3 shRNA and DN-hTERT also entered a crisis state that was accompanied by telomere erosion and fusion (Figures 5A and 5B), resulting in loss of viability, such that only seven LIG3

1068 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

Figure 5. LIG3 Is Required for the Escape from Crisis in MCF7 Cells (A) STELA of the 17p telomere. Two clones that escaped crisis are shown; the mean and SD of the telomere length profiles are detailed below. (B) Telomere fusion analysis at the PD points as indicated, showing control cells, cells in crisis, and those that escaped. Neo refers to the empty vector control for the shLIG3 and puro refers to the empty vector control for the DN-hTERT. (C and D) Cell growth curves of multiple independent single cell clonal populations. PDs are shown from the point of single cell cloning. Clones that did not escape crisis are highlighted in red. See also Figure S5.

shRNA clones could be isolated. After 120 days in culture, three clones emerged from crisis (Figure 5D); all of these clones had reestablished LIG3 expression (Figure S5A); these clones also showed telomere elongation and the cessation of telomere fusion (Figures 5A and 5B). These data further confirm the role of LIG3 in the escape from a telomere-driven crisis and demonstrate the reproducibility of our observations in cell types other than HCT116. Escape from Crisis Depends on the LIG3 BRCT Domain LIG3 is stabilized by its interaction with XRCC1 via the C-terminal BRCT domains of both proteins (Caldecott et al., 1994; Dulic et al., 2001). We therefore tested the requirement for this interaction in mediating the LIG3-dependent escape from a telomeredriven crisis. We complemented the HCT116 LIG3
/
:mL3 cells with nuclear-targeted LIG3 cDNAs encoding either a deletion in the LIG3 BRCT domain or a S847A point mutation (Cuneo et al., 2011), both of which are predicted to disrupt the interaction of LIG3 with XRCC1. The expression of the DN-hTERT in these cells, but not in empty vector controls, led to telomere erosion, telomere fusion, and the induction of crisis (Figures 6A, 6B, S2D and S2E). However, none of the clones containing the LIG3 S847A mutation (n = 13) or the LIG3 DBRCT mutation (n = 7) were able to escape crisis (Figures 2F–2H). These data further confirm the requirement for LIG3 in the escape from a

telomere-driven crisis and demonstrate that the BRCT domain is required for this process. Cells that Cannot Escape Crisis Exhibit a Higher Frequency of Interchromosomal Telomere Fusion Events We next considered whether there were specific features of the telomere fusions ligated by LIG3 or LIG4 that conferred a selective advantage or disadvantage, respectively, during the escape from crisis. We conducted a comprehensive characterization of telomere fusion events in these strains at the molecular level. By focusing our assays on single chromosome ends, we observed intrachromosomal fusion events, consistent with sister chromatid fusion, in both the LIG3
/
:mL3 and LIG4
/
cells (Figure 7A). However the proportions of inter- versus intrachromosomal fusion events was significantly increased in the LIG3
/
:mL3 cells (23%) compared to LIG4
/
cells (2%; p = 0.0006, c2 test; Figures 7B and S6A). A similar analysis of a fibroblast strain (MRC5) expressing HPVE6E7 that undergoes crisis in culture (Capper et al., 2007) but does not escape crisis, also showed a large skew in favor of interchromosomal fusions (Figures 7A and 7B), and these cells also displayed lower levels of LIG3 compared to HCT116 (Figure S6B). Together, these data indicate that one feature of cells that cannot escape a telomere-driven crisis is that they exhibit a greater proportion

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1069

Figure 6. A Functional LIG3 BRCT Domain Is Required for the Escape from Crisis (A) STELA of the 17p telomere is shown above. The PD from the point of single cell cloning is shown above, the mean and SD of the telomere length profiles are shown below. (B) Telomere fusion analysis at the PD points as indicated.

of interchromosomal fusion events compared to intrachromosomal events. Telomere Fusion Displays Characteristics of Both A- and C-NHEJ We next characterized the DNA sequences spanning telomere fusion junctions in LIG3
/
:mL3, LIG4
/
, WT HCT116, and MRC5HPVE6E7 cells (Figure 7C). Intrachromosomal fusion events create palindromic sequences, which make identification of the fusion site problematic and thus despite these events being more numerous, we were only able to characterize six such events. Therefore, our subsequent analysis was focused on interchromosomal fusions, and because these were rare in LIG4
/
cells, we used both direct next generation sequencing (Illumina) of telomere fusion products, as well as Sanger sequencing, to obtain sufficient events for analysis. We observed distinct distributions of the fusion events, with a bias

toward fusions containing telomere repeats on either side of the fusion point in the LIG3
/
:mL3 cells compared to LIG4
/
cells, where fusion was biased toward subtelomeric deletion (p < 0.0001, c2 test; Figure 7D). The WT HCT116 and MRC5HPVE6E7 cells displayed an identical profile (p = 0.58, c2 test) that was intermediate between the fusion profiles of the LIG3
/
:mL3 and LIG4
/
cells (LIG3
/
:mL3:WT, p < 0.05; LIG4
/
:WT, p = 0.0002, c2 test; Figure 7D). There was a slight, but not significant, increase in the extent of subtelomeric sequence deletion in LIG4
/
compared to LIG3
/
:mL3 cells (Figure 7E), indicating that, in contrast to nontelomeric loci (Lieber, 2008), both A- and CNHEJ processes are equally error prone at telomeres. We observed a decrease in the extent of microhomology use at the fusion points in LIG3
/
:mL3 cells (mean 1.6 nts) compared to LIG4
/
cells (mean 2.8 nts, p = 0.0001, t test; Figure 7F). Microhomology use in the WT HCT116 cells (1.9 nts) was intermediate between LIG3
/
:mL3 and LIG4
/
cells. The reduction in mean microhomology use in the LIG3
/
:mL3 cells was also accompanied by a reduction in the proportion of fusion events that did not use microhomology at all: in LIG4
/
cells, 4.6% displayed no microhomology compared to 26% in LIG3
/
:mL3 cells (p < 0.0001, Fisher’s exact test; Figure 7F). The LIG3
/
:mL3 and MRC5HPVE6E7 cells displayed a similar frequency of these events (p = 0.64; Figure 7F). Given the distinct profiles of telomere fusion observed, we considered that there may be a bias toward the use of microhomology depending on the type of fusion event in which it occurred. We therefore classified the microhomology data obtained from MRC5E6E7 cells (for which we had the largest data set) into the different fusion classes. In events that involved the telomere, there was a clear reduction in overall microhomology use, which was accompanied by an increase in events that displayed no microhomology (p < 0.0001; Figure 7G). Taken together, these data are consistent with both C- and A-NHEJ operating to catalyze the fusion of short dysfunctional telomeres in human cells and indicate a bias toward C-NHEJ catalyzing

1070 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

fusions within the telomere repeat array and A-NHEJ favoring fusion within the subtelomeric sequences. DISCUSSION Our data demonstrate a role for both C-NHEJ and A-NHEJ in distinct processes that catalyze the fusion of short dysfunctional telomeres. Whereas both C- and A-NHEJ can lead to inter- or intrachromosomal telomere fusion, interchromosomal events predominate with LIG4-dependent C-NHEJ; these events use less microhomology and are biased toward fusion within the telomere repeat regions. In contrast, LIG3-dependent A-NHEJ results in interchromosomal events that use microhomology at the fusion points, but are significantly less common relative to intrachromosomal events. Importantly, our data also show that the activity of LIG3 is required for cells to escape a telomere-driven crisis by clonal evolution. C-NHEJ versus A-NHEJ at Short Dysfunctional Telomeres A key observation from our data is that, irrespective of the lack of LIG3 or LIG4, short dysfunctional telomeres can readily be subjected to fusion. This is in contrast to the situation following the abrogation of TRF2 function at telomeres, where fusion is dependent on C-NHEJ involving ATM, Ku, 53BP1, and LIG4 (Celli et al., 2006; Rai et al., 2010; Smogorzewska et al., 2002). It is therefore apparent that telomeres rendered dysfunctional by replicative telomere erosion are processed differently; a situation exemplified in the telomerase-deficient mice where fusion of short telomeres was independent of DNA-PKcs, LIG4, and 53BP1 (Maser et al., 2007; Rai et al., 2010). Consistent with the use of two different pathways in mediating the fusion of short dysfunctional telomeres, our data revealed distinct profiles associated with fusion in the absence of LIG3 or LIG4. We observed an increase in microhomology at the fusion junctions in the absence of LIG4 compared to LIG3, with WT cells displaying an approximately intermediate phenotype. We also observed changes in the distribution of fusion events between those occurring within the telomere repeat region or in the telomere-adjacent sequences. In the absence of LIG3, telomere fusion was biased toward an increase in events occurring within the telomere repeat array, whereas in the absence of LIG4, fusion was biased toward deletion within the telomere-adjacent sequences. Again, WT cells displayed an intermediate distribution. These data indicate that LIG3-mediated fusion is subtly more mutagenic and results in the resection of the telomere repeat array into the telomere adjacent DNA. This is consistent with higher levels of resection known to occur in the context of A-NHEJ compared to C-NHEJ (Symington and Gautier, 2011). However, fusion was still detected within the telomere-adjacent DNA in the absence of LIG3, suggesting that LIG4-dependent C-NHEJ is intrinsically more error prone at telomeric loci presumably resulting from increased end-resection activity. These data are broadly consistent with the processing of experimentally induced I-SceI breaks at telomeric loci, which has also been shown to be more error prone, indicating that telomeres are processed differently from nontelomeric loci (Zschenker et al., 2009). Why the processing of short-dysfunctional telomeres is more error prone, compared

to nontelomeric double-stranded breaks, is not clear. Our data, however, indicate that this is not simply a consequence of the preferential utilization of error-prone A-NHEJ, over the less error-prone C-NHEJ pathway. One possible explanation may be related to the unique terminal structure of a telomere. At the point at which fusion is detected in human cells following replicative erosion, the telomeres are extremely short, with a mean number of just six TTAGGG repeats adjacent to the fusion point (Capper et al., 2007; Letsolo et al., 2010). Telomeres of this length are unlikely to be capable of binding sufficient TRF1 or TRF2 homodimers to form a functional shelterin complex and confer the end capping function of telomeres (Capper et al., 2007). However, telomeres of this size still display a 30 overhang at the terminus (Baird et al., 2003), the mean length of which in human cells has been estimated at 100-400 nts (Chai et al., 2006). The presence of a large 30 -overhang, that is capable of forming G-quadruplex structures (Gavathiotis et al., 2003), renders the structure of a short dysfunctional telomere distinct from double-stranded DNA breaks observed elsewhere in the genome, for example following RAG cleavage or exposure to ionizing radiation (van Gent et al., 2001). Consistent with this view, the length of a 30 overhang at nontelomeric double-stranded DSBs influences both the rate of repair and the pathway utilized: with longer overhangs, that bind less Ku dimers, being repaired slower and repair favored by Ku-independent A-NHEJ (Daley and Wilson, 2005; Dynan and Yoo, 1998). We therefore speculate that the apparent use of error-prone end-joining pathways, including A-NHEJ, may be related to the unique structure of the 30 overhang of a short dysfunctional telomere that is denuded of the primary components of the shelterin complex. Whatever the underlying mechanism, it is clear that both A- and C-NHEJ are utilized for the fusion of short dysfunctional telomeres. LIG3 Is Required for the Escape from a Telomere-Driven Crisis The cellular crisis induced by telomere erosion provides a highly selective environment in which telomere fusion can, by itself, only provide a temporary lifespan extension to the cell in which it occurs. However, by initiating widespread genomic instability, the appropriate genomic rearrangements that confer immortality can be selected for. We observed that the ability of cells to escape this selective environment depended on LIG3. The inability of LIG3
/
:mL3 cells to escape a telomere-driven crisis was completely rescued by complementation with WT LIG3, whereas, LIG3 constructs with mutations within the BRCT domain completely failed to rescue the phenotype. In addition, half of the clones derived from LIG3+/
heterozygotes escaped crisis and half failed to escape, indicating that the ability of cells to escape from crisis induced by gradual telomere shortening, might be sensitive to the dosage of LIG3 expressed in a cell. This view is also consistent with our data from the shRNA knockdown of MCF7 cells and the MRC5HPVE6E7 cells both of which display low levels of LIG3 and do not readily escape crisis. These key observations, together with the ability of LIG4
/
cells to escape a telomere-driven crisis, strongly implicate the LIG3dependent A-NHEJ pathway in facilitating the escape from a telomere-driven crisis.

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1071

Figure 7. Intrachromosomal Telomere Fusion Is Common in LIG3
/
:mL3 and LIG4
/
Cells but Interchromosomal Telomere Fusion Is Rare in LIG4
/
Cells (A) XpYp:17p and 17p:17p fusion analysis (Capper et al., 2007). Fusion fragments are amplified with primer combinations as detailed above and detected with probe as detailed on the right. Products in XpYp:17p reactions detected with both probes are consistent with interchromosomal events (examples arrowed in red). Products in both the XpYp:17p and 17p reactions, detected only with the 17p probe, only are consistent with 17p sister chromatid fusion events (examples arrowed in green). (B) Horizontal stacked bar chart depicting the proportions of 17p sister chromatid fusion to XpYp:17p fusion events; also depicted as ratios on the right. Error bars represent SE. (C) Examples of XpYp and 17p fusion junction sequences. Homology at the fusion point is indicated as bold underlined. Deletion sizes of the participating telomeres are taken from the beginning of the telomere repeat array.

(legend continued on next page)

1072 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

XRCC1 is a scaffold protein that interacts with numerous DNA repair processes (Caldecott, 2003); this includes an interaction with LIG3, via BRCT domains in both proteins (Caldecott et al., 1994). This interaction controls the stability of LIG3 and because XRCC1 contains a nuclear localization signal, but LIG3 does not, it may be required for the translocation of LIG3 into the nucleus (Masson et al., 1998). We demonstrated a requirement for a functional LIG3 BRCT domain for the escape from crisis and thereby implicated the interaction with XRCC1 in this process. Curiously, however, XRCC1 may not be required for A-NHEJ, or at least not in the context of A-NHEJ-dependent class switch recombination (Boboila et al., 2012); it therefore remains possible that additional proteins may interact with, and stabilize, LIG3 via the BRCT domain, thereby providing the activity required for the escape from crisis. Mechanism of Escape from Crisis We observed changes in the relative proportions of intra- to interchromosomal telomere fusion events; in the absence of LIG4, telomere fusion was biased toward intrachromosomal events (52:1), but in the absence of LIG3 there was an increase in the proportion of interchromosomal events (3:1; Figure 7B). This observation may appear to be in contrast to the apparent role of the A-NHEJ pathway at nontelomeric double-stranded DNA breaks, where it can catalyze the formation of translocations at specific break sites in the absence of C-NHEJ (Simsek et al., 2011a). However it is also apparent that in normal situations, when both C- and A-NHEJ are functional, that the majority of translocations are in fact mediated by C-NHEJ (Brunet et al., 2009; Bunting and Nussenzweig, 2013). The role of C-NHEJ in catalyzing interchromosomal telomere fusions is consistent with the roles of 53BP1 and LIG4-dependent C-NHEJ in mediating long-range end-joining during class-switch recombination (Difilippantonio et al., 2008) and telomere fusion following abrogation of TRF2 activity (Dimitrova et al., 2008). Our data imply that the operation of the A-NHEJ pathway on short dysfunctional telomeres has a suppressive effect on the longrange end-joining activity catalyzed by the C-NHEJ pathway. Cells that were capable of escaping crisis displayed lower levels of interchromosomal, compared to intrachromosomal, telomere fusion events. We therefore consider that the relative balance between these events dictates the ability of cells to escape crisis and that this is, in turn, influenced by the relative activities of the A- and C- NHEJ pathways operating at short dysfunctional telomeres (Figure S6C). Consistent with this hypothesis, sister-chromatid telomere fusion events predominate in experimental systems that use the selection of marker genes integrated in telomereadjacent sequences to identify telomeric mutation events (Lo et al., 2002; Murnane, 2012; Sabatier et al., 2005). Sister chromatid telomere fusion events can, via breakagefusion-bridge cycles, drive the formation of inverted duplications

and the localized amplification or deletion of the chromosome arm of the fused telomere (Murnane, 2012). These types of events are common in many tumor types, including mouse models of telomere dysfunction (Artandi et al., 2000) and in human tumors (Campbell et al., 2010; Tanaka et al., 2005). The escape from the telomere-driven crisis described here also accompanied by genomic rearrangements that resulted in copy number changes at the chromosomal termini: these are consistent with sister chromatid telomere fusion events initiating breakage-fusion-bridge cycles. Consistent with the possible key role of A-NHEJ mediating fusion and the escape from crisis, LIG3 is overexpressed in many tumor cell lines and is associated with increased A-NHEJ activity (Chen et al., 2008; Sallmyr et al., 2008; Tobin et al., 2012). We propose that sister chromatid fusions, as opposed to interchromosomal fusions, can create localized copy number changes at specific chromosome ends, creating the appropriate genetic alterations that initiate clonal evolution and the escape from crisis. The differences in the proportions of inter- compared to intrachromosomal fusion are striking and significant; therefore, we consider that the relative mutational impact of these processes may account for the differences in the ability of these cells to escape crisis. However, alternative explanations for our observations could be envisaged; for example, because the LIG3/ XRCC1 complex plays a role in the short-patch base excision repair pathway (Ellenberger and Tomkinson, 2008), a deficiency in LIG3 together with telomere dysfunction may have an additive effect, rendering the cells more sensitive to telomere dysfunction. However, these cells, as well as LIG3 mouse KO cells (Oh et al., 2014; Simsek et al., 2011b), are not sensitive to the alkylating agent methyl methanesulphonate, indicating no overt BER defect. Alternatively, LIG3 may, via an unspecified mechanism, control gene amplification and deletion and thus LIG3-deficient cells may be unable to amplify the hTERT locus or delete the inserted DN-hTERT. Ultimately, further experimentation, including a full characterization of the mutation events that facilitate the escape crisis, is required to define the exact mechanism. In conclusion, our data show that the activity of LIG3, dependent on a functional BRCT domain, confers a selective advantage to cells undergoing a telomere-driven crisis that facilitates clonal evolution and the escape from crisis, which is a key stage during the progression of many tumor types. Therefore, the ANHEJ pathway may represent a promising target for therapeutic intervention in the early stages of tumorigenesis.

EXPERIMENTAL PROCEDURES Cell Culture and Analysis HCT116 knockout cells and derivative complemented versions were generated using recombinant adeno-associated virus-mediated gene targeting

(D) Histograms showing the distribution of telomere fusion events, in the various cell strains analyzed. Tel-Tel, telomere repeats on each side of the fusion event; Del-Tel, subtelomeric deletion on one side and telomere repeats on the other side; Del-Del, subtelomeric deletion on both sides of the fusion event. Error bars represent SE. (E) Scatter plot depicting the extent of subtelomeric deletion at XpYp and 17p telomeres; deletion is taken from the start of the telomere repeat array. The 0 bp deletion represents telomeres that have fused within the telomere repeat array (mean ± 95% CI). (F) Scatter plots showing the extent of homology (nucleotides) at the fusion point (mean ± 95% CI). (G) Scatter plots displaying homology at fusion breakpoints in MRC5E6E7 cells, by fusion type (mean ± 95% CI). See also Figure S6.

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1073

as described (Oh et al., 2013, 2014). LIG3 was knocked down in MCF7 cells using a pSUPER shRNA derivative (Oligoengine) containing the LIG3 specific 19-mer GGATGTGAAGGGTACATAT (provided by Professor K. Caldecott, University of Sussex). To initiate telomere erosion, cells were transduced with amphotropic retroviral vectors containing a DN-hTERT cDNA (Hahn et al., 1999) as described (Preto et al., 2004). Apoptosis was assessed by Annexin V staining with the Annexin V-FITC Apoptosis Detection Kit (eBioscience) and cell cycle analysis using propidium iodide, both of which were quantified with an AccuriC6 flow cytometer (Becton Dickson). Cell images were taken at 403 magnification with phase contrast microscopy using a Zeiss Axiovert S100 TV. STELA, Telomere Fusion Assay, and TRAP For telomere length analysis, we used the STELA protocol (Capper et al., 2007). Telomere fusions were analyzed with single-molecule PCR using either the two-telomere protocol (Capper et al., 2007) or multiple-telomere protocol (Letsolo et al., 2010) as indicated. Telomere fusions were characterized following reamplification and Sanger sequencing as described (Capper et al., 2007) or directly using Next Generation Sequencing technology carried out using the HiSeq 2000 platform (Illumina) by the Beijing Genome Institute. Telomerase activity was detected using the TRAPeze XL Telomerase detection kit (Chemicon International). Next Generation Sequencing Bioinformatics Illumina HiSeq paired-end (23 90 bp) fastq data sets were trimmed and filtered with Trimmomatic (version 0.30; Lohse et al., 2012) to remove primers and low quality bases and reads. Resulting reads were then mapped to a bespoke reference constructed from telomeric regions 17p and XpYp using the MEM algorithm of BWA (version 0.7.4; Heng, 2013) with -M flag deployed to mark secondary mappings, which were then removed using SAMtools (version 0.1.19; Li et al., 2009) using the view -F 256 command. MergeSamFiles.jar, part of Picard toolkit and API (version 1.102; http://picard.sourceforge.net), was then used to merge the resultant SAM files, and sorted, indexed BAMs were created from these using SAMtools. Having mapped the data, an in-house Java script, based on the Picard API, was deployed to filter for paired-end reads that straddled the 17p and XpYp telomeric regions and thus likely to describe interchromosomal fusion events. SAMtools view -F48 was then used to filter for only those paired-end reads that mapped with the correct orientation with respect to the telomeric reference used. Sequence tags that straddled fusion events were identified by using a combination of SAMtools and command-line perl to filter for any mapped tag, in a fusion event straddling pair, that had been heavily (R10 bp) softtrimmed at the 30 end (which would be how the BWA mapper would have to respond to such tags within our data sets). Sequences so identified were then de novo assembled with Velvet (version 1.2.09; Zerbino and Birney, 2008). Resultant contigs and unassembled reads, were blastn mapped (parameter settings -G 2 -E 1 -F F -m 8; Altschul et al., 1990) to our bespoke reference to obtain positional information that was summarized in a spreadsheet. Array-CGH Analysis Array CGH was carried out as a service by NimbleGen using the 123135 k array (HG18 WG CGH v3.1; Roche NimbleGen); copy number gains and losses were identified using the segMNT algorithm included in NimbleScan software. Statistical Analyses All statistical analyses were performed using GraphPad Prism 6 and all statistical tests were two-sided. The Student’s t test and Chi-square test were used; a p value of less than 0.05 was considered statistically significant. ACCESSION NUMBERS The ArrayExpress accession number for the sequence data reported in this paper is _E-MTAB-2769.

SUPPLEMENTAL INFORMATION Supplemental Information includes six figures and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2014.07.007. AUTHOR CONTRIBUTIONS R.E.J., S.O., and J.W.G. carried out experimental work and edited the manuscript; J.Z., L.R., and N.H.H., carried out experimental work; K.E.A. provided bioinformatics support; K.L. carried out experimental work and edited the manuscript; E.A.H. and D.M.B. jointly conceived and supervised the study, analyzed the data, and wrote the manuscript. ACKNOWLEDGMENTS We thank Professor Keith Caldecott (University of Sussex) for providing the LIG3 shRNA. This work was funded by Cancer Research UK (C17199/ A13490) and the National Institute for Social Care and Health Research (NISCHR) Cancer Genetics Biomedical Research Unit. The work in the Hendrickson laboratory was funded, in part, by grants from the NIH (GM088351), the National Cancer Institute (CA15446), and a research contract from Horizon Discovery. Received: March 7, 2014 Revised: May 28, 2014 Accepted: July 10, 2014 Published: August 7, 2014 REFERENCES Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. (1990). Basic local alignment search tool. J. Mol. Biol. 215, 403–410. Artandi, S.E., Chang, S., Lee, S.L., Alson, S., Gottlieb, G.J., Chin, L., and DePinho, R.A. (2000). Telomere dysfunction promotes non-reciprocal translocations and epithelial cancers in mice. Nature 406, 641–645. Baird, D.M., Rowson, J., Wynford-Thomas, D., and Kipling, D. (2003). Extensive allelic variation and ultrashort telomeres in senescent human cells. Nat. Genet. 33, 203–207. Boboila, C., Yan, C., Wesemann, D.R., Jankovic, M., Wang, J.H., Manis, J., Nussenzweig, A., Nussenzweig, M., and Alt, F.W. (2010). Alternative endjoining catalyzes class switch recombination in the absence of both Ku70 and DNA ligase 4. J. Exp. Med. 207, 417–427. Boboila, C., Oksenych, V., Gostissa, M., Wang, J.H., Zha, S., Zhang, Y., Chai, H., Lee, C.S., Jankovic, M., Saez, L.M., et al. (2012). Robust chromosomal DNA repair via alternative end-joining in the absence of X-ray repair crosscomplementing protein 1 (XRCC1). Proc. Natl. Acad. Sci. USA 109, 2473– 2478. Boulton, S.J., and Jackson, S.P. (1996). Identification of a Saccharomyces cerevisiae Ku80 homologue: roles in DNA double strand break rejoining and in telomeric maintenance. Nucleic Acids Res. 24, 4639–4648. Brunet, E., Simsek, D., Tomishima, M., DeKelver, R., Choi, V.M., Gregory, P., Urnov, F., Weinstock, D.M., and Jasin, M. (2009). Chromosomal translocations induced at specified loci in human stem cells. Proc. Natl. Acad. Sci. USA 106, 10620–10625. Bunting, S.F., and Nussenzweig, A. (2013). End-joining, translocations and cancer. Nat. Rev. Cancer 13, 443–454. Caldecott, K.W. (2003). Protein-protein interactions during mammalian DNA single-strand break repair. Biochem. Soc. Trans. 31, 247–251. Caldecott, K.W., McKeown, C.K., Tucker, J.D., Ljungquist, S., and Thompson, L.H. (1994). An interaction between the mammalian DNA repair protein XRCC1 and DNA ligase III. Mol. Cell. Biol. 14, 68–76. Campbell, P.J., Yachida, S., Mudie, L.J., Stephens, P.J., Pleasance, E.D., Stebbings, L.A., Morsberger, L.A., Latimer, C., McLaren, S., Lin, M.L., et al. (2010). The patterns and dynamics of genomic instability in metastatic pancreatic cancer. Nature 467, 1109–1113.

1074 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors

Capper, R., Britt-Compton, B., Tankimanova, M., Rowson, J., Letsolo, B., Man, S., Haughton, M., and Baird, D.M. (2007). The nature of telomere fusion and a definition of the critical telomere length in human cells. Genes Dev. 21, 2495–2508. Celli, G.B., Denchi, E.L., and de Lange, T. (2006). Ku70 stimulates fusion of dysfunctional telomeres yet protects chromosome ends from homologous recombination. Nat. Cell Biol. 8, 885–890. Chai, W., Du, Q., Shay, J.W., and Wright, W.E. (2006). Human telomeres have different overhang sizes at leading versus lagging strands. Mol. Cell 21, 427–435. Chen, X., Zhong, S., Zhu, X., Dziegielewska, B., Ellenberger, T., Wilson, G.M., MacKerell, A.D., Jr., and Tomkinson, A.E. (2008). Rational design of human DNA ligase inhibitors that target cellular DNA replication and repair. Cancer Res. 68, 3169–3177. Counter, C.M., Avilion, A.A., LeFeuvre, C.E., Stewart, N.G., Greider, C.W., Harley, C.B., and Bacchetti, S. (1992). Telomere shortening associated with chromosome instability is arrested in immortal cells which express telomerase activity. EMBO J. 11, 1921–1929. Cuneo, M.J., Gabel, S.A., Krahn, J.M., Ricker, M.A., and London, R.E. (2011). The structural basis for partitioning of the XRCC1/DNA ligase III-a BRCT-mediated dimer complexes. Nucleic Acids Res. 39, 7816–7827. d’Adda di Fagagna, F., Reaper, P.M., Clay-Farrace, L., Fiegler, H., Carr, P., Von Zglinicki, T., Saretzki, G., Carter, N.P., and Jackson, S.P. (2003). A DNA damage checkpoint response in telomere-initiated senescence. Nature 426, 194–198. Daley, J.M., and Wilson, T.E. (2005). Rejoining of DNA double-strand breaks as a function of overhang length. Mol. Cell. Biol. 25, 896–906.

Harley, C.B., Futcher, A.B., and Greider, C.W. (1990). Telomeres shorten during ageing of human fibroblasts. Nature 345, 458–460. Hendrickson, E.A. (2008). Gene targeting in human somatic cells. In Sourcebook of Models for Biomedical Research, M. Conn, ed. (Totowa, NJ: Humana Press), pp. 509–525. Heng, L. (2013). Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv:1303.3997v2 [q-bio.GN]. Khan, I.F., Hirata, R.K., and Russell, D.W. (2011). AAV-mediated gene targeting methods for human cells. Nat. Protoc. 6, 482–501. Letsolo, B.T., Rowson, J., and Baird, D.M. (2010). Fusion of short telomeres in human cells is characterized by extensive deletion and microhomology, and can result in complex rearrangements. Nucleic Acids Res. 38, 1841–1852. Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., and Durbin, R.; 1000 Genome Project Data Processing Subgroup (2009). The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079. Lieber, M.R. (2008). The mechanism of human nonhomologous DNA end joining. J. Biol. Chem. 283, 1–5. Lin, T.T., Letsolo, B.T., Jones, R.E., Rowson, J., Pratt, G., Hewamana, S., Fegan, C., Pepper, C., and Baird, D.M. (2010). Telomere dysfunction and fusion during the progression of chronic lymphocytic leukemia: evidence for a telomere crisis. Blood 116, 1899–1907. Lo, A.W., Sabatier, L., Fouladi, B., Pottier, G., Ricoul, M., and Murnane, J.P. (2002). DNA amplification by breakage/fusion/bridge cycles initiated by spontaneous telomere loss in a human cancer cell line. Neoplasia 4, 531–538.

de Lange, T. (2005). Shelterin: the protein complex that shapes and safeguards human telomeres. Genes Dev. 19, 2100–2110.

Lohse, M., Bolger, A.M., Nagel, A., Fernie, A.R., Lunn, J.E., Stitt, M., and Usadel, B. (2012). RobiNA: a user-friendly, integrated software solution for RNASeq-based transcriptomics. Nucleic Acids Res. 40, W622–W627.

Difilippantonio, M.J., Petersen, S., Chen, H.T., Johnson, R., Jasin, M., Kanaar, R., Ried, T., and Nussenzweig, A. (2002). Evidence for replicative repair of DNA double-strand breaks leading to oncogenic translocation and gene amplification. J. Exp. Med. 196, 469–480.

Maser, R.S., Wong, K.K., Sahin, E., Xia, H., Naylor, M., Hedberg, H.M., Artandi, S.E., and DePinho, R.A. (2007). DNA-dependent protein kinase catalytic subunit is not required for dysfunctional telomere fusion and checkpoint response in the telomerase-deficient mouse. Mol. Cell. Biol. 27, 2253–2265.

Difilippantonio, S., Gapud, E., Wong, N., Huang, C.Y., Mahowald, G., Chen, H.T., Kruhlak, M.J., Callen, E., Livak, F., Nussenzweig, M.C., et al. (2008). 53BP1 facilitates long-range DNA end-joining during V(D)J recombination. Nature 456, 529–533.

Masson, M., Niedergang, C., Schreiber, V., Muller, S., Menissier-de Murcia, J., and de Murcia, G. (1998). XRCC1 is specifically associated with poly(ADPribose) polymerase and negatively regulates its activity following DNA damage. Mol. Cell. Biol. 18, 3563–3571.

Dimitrova, N., Chen, Y.C., Spector, D.L., and de Lange, T. (2008). 53BP1 promotes non-homologous end joining of telomeres by increasing chromatin mobility. Nature 456, 524–528.

Meeker, A.K., Hicks, J.L., Iacobuzio-Donahue, C.A., Montgomery, E.A., Westra, W.H., Chan, T.Y., Ronnett, B.M., and De Marzo, A.M. (2004). Telomere length abnormalities occur early in the initiation of epithelial carcinogenesis. Clin. Cancer Res. 10, 3317–3326.

Dulic, A., Bates, P.A., Zhang, X., Martin, S.R., Freemont, P.S., Lindahl, T., and Barnes, D.E. (2001). BRCT domain interactions in the heterodimeric DNA repair protein XRCC1-DNA ligase III. Biochemistry 40, 5906–5913. Dynan, W.S., and Yoo, S. (1998). Interaction of Ku protein and DNA-dependent protein kinase catalytic subunit with nucleic acids. Nucleic Acids Res. 26, 1551–1559. Ellenberger, T., and Tomkinson, A.E. (2008). Eukaryotic DNA ligases: structural and functional insights. Annu. Rev. Biochem. 77, 313–338. Feldmann, E., Schmiemann, V., Goedecke, W., Reichenberger, S., and Pfeiffer, P. (2000). DNA double-strand break repair in cell-free extracts from Ku80-deficient cells: implications for Ku serving as an alignment factor in non-homologous DNA end joining. Nucleic Acids Res. 28, 2585–2596.

Murnane, J.P. (2012). Telomere dysfunction and chromosome instability. Mutat. Res. 730, 28–36. Oh, S., Wang, Y., Zimbric, J., and Hendrickson, E.A. (2013). Human LIGIV is synthetically lethal with the loss of Rad54B-dependent recombination and is required for certain chromosome fusion events induced by telomere dysfunction. Nucleic Acids Res. 41, 1734–1749. Oh, S., Harvey, A., Zimbric, J., Wang, Y., Nguyen, T., Jackson, P.J., and Hendrickson, E.A. (2014). DNA ligase III and DNA ligase IV carry out genetically distinct forms of end joining in human somatic cells. DNA Repair (Amst.) http://dx.doi.org/10.1016/j.dnarep.2014.04.015.

Gavathiotis, E., Heald, R.A., Stevens, M.F., and Searle, M.S. (2003). Drug recognition and stabilisation of the parallel-stranded DNA quadruplex d(TTAGGGT)4 containing the human telomeric repeat. J. Mol. Biol. 334, 25–36.

Preto, A., Singhrao, S.K., Haughton, M.F., Kipling, D., Wynford-Thomas, D., and Jones, C.J. (2004). Telomere erosion triggers growth arrest but not cell death in human cancer cells retaining wild-type p53: implications for antitelomerase therapy. Oncogene 23, 4136–4145.

Guirouilh-Barbat, J., Huck, S., Bertrand, P., Pirzio, L., Desmaze, C., Sabatier, L., and Lopez, B.S. (2004). Impact of the KU80 pathway on NHEJ-induced genome rearrangements in mammalian cells. Mol. Cell 14, 611–623.

Rai, R., Zheng, H., He, H., Luo, Y., Multani, A., Carpenter, P.B., and Chang, S. (2010). The function of classical and alternative non-homologous end-joining pathways in the fusion of dysfunctional telomeres. EMBO J. 29, 2598–2610.

Hahn, W.C., Stewart, S.A., Brooks, M.W., York, S.G., Eaton, E., Kurachi, A., Beijersbergen, R.L., Knoll, J.H., Meyerson, M., and Weinberg, R.A. (1999). Inhibition of telomerase limits the growth of human cancer cells. Nat. Med. 5, 1164–1170.

Roger, L., Jones, R.E., Heppel, N.H., Williams, G.T., Sampson, J.R., and Baird, D.M. (2013). Extensive telomere erosion in the initiation of colorectal adenomas and its association with chromosomal instability. J. Natl. Cancer Inst. 105, 1202–1211.

Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors 1075

Sabatier, L., Ricoul, M., Pottier, G., and Murnane, J.P. (2005). The loss of a single telomere can result in instability of multiple chromosomes in a human tumor cell line. Mol. Cancer Res. 3, 139–150. Sallmyr, A., Tomkinson, A.E., and Rassool, F.V. (2008). Up-regulation of WRN and DNA ligase IIIalpha in chronic myeloid leukemia: consequences for the repair of DNA double-strand breaks. Blood 112, 1413–1423. Sfeir, A., and de Lange, T. (2012). Removal of shelterin reveals the telomere end-protection problem. Science 336, 593–597. Shih, I.M., Zhou, W., Goodman, S.N., Lengauer, C., Kinzler, K.W., and Vogelstein, B. (2001). Evidence that genetic instability occurs at an early stage of colorectal tumorigenesis. Cancer Res. 61, 818–822. Simsek, D., and Jasin, M. (2011). DNA ligase III: a spotty presence in eukaryotes, but an essential function where tested. Cell Cycle 10, 3636–3644. Simsek, D., Brunet, E., Wong, S.Y., Katyal, S., Gao, Y., McKinnon, P.J., Lou, J., Zhang, L., Li, J., Rebar, E.J., et al. (2011a). DNA ligase III promotes alternative nonhomologous end-joining during chromosomal translocation formation. PLoS Genet. 7, e1002080. Simsek, D., Furda, A., Gao, Y., Artus, J., Brunet, E., Hadjantonakis, A.K., Van Houten, B., Shuman, S., McKinnon, P.J., and Jasin, M. (2011b). Crucial role for DNA ligase III in mitochondria but not in Xrcc1-dependent repair. Nature 471, 245–248. Smogorzewska, A., Karlseder, J., Holtgreve-Grez, H., Jauch, A., and de Lange, T. (2002). DNA ligase IV-dependent NHEJ of deprotected mammalian telomeres in G1 and G2. Curr. Biol. 12, 1635–1644. Symington, L.S., and Gautier, J. (2011). Double-strand break end resection and repair pathway choice. Annu. Rev. Genet. 45, 247–271. Tanaka, H., Bergstrom, D.A., Yao, M.C., and Tapscott, S.J. (2005). Widespread and nonrandom distribution of DNA palindromes in cancer cells provides a structural platform for subsequent gene amplification. Nat. Genet. 37, 320–327. Tankimanova, M., Capper, R., Letsolo, B.T., Rowson, J., Jones, R.E., BrittCompton, B., Taylor, A.M., and Baird, D.M. (2012). Mre11 modulates the fidel-

ity of fusion between short telomeres in human cells. Nucleic Acids Res. 40, 2518–2526. Tobin, L.A., Robert, C., Nagaria, P., Chumsri, S., Twaddell, W., Ioffe, O.B., Greco, G.E., Brodie, A.H., Tomkinson, A.E., and Rassool, F.V. (2012). Targeting abnormal DNA repair in therapy-resistant breast cancers. Mol. Cancer Res. 10, 96–107. van Gent, D.C., Hoeijmakers, J.H., and Kanaar, R. (2001). Chromosomal stability and the DNA double-stranded break connection. Nat. Rev. Genet. 2, 196–206. van Steensel, B., Smogorzewska, A., and de Lange, T. (1998). TRF2 protects human telomeres from end-to-end fusions. Cell 92, 401–413. Yan, C.T., Boboila, C., Souza, E.K., Franco, S., Hickernell, T.R., Murphy, M., Gumaste, S., Geyer, M., Zarrin, A.A., Manis, J.P., et al. (2007). IgH class switching and translocations use a robust non-classical end-joining pathway. Nature 449, 478–482. Zerbino, D.R., and Birney, E. (2008). Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 18, 821–829. Zhang, Y., and Jasin, M. (2011). An essential role for CtIP in chromosomal translocation formation through an alternative end-joining pathway. Nat. Struct. Mol. Biol. 18, 80–84. Zhang, A., Zheng, C., Lindvall, C., Hou, M., Ekedahl, J., Lewensohn, R., Yan, Z., Yang, X., Henriksson, M., Blennow, E., et al. (2000). Frequent amplification of the telomerase reverse transcriptase gene in human tumors. Cancer Res. 60, 6230–6235. Zhu, C., Mills, K.D., Ferguson, D.O., Lee, C., Manis, J., Fleming, J., Gao, Y., Morton, C.C., and Alt, F.W. (2002). Unrepaired DNA breaks in p53-deficient cells lead to oncogenic gene amplification subsequent to translocations. Cell 109, 811–821. Zschenker, O., Kulkarni, A., Miller, D., Reynolds, G.E., Granger-Locatelli, M., Pottier, G., Sabatier, L., and Murnane, J.P. (2009). Increased sensitivity of subtelomeric regions to DNA double-strand breaks in a human cancer cell line. DNA Repair (Amst.) 8, 886–900.

1076 Cell Reports 8, 1063–1076, August 21, 2014 ª2014 The Authors