Role of interaction of XPF with RPA in nucleotide excision repair

Nucleotide excision repair (NER) is a very important defense system against various types of DNA damage, and it is necessary for maintaining genomic stability. The molecular mechanism of NER has been studied in considerable detail, and it has been shown that proper protein-protein interactions among NER factors are critical for efficient repair. A structure-specific endonuclease, XPF-ERCC1, which makes the 5′ incision in NER, was shown to interact with a single-stranded DNA binding protein, RPA. However, the biological significance of this interaction was not studied in detail. We used the yeast two-hybrid assay to determine that XPF interacts with the p70 subunit of RPA. To further examine the role of this XPF-p70 interaction, we isolated a p70-interaction-deficient mutant form of XPF that contains a single amino acid substitution in the N-terminus of XPF by the reverse yeast two-hybrid assay using randomly mutagenized XPF. The biochemical properties of this RPA-interaction-deficient mutant XPF-ERCC1 are very similar to those of wild-type XPF-ERCC1 in vitro. Interestingly, expression of this mutated form of XPF in the XPF-deficient Chinese hamster ovary cell line, UV41, only partially restores NER activity and UV resistance in vivo compared to wild-type XPF. We discovered that the RPA-interaction-deficient XPF is not localized in nuclei and the mislocalization of XPF-ERCC1 prevents the complex from functioning in NER.     

Differential processing of UV mimetic and interstrand crosslink damage by XPF cell extracts

We have recently developed a mammalian cell free assay in which interstrand crosslinks induce DNA synthesis in both damaged and undamaged plasmids co-incubated in the same extract. We have also shown using hamster mutants that both ERCC1 and XPF are required for the observed incorporation. Here, we show that extracts from an XPF patient cell line differentially process UV mimetic damage and interstrand crosslinks in vitro. XPF extracts are highly defective in the stimulation of repair synthesis by N-acetoxy-N- acetylaminofluorene, but are proficient in the stimulation of DNA synthesis by psoralen interstrand crosslinks. In addition, we show that extracts from the hamster UV140 mutant, which has high UV sensitivity, but moderate mitomycin C sensitivity, are similar in both assays to XPF cell extracts. These findings support the hypothesis that the activities of XPF in nucleotide excision repair (NER) and crosslink repair are separable, and that mutations in XPF patients result in the abolition of NER, but not recombinational repair pathways, which are likely to be essential as has been observed in ERCC1 homozygous –/– mice.     

Involvement of nucleotide excision repair (NER) system in repair of mono ADP-ribosylated dG adducts produced by pierisin-1, a cytotoxic protein from cabbage butterfly

Pierisin-1, a cytotoxic protein from the cabbage butterfly (Pieris rapae), induces apoptosis in mammalian cell lines. Binding of its C-terminal region to glycosphingolipid Gb3 and Gb4 receptors on cell membrane is necessary for incorporation into cells, while the N-terminal polypeptide catalyzes transfer of the ADP-ribose moiety of NAD at N2 of dG in DNA. Resulting DNA adducts cause mutation if they are present at low levels. If the DNA damage is more severe, the cells undergo apoptosis. In the present study, we examined the repair system for ADP-ribosylated dG adducts using nucleotide excision repair (NER) mutants of Chinese hamster ovary (CHO) cell lines. Pierisin-1 showed cytotoxic effects in all cases: IC50 values of them were; 650 ng/ml for AA8 (wild), 230 ng/ml for UV5, 190 ng/ml for UV20, 260 ng/ml for UV41, and 240 ng/ml for UV135. Thus, wild-type AA8 proved most resistant to pierisin-1-induced cytotoxicity. When these CHO cell lines were treated with pierisin-1, the adduct levels of ADP-ribosylated dG increased to 2.5-4.8/10(5) nucleotides time-dependently in all cell lines at 12 h. After removal of pierisin-1, the adduct levels remained constant or increased to 4-14/10(5) nucleotides in all NER mutant cells (UV5, UV20, UV41, UV135), while those rapidly decreased to 0.27/10(5) nucleotides in the repair proficient AA8 cells for 24 h. From these results, it is suggested that the NER system is involved in the repair of ADP-ribosylated dG adducts in DNA.     

The knock-down of ERCC1 but not of XPF causes multinucleation

Excision repair cross complementing gene 1 (ERCC1) associated with xeroderma pigmentosum group F (XPF) is a heterodimeric endonuclease historically involved in the excision of bulky helix-distorting DNA lesions during nucleotide excision repair (NER) but also in the repair of DNA interstrand crosslinks. ERCC1 deficient mice show severe growth retardation associated with premature replicative senescence leading to liver failure and death at four weeks of age. In humans, ERCC1 is overexpressed in hepatocellular carcinoma and in the late G1 phase of hepatocyte cell cycle. To investigate whether ERCC1 could be involved in human hepatocyte cell growth and cell cycle progression, we knocked-down ERCC1 expression in the human hepatocellular carcinoma cell line Huh7 by RNA interference. ERCC1 knocked-down cells were delayed in their cell cycle and became multinucleated. This phenotype was rescued by ERCC1 overexpression. Multinucleation was not liver specific since it also occurred in HeLa and in human fibroblasts knocked-down for ERCC1. Multinucleated cells arose after drastic defects leading to flawed metaphase and cytokinesis. Interestingly, multinucleation did not appear after knocking-down other NER enzymes such as XPC and XPF, suggesting that NER deficiency was not responsible for multinucleation. Moreover, XPF mutant human fibroblasts formed multinucleated cells after ERCC1 knock-down but not after XPF knock-down. Therefore our results seem consistent with ERCC1 being involved in multinucleation but not XPF. This work reveals a new role for ERCC1 distinct from its known function in DNA repair, which may be independent of XPF. The role for ERCC1 in mitotic progression may be critical during development, particularly in humans.     

Activity of individual ERCC1 and XPF subunits in DNA nucleotide excision repair

ERCC1–XPF is a structure-specific nuclease with two subunits, ERCC1 and XPF. The enzyme cuts DNA at junctions where a single strand moves 5′ to 3′ away from a branch point with duplex DNA. This activity has a central role in nucleotide excision repair (NER), DNA cross-link repair and recombination. To dissect the activities of the nuclease it is necessary to investigate the subunits individually, as studies of the enzyme so far have only used the heterodimeric complex. We produced recombinant ERCC1 and XPF separately in Escherichia coli as soluble proteins. Activity was monitored by a sensitive dual incision assay for NER by complementation of cell extracts. XPF and ERCC1 are unstable in mammalian cells in the absence of their partners but we found, surprisingly, that ERCC1 alone could confer some repair to extracts from ERCC1-defective cells. A version of ERCC1 lacking the first 88 non-conserved amino acids was also functional. This indicated that a small amount of active XPF was present in ERCC1 extracts, and immunoassays showed this to be the case. Some repair in XPF-defective extracts could be achieved by adding ERCC1 and XPF proteins together, but not by adding only XPF. The results show for the first time that functional ERCC1–XPF can be formed from separately produced subunits. Protein sequence comparison revealed similarity between the ERCC1 family and the C-terminal region of the XPF family, including the regions of both proteins that are necessary for the ERCC1–XPF heterodimeric interaction. This suggests that the ERCC1 and XPF families are related via an ancient duplication.     

Defining the roles of nucleotide excision repair and recombination in the repair of DNA interstrand cross-links in mammalian cells

The mechanisms by which DNA interstrand cross-links (ICLs) are repaired in mammalian cells are unclear. Studies in bacteria and yeasts indicate that both nucleotide excision repair (NER) and recombination are required for their removal and that double-strand breaks are produced as repair intermediates in yeast cells. The role of NER and recombination in the repair of ICLs induced by nitrogen mustard (HN2) was investigated using Chinese hamster ovary mutant cell lines. XPF and ERCC1 mutants (defective in genes required for NER and some types of recombination) and XRCC2 and XRCC3 mutants (defective in RAD51-related homologous recombination genes) were highly sensitive to HN2. Cell lines defective in other genes involved in NER (XPB, XPD, and XPG), together with a mutant defective in nonhomologous end joining (XRCC5), showed only mild sensitivity. In agreement with their extreme sensitivity, the XPF and ERCC1 mutants were defective in the incision or "unhooking" step of ICL repair. In contrast, the other mutants defective in NER activities, the XRCC2 and XRCC3 mutants, and the XRCC5 mutant all showed normal unhooking kinetics. Using pulsed-field gel electrophoresis, DNA double-strand breaks (DSBs) were found to be induced following nitrogen mustard treatment. DSB induction and repair were normal in all the NER mutants, including XPF and ERCC1. The XRCC2, XRCC3, and XRCC5 mutants also showed normal induction kinetics. The XRCC2 and XRCC3 homologous recombination mutants were, however, severely impaired in the repair of DSBs. These results define a role for XPF and ERCC1 in the excision of ICLs, but not in the recombinational components of cross-link repair. In addition, homologous recombination but not nonhomologous end joining appears to play an important role in the repair of DSBs resulting from nitrogen mustard treatment.     

Repair of site-specific double-strand breaks in a mammalian chromosome by homologous and illegitimate recombination.

In mammalian cells, chromosomal double-strand breaks are efficiently repaired, yet little is known about the relative contributions of homologous recombination and illegitimate recombination in the repair process. In this study, we used a loss-of-function assay to assess the repair of double-strand breaks by homologous and illegitimate recombination. We have used a hamster cell line engineered by gene targeting to contain a tandem duplication of the native adenine phosphoribosyltransferase (APRT) gene with an I-SceI recognition site in the otherwise wild-type APRT+ copy of the gene. Site-specific double-strand breaks were induced by intracellular expression of I-SceI, a rare-cutting endonuclease from the yeast Saccharomyces cerevisiae. I-SceI cleavage stimulated homologous recombination about 100-fold; however, illegitimate recombination was stimulated more than 1,000-fold. These results suggest that illegitimate recombination is an important competing pathway with homologous recombination for chromosomal double-strand break repair in mammalian cells.     

Correction of a deletion mutant by gene targeting with an adenovirus vector.

The usefulness of adenovirus type 5 as a vector for homologous recombination was examined in CHO cells by using the adenine phosphoribosyltransferase (aprt) gene. Infection of a hemizygous CHO APRT- cell line containing a 3-bp deletion in exon 5 of the aprt gene with a recombinant adenovirus containing the wild-type gene resulted in restoration of the APRT+ phenotype at a frequency of 10(-5) to 10(-6) per infected cell. A relatively high frequency (approximately 6 to 20%) of the transductants appears to result from a homologous recombination event. The mutation on the chromosomal aprt gene is corrected in the homologous recombinants, and APRT expression is restored to a normal hemizygous level. Neither adenovirus nor exogenous promoter sequences are detected in the homologous recombinants. The remaining transductants result from random integration of the aprt gene with the adenovirus sequence. A number of adenovirus vectors containing different promoter sequences linked to the hamster aprt gene were constructed. A possible role for the promoter region in the homologous recombination event was indicated by the lack of homologous recombination in constructs lacking an active promoter.     

DNA repair deficiency and BPDE-induced chromosomal alterations in CHO cells

The induction of chromosomal aberrations and sister chromatid exchanges by BPDE was evaluated in parental and different DNA repair deficient Chinese hamster ovary cell lines in order to elucidate the mechanisms involved in their induction. These included the parental line (AA8), nucleotide excision repair (UV4, UV5, UV61), base excision repair (EM9), homologous recombination repair (Irs1SF) and non-homologous end joining (V3-3) deficient ones. The ranking of different cell lines for BPDE-induced chromosome aberrations was: UV4, Irs1SF, UV5, UV 61, EM9, V3-3, and AA8 in a descending order. Cells deficient in NER and HRR were found to be very sensitive, indicating the importance of these pathways in the repair of lesions induced by BPDE. For induction of SCEs, HRR and BER deficient cells were refractory, whereas the other cell lines responded with a dose-dependent increase. The possible mechanisms involved in BPDE-induced chromosomal alterations are discussed.     

Growth retardation, early death, and DNA repair defects in mice deficient for the nucleotide excision repair enzyme XPF

Xeroderma pigmentosum (XP) is a human genetic disease which is caused by defects in nucleotide excision repair. Since this repair pathway is responsible for removing UV irradiation-induced damage to DNA, XP patients are hypersensitive to sunlight and are prone to develop skin cancer. Based on the underlying genetic defect, the disease can be divided into the seven complementation groups XPA through XPG. XPF, in association with ERCC1, constitutes a structure-specific endonuclease that makes an incision 5' to the photodamage. XPF-ERCC1 has also been implicated in both removal of interstrand DNA cross-links and homology-mediated recombination and in immunoglobulin class switch recombination (CSR). To study the function of XPF in vivo, we inactivated the XPF gene in mice. XPF-deficient mice showed a severe postnatal growth defect and died approximately 3 weeks after birth. Histological examination revealed that the liver of mutant animals contained abnormal cells with enlarged nuclei. Furthermore, embryonic fibroblasts defective in XPF are hypersensitive to UV irradiation and mitomycin C treatment. No defect in CSR was detected, suggesting that the nuclease is dispensable for this recombination process. These phenotypes are identical to those exhibited by the ERCC1-deficient mice, consistent with the functional association of the two proteins. The complex phenotype suggests that XPF-ERCC1 is involved in multiple DNA repair processes.     

UV mutagenesis, cytotoxicity and split-dose recovery in a human-CHO cell hybrid having intermediate (6-4) photoproduct repair

Somatic cell hybrids constructed between UV-hypersensitive Chinese hamster ovary cell line UV20 and human lymphocytes were used to examine the influence of a human DNA repair gene, ERCC1, on UV photoproduct repair, mutability at several drug-resistance loci, UV cytotoxicity and UV split-dose recovery. In hybrid cell line 20HL21-4, which contains human chromosome 19, UV-induced mutagenesis at the APRT, HPRT and Na+/K+-ATPase loci was comparable to that in repair-proficient CHO AA8 cells, whereas cell line 20HL21-7, a reduced human-CHO hybrid not containing human chromosome 19, exhibited a hypermutable phenotype at all 3 loci indistinguishable from that of UV20 cells. The response of 20HL21-4 cells to UV cytotoxicity reflected substantial but incomplete restoration of wild-type UV cytotoxic response, whereas responses of UV20 and 20HL21-7 cell lines to UV cytotoxicity were essentially the same, reflecting several-fold UV hypersensitivity. Repair of UV-induced (5-6) cyclobutane dimers and (6-4) photoproducts was examined by radioimmunoassay; (6-4) photoproduct repair was deficient in UV20 and 20HL21-7 cell lines, and intermediate in 20HL21-4 cells relative to wild-type CHO AA8 cells. UV split-dose recovery in 20HL21-4 cells was also intermediate relative to AA8 cells. These results show that the human ERCC1 gene on chromosome 19 is responsible for substantial restoration of UV survival and mutation responses in repair-deficient UV20 cells, but only partially restores (6-4) UV photoproduct repair and UV split-dose recovery.     

Homology dependence of targeted recombination at the Chinese hamster APRT locus.

Using simple linear fragments of the Chinese hamster adenine phosphoribosyltransferase (APRT) gene as targeting vectors, we have investigated the homology dependence of targeted recombination at the endogenous APRT locus in Chinese hamster ovary (CHO) cells. We have examined the effects of varying either the overall length of targeting sequence homology or the length of 5' or 3' flanking homology on both the frequency of targeted homologous recombination and the types of recombination events that are obtained. We find an exponential (logarithmic) relationship between length of APRT targeting homology and the frequency of targeted recombination at the CHO APRT locus, with the frequency of targeted recombination dependent upon both the overall length of targeting homology and the length of homology flanking each side of the target gene deletion. Although most of the APRT+ recombinants analyzed reflect simple targeted replacement or conversion of the target gene deletion, a significant fraction appear to have arisen by target gene-templated extension and correction of the targeting fragment sequences. APRT fragments with limited targeting homology flanking one side of the target gene deletion yield proportionately fewer target gene conversion events and proportionately more templated extension and vector correction events than do fragments with more substantial flanking homology.     

Nucleotide Excision Repair Is Not Required for the Antiapoptotic Function of Insulin-like Growth Factor 1

The expression of ERCC1, a member of the nucleotide excision repair (NER) family, is enhanced in cells transfected with insulin-like growth factor 1 (IGF-1) receptors. Of interest, an excellent concordance between ERCC1 expression and NER-mediated cell survival has been demonstrated. The two aims of the present study were to determine the signaling pathways used by IGF-1 to confer protection against apoptotic cell death in Chinese hamster ovary (CHO) cells and to assess the role of NER in this IGF-1 action. Experiments with pharmacological inhibitors indicated that phosphatidylinositol 3-kinase (PI 3-kinase) but not mitogen-activated protein kinase (ERK1/ERK2) mediates IGF-1 antiapoptotic activity. Using two series of CHO cells that have altered expression of ERCC1 or XPB/ERCC3, we examined IGF-1's ability to delay apoptotic death and reduction of mitochondrial oxidative function mediated by growth factor withdrawal. IGF-1 effectively blocked apoptosis, concomitant with increased MTT activity, in a pair of CHO cell lines expressing inactive ERCC1 (43-3B cells) and the transfected line of the mutant carrying the expressed human ERCC1 gene (83-G5 cells). Similarly, repair-deficient UV24 cells, which lack XPB/ERCC3, and their parental line AA8 were also responsive to the IGF-1's antiapoptotic capacity. In the presence of IGF-1, these cell lines became resistant to the cleavage of poly(ADP-ribose) polymerase, a key player in DNA damage recognition and DNA repair. These results suggest that PI 3-kinase activation plays a determinant role in the antiapoptotic function of IGF-1, but that functional NER does not play a critical part in mediating this IGF-1 response.     

A DNA repair gene of Caenorhabditis elegans: a homolog of human XPF

The xeroderma pigmentosum complementation group F (XPF) protein is a structure-specific endonuclease in a complex with ERCC1 and is essential for nucleotide excision repair (NER). We report a single cDNA of Caenorhabditis elegans (C. elegans) encoding highly similar protein to human XPF and other XPF members. We propose to name the corresponding C. elegans gene xpf. Messenger RNA for C. elegans xpf is 5'-tagged with a SL2 splice leader, suggesting an operon-like expression for xpf. Using RNAi, we showed that loss of C. elegans xpf function caused hypersensitivity to ultra-violet (UV) irradiation, as observed in enhanced germ cell apoptosis and increased embryonic lethality. This study suggests that C. elegans xpf is conserved in evolution and plays a role in the repair of UV-damaged DNA in C. elegans.     

Induction of chromosomal aberrations in the CHO mutant EM9 and its parental line AA8 by EcoRI restriction endonuclease: electroporation experiments

EcoRI restriction endonuclease (RE), which produces cohesive-ended double-strand breaks (dsb) in DNA, was tested in the ethyl methanesulfonate- and X-ray-sensitive CHO mutant EM9 and its parental cell strain AA8 for its chromosomal aberration-inducing effect. The RE was efficiently introduced by electroporation into AA8 cells, while the mutant cells showed a very high sensitivity to electroporation, which consistently resulted in cell death. Nevertheless, the incubation of EM9 cells in the presence of EcoRI, without electroporation, was sufficient to induce about three times the chromosome damage observed in the electroporated parental cell line AA8 for any given dose of the RE.     

Inhibition of transient gene expression in Chinese hamster ovary cells by cyclobutane dimers and (6-4) photoproducts in transfected ultraviolet-irradiated plasmid DNA

Using a transient gene expression assay to measure host cell reactivation, the effects of cyclobutane dimer and noncyclobutane dimer uv photoproducts on expression of a reporter gene were examined in normal and repair-deficient Chinese hamster ovary (CHO) cell lines. Ultraviolet damage in plasmid pRSV beta gal DNA, containing the Escherichia coli beta-galactosidase gene, resulted in reduced reporter gene expression in both uv-hypersensitive mutant CHO cell lines UV5 and UV61 relative to wild-type, parental AA8 cells. However, the effects of uv irradiation of transfected plasmid DNA on gene activity were reduced in UV61, a mutant with normal (6-4) photoproduct repair, compared to UV5, which is deficient in (6-4) photoproduct repair; this reduction correlated with the intermediate uv-hypersensitivity of UV61. Selective removal of cyclobutane dimers by in vitro photoreactivation of uv-irradiated plasmid DNA prior to transfection substantially increased reporter gene activity in both uv-hypersensitive mutant cell lines. This increase was significantly greater in UV61 than in UV5, consistent with UV5 being deficient in repair of both (6-4) photoproducts and cyclobutane dimers. These results suggest that unrepaired (6-4) photoproducts in transfected pRSV beta gal plasmid DNA are responsible for a significant fraction of the reduction in transient gene expression observed in recipient uv-hypersensitive CHO cell mutants.     

A method to monitor replication fork progression in mammalian cells: nucleotide excision repair enhances and homologous recombination delays elongation along damaged DNA

The capacity to rescue stalled replication forks (RFs) is important for the maintenance of cell viability and genome integrity. Here, we have developed a novel method for monitoring RF progression and the influence of DNA lesions on this process. The method is based on the principle that each RF is expected to be associated with a pair of single-stranded ends, which can be analyzed by employing strand separation in alkali. This method was applied to examine the rate of RF progression in Chinese hamster cell lines deficient in ERCC1, which is involved in nucleotide excision repair (NER), or in XRCC3, which participates in homologous recombination repair, following irradiation with ultraviolet (UV) light or exposure to benzo(a)pyrene-7,8-diol-9,10-epoxide (BPDE). The endpoints observed were cell survival, NER activity, formation of double-strand breaks and the rate of RF progression. Subsequently, we attempted to explain our observation that cells deficient in XRCC3 (irs1SF) exhibit enhanced sensitivity to UV radiation and BPDE. irs1SF cells demonstrated a capacity for NER that was comparable with wild-type AA8 cells, but the rate of RF progression was even higher than that for the wild-type AA8 cells. As expected, cells deficient in ERCC1 (UV4) showed no NER activity and were hypersensitive to both UV radiation and BPDE. The observation that cells deficient in NER displayed a pronounced delay in RF progression indicates that NER plays an important role in maintaining fork progression along damaged DNA. The elevated rate of RF progression in XRCC3-deficient cells indicates that this protein is involved in a time-consuming process which resolves stalled RFs.     

ERCC4 (XPF) encodes a human nucleotide excision repair protein with eukaryotic recombination homologs.

ERCC4 is an essential human gene in the nucleotide excision repair (NER) pathway, which is responsible for removing UV-C photoproducts and bulky adducts from DNA. Among the NER genes, ERCC4 and ERCC1 are also uniquely involved in removing DNA interstrand cross-linking damage. The ERCC1-ERCC4 heterodimer, like the homologous Rad10-Rad1 complex, was recently found to possess an endonucleolytic activity that incises on the 5' side of damage. The ERCC4 gene, assigned to chromosome 16p13.1-p13.2, was previously isolated by using a chromosome 16 cosmid library. It corrects the defect in Chinese hamster ovary (CHO) mutants of NER complementation group 4 and is implicated in complementation group F of the human disorder xeroderma pigmentosum. We describe the ERCC4 gene structure and functional cDNA sequence encoding a 916-amino-acid protein (104 kDa), which has substantial homology with the eukaryotic DNA repair and recombination proteins MEI-9 (Drosophila melanogaster), Rad16 (Schizosaccharomyces pombe), and Rad1 (Saccharomyces cerevisiae). ERCC4 cDNA efficiently corrected mutants in rodent NER complementation groups 4 and 11, showing the equivalence of these groups, and ERCC4 protein levels were reduced in mutants of both groups. In cells of an XP-F patient, the ERCC4 protein level was reduced to less than 5%, consistent with XPF being the ERCC4 gene. The considerable identity (40%) between ERCC4 and MEI-9 suggests a possible involvement of ERCC4 in meiosis. In baboon tissues, ERCC4 was expressed weakly and was not significantly higher in testis than in nonmeiotic tissues.     

ERCC1/XPF removes the 3' overhang from uncapped telomeres and represses formation of telomeric DNA-containing double minute chromosomes

Human telomeres are protected by TRF2. Inhibition of this telomeric protein results in partial loss of the telomeric 3' overhang and chromosome end fusions formed through nonhomologous end-joining (NHEJ). Here we report that ERCC1/XPF-deficient cells retained the telomeric overhang after TRF2 inhibition, identifying this nucleotide excision repair endonuclease as the culprit in overhang removal. Furthermore, these cells did not accumulate telomere fusions, suggesting that overhang processing is a prerequisite for NHEJ of telomeres. ERCC1/XPF was also identified as a component of the telomeric TRF2 complex. ERCC1/XPF-deficient mouse cells had a novel telomere phenotype, characterized by Telomeric DNA-containing Double Minute chromosomes (TDMs). We speculate that TDMs are formed through the recombination of telomeres with interstitial telomere-related sequences and that ERCC1/XPF functions to repress this process. Collectively, these data reveal an unanticipated involvement of the ERCC1/XPF NER endonuclease in the regulation of telomere integrity and establish that TRF2 prevents NHEJ at telomeres through protection of the telomeric overhang from ERCC1/XPF.