Interplay between Cernunnos-XLF and Nonhomologous End-joining Proteins at DNA Ends in the Cell*

Cernunnos-XLF is the most recently identified core component in the nonhomologous end-joining (NHEJ) pathway for the repair of DNA double strand breaks (DSBs) in mammals. It associates with the XRCC4/ligase IV ligation complex and stimulates its activity in a still unknown manner. NHEJ also requires the DNA-dependent protein kinase that contains a Ku70/Ku80 heterodimer and the DNA-dependent protein kinase catalytic subunit. To understand the interplay between Cernunnos-XLF and the other proteins implicated in the NHEJ process, we have analyzed the interactions of Cernunnos-XLF and NHEJ proteins in cells after treatment with DNA double strand-breaking agents by means of a detergent-based cellular fractionation protocol. We report that Cernunnos-XLF is corecruited with the core NHEJ components on chromatin damaged with DSBs in human cells and is phosphorylated by the DNA-dependent protein kinase catalytic subunit. Our data show a pivotal role for DNA ligase IV in the NHEJ ligation complex assembly and recruitment to DSBs because the association of Cernunnos-XLF with the XRCC4/ligase IV complex relies primarily on the DNA ligase IV component, and an intact XRCC4/ligase IV complex is necessary for Cernunnos-XLF mobilization to damaged chromatin. Conversely, a Cernunnos-XLF defect has no apparent impact on the XRCC4/ligase IV association and recruitment to the DSBs or on the stimulation of the DNA-dependent protein kinase on DNA ends.

DNA double strand breaks (DSBs) 3 occur physiologically in lymphocytes during V(D)J recombination or class-switch recombination (1). However, DSBs in most cells are pathologic like those arising from genomic replication in the presence of nicks or from cell treatment with ionizing radiation (IR) or radiomimetic molecules. Improper repair of DSBs may lead to cell death or cancer-prone genomic rearrangements (2,3).
In mammalian cells, apart from homologous recombination between sister chromatids, DSBs are mainly processed by in situ religation relying on the nonhomologous end-joining (NHEJ) pathway (4). Although alternative subpathways may operate (5,6), the major NHEJ pathway relies on a set of core proteins, the individual deficiency of which elicits a radiosensitive severe combined immunodeficiency syndrome in human or animals (7). The two DNA ends of the DSB are recognized and bound by the ring-shaped heterodimer Ku70/Ku80 that recruits the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) (8). The assembled DNA-dependent protein kinase (DNA-PK) holoenzyme then exhibits serine-threonine protein kinase and DNA end-bridging activities (9,10). Among other functions, the kinase activity regulates DNA end access to processing enzymes like the DNA-PKcs-associated Artemis nuclease (11)(12)(13). This explains why DNA-PK is a favored target in radiosensitization strategies of tumors (14). Finally, the XRCC4/DNA ligase IV complex is responsible for the ligation step (15,16).
Another core NHEJ factor is Cernunnos-XLF, a factor with a predicted structural similarity to XRCC4 that has been identified as an XRCC4-interacting protein (18) that is deficient in a human radiosensitive severe combined immunodeficiency syndrome (17). Cernunnos-XLF is the homologue of the yeast protein Nej1p in Saccharomyces cerevisiae (19) and belongs to a larger family of functionally conserved proteins that are required for NHEJ (20). This factor has been postulated to function in NHEJ events based on its interaction with the XRCC4/ ligase IV complex (18,19,21), the lack of V(D)J recombination activity in plasmid transfection assays and on the high IR sensitivity of the corresponding deficient cells (17,18,22) associated with an absence of NHEJ activity in vitro (17,23). Cernunnos-defective embryonic stem cells show an impaired ability to form both V(D)J coding and signal joins in transient recombination assays (22). Cernunnos-XLF is therefore likely involved in most NHEJ reactions and not just those that require end processing in contrast to DNA-PKcs and Artemis. In addition, like XRCC4 and DNA ligase IV, Cernunnos-XLF may be implicated in the development of the central nervous system (24). Its participation in the XRCC4/ligase IV complex, its structural resemblance with XRCC4 (18,19), and its specific stimulation of ligation in ligase IV-dependent assays (20,21) suggest that this protein might function by activating or enhancing the basic NHEJ ligation reaction (25), but its precise function is still unknown.
To get insight into the relationship between Cernunnos-XLF and the other members of the NHEJ process, we have analyzed the interactions of Cernunnos-XLF and NHEJ proteins in cells after treatment with double strand-breaking agents. In particular, we have used a detergent-based cellular fractionation protocol that allows assessment in situ of the DSB-induced recruitment of NHEJ repair proteins after cell treatment with IR or radiomimetic molecules (26).

MATERIALS AND METHODS
Chemicals-Calicheamicin ␥1 (Cali), a generous gift from P. R. Hamann (Wyeth Research, Pearl River, NY), was dissolved at 4 mM in ethanol and stored at Ϫ70°C. Wortmannin (Sigma) and NU7026 (Calbiochem) were dissolved in Me 2 SO (10 mM stock solution) and stored at Ϫ20°C. Small aliquots of stock solutions chemicals were used once.
Cell Culture and Cloning-All culture media were from Invitrogen and were supplemented with 10% fetal calf serum unless indicated, 2 mM glutamine, 125 units/ml penicillin, and 125 g/ml streptomycin. All cells were grown in a humidified atmosphere at 37°C with 5% CO 2 . MRC5-SV2 was from the European Collection of Cell Cultures (ECACC, Salisbury, Wiltshire, UK) and was grown in Dulbecco's modified Eagle's medium. BuS cells are SV40T-transformed, telomerase-immortalized radiosensitive fibroblasts from the Cernunnos-deficient P2 patient as published (17). BuC cells were obtained after transduction of BuS with the pMND-Cernunnos-Myc-ires-GFP retroviral vector expressing a C-terminal Myc-His-tagged Cernunnos protein. 4 Both cell lines were grown in RPMI 1640 medium. DNA-PKcs-deficient and -complemented cell lines (Fus9-alias M059J and Fus1, respectively (27), a gift from Dr. C. Kirchgessner, Stanford University School of Medicine, CA) were maintained in Dulbecco's modified Eagle's medium-F12 1/1. The LIG4-defective N114P2 cells and the parental cell line Nalm-6 (gifts from Dr. M. R. Lieber, University of Southern California, Los Angeles) were isolated as described previously (15) and cultured in RPMI 1640 medium. Small interfering RNA design and cloning in pEBV-based small interfering RNA vectors carrying a hygromycin B resistance cassette and establishment of knockdown and control HeLa clones were as described elsewhere (28,29). HeLa clones were grown in Dulbecco's modified Eagle's medium in the presence of 125 g/ml hygromycin B (Invitrogen). The same procedure as for HeLa was followed to establish an MRC5-shL4 clone expressing a short hairpin RNA (shRNA) silencing LIG4 and a BuC-shX4 clone expressing a shRNA silencing XRCC4. The RNA-interfering sequences for LIG4 (NM_002312) and for XRCC4 (NM_022550) were nucleotides 1939 to 1957 and nucleotides 674 to 692, respectively.
DNA-damaging Treatments-For drug exposure, exponentially growing cells were either mock-treated or treated with freshly diluted calicheamicin at the specified concentrations in medium at 37°C in culture dishes and then harvested at the indicated time points. For cell treatment with IR, irradiation was carried out in a Faxitron RX-650 irradiator (Faxitron X-ray Corp., Buffalo Grove, IL) at a dose rate of 5.72 grays/min. For UV irradiation, cells were washed with phosphate-buffered saline (PBS) and then exposed to UVC irradiation (254 nm) from a germicidal lamp (Bioblock Scientific). Immediately after irradiation, unsupplemented medium was added, and cells were postincubated as above.
Biochemical Fractionation and Immunoblotting-Treated or mock-treated cells in culture dishes were washed twice with ice-cold PBS, collected by scraping, and centrifuged. Cell fractionation was carried out by two consecutive extractions. Pellets of about 1 ϫ 10 6 cells were first resuspended for 10 min on ice in 200 l of extraction buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA) containing 0.1% Triton X-100, supplemented with protease inhibitor mixture tablets (Complete Mini TM , Roche Diagnostics) and phosphatase inhibitors (10 mM NaF, 10 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, and 1 mM cantharidin, all from Sigma). Following centrifugation at 14,000 ϫ g for 3 min, the supernatant was collected (fraction S1), and the pellet was washed with extraction buffer without Triton. The pellet was further incubated in 100 l of extraction buffer without Triton but supplemented with 200 g/ml Rnase A (Sigma) for 30 min at 25°C under agitation. Following centrifugation at 14,000 ϫ g for 3 min, the pellet was washed with extraction buffer without Triton (fraction P2). Insoluble P2 fraction was resuspended in PBS buffer supplemented with 1% SDS, heated 10 min at 100°C, and sonicated for 10 s (Vibracel, Bioblock Scientific). Whole cell extracts (WCEs) of treated or mock-treated cells were obtained by direct lysis in PBS buffer supplemented with 1% SDS and treatment as above. When necessary, the treated or mock-treated cell pellets were resuspended in 1ϫ lambda phosphatase buffer (New England Biolabs) with 1% Triton X-100 in the presence of 2 mM magnesium chloride and protease inhibitors as above, sonicated on ice, and incubated for 1 h at 37°C with 400 units of lambda phosphatase (New England Biolabs). Concentrated loading sample buffer was added for 1ϫ final concentration in all fractions, and the samples were boiled for 5 min. Equal aliquots of each fraction derived from equivalent cell numbers were sepa-rated on SDS-polyacrylamide gels (10% for standard separation or 15% for ␥-H2AX isolation) and blotted onto polyvinylidene difluoride membranes (Immobilon-P, Millipore). Membranes were blocked for 1 h in 5% dry milk in PBS containing 0.1% Tween 20 (PBS-T) and incubated for 1 h with primary antibody diluted in PBS containing 0.02% Tween 20 and 1% bovine serum albumin (fraction V, Sigma). After three washes with PBS-T, membranes were incubated for 1 h with secondary antibodies in PBS containing 0.02% Tween 20 and 5% dry milk. Immunoblots were visualized by enhanced chemiluminescence (Immunofax A, Yelen). When necessary, successive immunoblotting was performed on the same membranes after stripping (Restore Western blot stripping buffer, Pierce). For data presentation, films were scanned and processed with Adobe Photo-Shop 3.0 software.
Coimmunoprecipitation Assay-Cell extracts were obtained as follows. Cells were washed with cold PBS, spun at 4°C, 300 ϫ g for 5 min, resuspended in hypotonic buffer HB (10 mM Hepes pH 7.5, 25 mM KCl, 10 mM NaCl, 1 mM MgCl 2 , 0.1 mM EDTA) supplemented with protease inhibitors (Complete Mini TM , Roche Diagnostics) and phosphatase inhibitors (10 mM NaF, 10 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, and 1 mM cantharidin, all from Sigma), and then lysed by freezing in liquid nitrogen and thawing at 37°C three times. Lysates were then supplemented with NaCl to final 350 mM and cleared by spinning at 4°C, 15,000 ϫ g for 30 min. The soluble proteins were diluted with hypotonic buffer to final 120 mM NaCl, and protein concentration was measured. 100 g of proteins were mixed in the reaction volume completed to 100 l with immunoprecipitation buffer (10 mM Hepes, pH 7.5, 25 mM KCl, 120 mM NaCl, 1 mM MgCl 2 , 0.1 mM EDTA) with proteinase inhibitors and phosphatase inhibitors as above. When necessary, ethidium bromide was added at 100 g/ml final concentration. For anti-Cernunnos-XLF immunoprecipitation, the mixture was mixed with 10 l of magnetic anti-rabbit IgG Immunobeads suspension coated with the anti-XLF primary antibody according to the manufacturer's protocol (Dynal), and the beads were mixed gently on a wheel for 3 h at 4°C. The beads were pulled down over a magnet, the supernatant extract was removed, the beads were washed three times with 1 ml of icecold PBS-T, and proteins in the immunoprecipitates were eluted by boiling in SDS sample buffer. Samples were thereafter incubated for 30 min at room temperature in the presence of iodoacetamide (100 mM) and then separated in a 10% acrylamide Tris-glycine-SDS gel. For immunodetection after transfer on polyvinylidene difluoride membranes, rabbit TrueBlot horseradish peroxidase-conjugated anti-rabbit secondary antibodies (eBioscience) were used as secondary antibodies to reduce the interfering signal of the immunoglobulins used for the immunoprecipitation.

RESULTS
Cernunnos-XLF Is Phosphorylated by DNA-PKcs during the Cellular Response to DSBs-Cernunnos-XLF has been identified recently as a new member of the NHEJ apparatus. Because other members such as XRCC4 (26), Artemis (11), and DNA-PKcs itself (30,31) are phosphorylated in the nucleus after gen-eration of DSBs, we first addressed the possibility that Cernunnos-XLF could also be a substrate of DNA-PKcs.
Cernunnos-XLF was easily detected by Western blot on WCEs from the human MRC5-SV2 fibroblasts, as compared with extracts from XLF-deficient human BuS cells carrying a truncating mutation at amino acid 177 in which it was undetectable (Fig. 1A). For production of DSBs, MRC5-SV2 cells were treated with IR or with Cali, which yields a 1:3 ratio of DNA double strand breaks to single strand breaks in vivo compared with a 1:20 ratio for IR (32). As opposed to non-or UVtreated cells, WCE from IR-and Cali-treated cells contained ␥-H2AX, the phosphorylated form on Ser139 of the histone H2AX variant, which is admitted to be a quantitative nuclear marker of DSBs (33) (Fig. 1B). Under both of these latter conditions, XRCC4 exhibited a slower migrating form that was sensitive to lambda phosphatase (Fig. 1C), indicating that it corresponded to phosphorylation as reported previously (26). Similarly, a slower migrating form of Cernunnos-XLF sensitive to lambda phosphatase was detected which was faint after IR but predominant after Cali (Fig. 1, B and C). This indicates that, like XRCC4, Cernunnos-XLF is phosphorylated in the cell in the presence of DSBs.
We sought to determine whether DNA-PKcs activity was necessary for Cernunnos-XLF phosphorylation after DSBs. We first used the selective DNA-PKcs inhibitor NU7026 that has been shown to exhibit a strong DNA-PKcs-dependent radiosensitization effect on cells (34). MRC5-SV2 cells were pretreated or not with NU7026 before treatment with Cali. As shown in Fig. 2A, the DNA-PKcs inhibitor strongly reduced both XRCC4 and Cernunnos-XLF phosphorylation. This suggests that the phosphorylation observed under these conditions A, whole cell extracts of BuS and MRC5-SV2 cells were denatured and separated on 10% SDS-PAGE gels followed by electrotransfer onto membranes. The membranes were blotted with the antibodies as indicated. B, MRC5-SV2 cells in culture dishes were not treated, irradiated with UV-C light (UV, 60 J/m 2 ) or X-rays (IR, 60 Gy), or treated with calicheamicin (Cali, 10 nM for 1 h at 37°C). Cells were collected and lysed in denaturating buffer. WCEs were denatured and separated on SDS-PAGE gels (10% for standard separation or 15% for ␥H2AX isolation) followed by electrotransfer onto membranes. C, whole cell extracts of MRC5-SV2 cells treated or not with calicheamicin as above were prepared in lambda phosphatase buffer as described under "Materials and Methods" and then incubated or not with the enzyme (PPase, 400 units) as indicated before protein separation and blotting.
mostly relies on the NU7026-sensitive DNA-PKcs activity. Then we treated with Cali the M059J glioblastoma cells that do not express DNA-PKcs (DNA-PKcs-deficient cells, Fus9) and M059J-complemented cells that contain an extra copy of the human gene coding for DNA-PKcs (DNA-PKcs-complemented cells, Fus1) (27). As shown in Fig. 2B, although both XRCC4 and Cernunnos-XLF were phosphorylated upon Cali treatment of the DNA-PKcs proficient Fus1 cells, no shift was observed for either protein after treatment of the DNA-PKcsdeficient Fus9 cells. These data clearly implicate DNA-PKcs in the phosphorylation of Cernunnos-XLF upon generation of DSBs in DNA of human cells.

Cernunnos-XLF Is Corecruited with the Major NHEJ Components but Is Not Necessary for Their Mobilization to Chromatin
Damaged with DSBs-Cernunnos-XLF has been shown to be part of a larger complex with XRCC4 and DNA ligase IV (18,19). Although it stimulates ligation in a ligase IV-dependent assay (20,21) and it is indispensable for the repair of DSBs in the cell (17,18), the precise role of Cernunnos-XLF is still unknown. We have recently examined nuclear compartmentalization of NHEJ factors and demonstrated that in response to drugs that induce DSBs, NHEJ components mobilize to a detergent-insoluble nuclear fraction corresponding to the damaged chromatin (26). Therefore we have addressed the question of Cernunnos-XLF mobilization to damaged chromatin together with the other NHEJ factors as a result of production of DSBs in the cell.
After 1 h of treatment with calicheamicin, MRC5-SV2 cells were extracted with a buffer containing Triton; thereafter, the cell pellet was treated with RNaseA in the same buffer but without detergent, and insoluble P2 fraction was collected. A parallel extraction procedure was performed on untreated cells and cells treated with Cali. Fig. 3A shows the immunoblot analysis following SDS-PAGE of cell-equivalent aliquots of P2 fraction compared with WCEs under both untreated and Cali-treated conditions. In contrast to extracts from mock-treated cells, extracts from Cali-treated cells contain ␥-H2AX consistent with the high DNA double strand-breaking potency of Cali. In untreated cells, the majority of NHEJ proteins were released during the two extraction steps, and only a marginal amount was detected in the insoluble P2 fraction. However, the P2 fraction from Cali-treated cells was highly enriched for NHEJ proteins, including Cernunnos-XLF. In contrast, ␥-tubulin protein was detected identically in the P2 fraction of drug-treated and nontreated cells. Furthermore, XRCC4 exhibited additional phosphorylated forms in the fractions of Cali-treated cells.
Because Cali also produces about 60% of non-DSBs lesions in DNA, the recruitment of NHEJ proteins observed may rely partly on lesions other than DSBs. However, the comparison of XRCC4 and Cernunnos-XLF recruitments to the P2 fraction of cells treated either with Cali or high doses of the methylating molecule methyl-methanesulfonate clearly showed an exclusive mobilization of these proteins at sites of DSBs (supplemental Fig. 1). We conclude from these data that Cernunnos-XLF is corecruited with the other NHEJ components to chromatin damaged by DSBs.
In addition to XRCC4, Cernunnos-XLF was phosphorylated in the P2 fraction of Cali-treated cells (Fig. 3B). As expected, this phosphorylation was prevented by pretreatment of the cells with Nu7026. Interestingly, Cernunnos-XLF was still recruited under this condition, implying that DNA-PKcs-dependent phosphorylation of Cernunnos-XLF is not required for its stable recruitment to the damaged chromatin, which is similar to what we observed previously with XRCC4 (Fig. 3B) (26).
It has been shown that Cernunnos-XLF binds to DNA and that it stimulates ligation dependent on the XRCC4/ligase IV complex (20,21). Because it is indispensable for the NHEJ in the cell, one possibility is that it mediates the recruitment of the ligation complex to DNA ends. Thus, we tested whether XRCC4/ligase IV relied on Cernunnos-XLF for its recruitment to DNA damaged sites. BuS cells that do not express detectable Cernunnos-XLF were treated with Cali, and the recruitment of NHEJ proteins to the insoluble P2 chromatin fraction was assessed. As shown in Fig. 4, NHEJ proteins including DNA ligase IV were depleted from the S1 soluble nuclear fraction and heavily mobilized to the detergent-resistant nucleoplasmic compartment of BuS cells. After Cali treatment of BuS cells, the DNA-PKcs substrate XRCC4 was phosphorylated, and DNA-  PKcs itself was normally autophosphorylated on the Ser2056 site (Fig. 4), indicating that DNA-PKcs was stimulated efficiently by DSBs in the absence of Cernunnos-XLF. No difference was observed in the recruitment of NHEJ proteins and phosphorylation of XRCC4 after Cali treatment, between the BuS cell line and its complemented control BuC cell line (supplemental Fig. 2). Thus, we conclude from these data that Cernunnos-XLF in the cell is not necessary either for the mobilization of the major NHEJ components to damaged chromatin or for the stimulation of the DNA-PKcs on DNA-ends.
Cernunnos-XLF Mobilization to the Damaged Chromatin Is Dependent on XRCC4/DNA Ligase IV-To understand the requirement for Cernunnos-XLF recruitment to damaged chromatin, we used stable clones of human cells expressing shRNA directed against NHEJ proteins. First, we used HeLa cells targeted for DNA-PKcs, XRCC4, or ligase IV expression and control cells as published recently (28,29). As shown in Fig.  5A, each of the HeLa-shL4, -shX4, and -shPKcs clones exhibited a strong reduction in the expression of the corresponding targeted protein as compared with the control HeLa-Vec and with Ku70 expression. Because XRCC4 is responsible for ligase IV stability (35), a strong reduction in ligase IV expression was also observed in HeLa-shX4 cells (Fig. 5A, lane 3). In contrast, HeLa-shL4 showed a strong reduction in ligase IV expression, whereas XRCC4 was still expressed at a normal level (Fig. 5A,  lane 2). As a control, the expression of both Ku70 and Cernunnos-XLF remained broadly unchanged in all the clones. Then the recruitment of NHEJ proteins in the P2 chromatin fraction was checked after Cali treatment (Fig. 5B). As compared with the control HeLa-Vec, no significant change in the recruitment of Ku70 was observed for the three shRNA HeLa clones in the presence of DSBs as revealed by the amount of ␥-H2AX (Fig.  5B, lanes 2, 4, 6, and 8). Indeed, we have shown elsewhere that the mobilization of Ku to damaged chromatin is not dependent on DNA-PKcs and XRCC4/ligase IV (26). In contrast, the XRCC4/ligase IV complex accumulated on damaged chromatin in both HeLa-shX4 and HeLa-shPKcs at a much lower level than in control cells ( Fig. 5B and supplemental Fig. 3, compare lane 2 with lanes 6 and 8). Concomitantly, the recruitment of Cernunnos-XLF was also impaired in both HeLa-shX4 and HeLa-shPKcs cells. Similar results were obtained in a BuC-shX4 clone (data not shown). This indicates that the recruitment of Cernunnos-XLF to DSBs parallels that of the XRCC4/ ligase IV complex. Notably, the marginal amount of Cernunnos-XLF present on damaged chromatin was still phosphorylated in HeLa-shX4 cells but no longer shifted in HeLa-shPKcs cells (Fig. 5B, compare lanes 6 and 8), strengthening again our previous conclusion of a major role of DNA-PKcs for this modification (Fig. 2). In HeLa-shL4 cells, the defect in ligase IV expression impaired XRCC4 accumulation on damaged chromatin (Fig. 5B and supplemental Fig. 3, lane 4), as expected from our published data (26). Strikingly, Cernunnos-XLF was hardly detected in the damaged chromatin fraction in the absence of ligase IV (Fig. 5B and supplemental Fig. 3, lane 4). Similar results were obtained in an MRC5-shL4 clone (supplemental Fig. 4). These observations emphasize the requirement of an intact XRCC4/ligase IV complex for Cernunnos-XLF stable recruitment to chromatin damaged by DSBs.
The ligase IV defect could impair the recruitment of a preformed XRCC4/Cernunnos-XLF or, alternatively, it could prevent the association between Cernunnos-XLF and XRCC4. To distinguish between these possibilities, we performed coimmunoprecipitation experiments with anti-XLF antibodies in a human pre-B cell line, N114P2, with targeted disruption in both ligase IV alleles and its parental line Nalm-6 (15). Control experiments indicated that the antibodies directed against Cernunnos-XLF did not cross-react with XRCC4 in immunopre- . Effect of Cernunnos-XLF defect on the NHEJ proteins mobilization in response to DNA DSBs. BuS cells in culture dishes were treated or not with Cali for 1 h at 37°C. Cells were collected and lysed in denaturating buffer (WCE) or fractionated by two consecutive extractions as described under "Materials and Methods" leading to S1 soluble fraction and P2 insoluble material. Whole cell extracts of MRC5-SV2 were used as control for normal Cernunnos expression. Protein samples were denatured and separated on SDS-PAGE gels (10% for standard separation or 6% for DNA-PKcs isolation) followed by electrotransfer onto membranes. The membranes were blotted with the antibodies as indicated. cipitation and neither Cernunnos-XLF nor XRCC4 interacted nonspecifically with the magnetic Immunobeads (supplemental Fig. 5). As shown in Fig. 6, Cernunnos-XLF coprecipitated with the XRCC4/ligase IV complex as reported (18,19,21) even in the presence of ethidium bromide, which disrupts the protein-DNA interactions (Fig. 6, lanes 1 and 5). In addition, we found that this association was not disrupted upon XRCC4 and Cernunnos-XLF phosphorylation after Cali treatment of the cells (Fig. 6, lane 3). In contrast, a total defect in ligase IV strongly reduced the association between Cernunnos-XLF and XRCC4 under all the conditions tested for coimmunoprecipitation (Fig. 6, lanes 2, 4, and 6), implying that the formation of a Cernunnos-XLF/XRCC4/ligase IV complex mostly relies on the DNA ligase IV component.

DISCUSSION
Cernunnos-XLF is the most recently identified core component in the NHEJ pathway in mammals, but its precise role for the repair of DSBs is still unknown. We report here that Cernunnos-XLF is phosphorylated in human cells in response to DSBs. Because this phosphorylation was abolished in two DNA-PKcs-defective lines (Fus9 and HeLa-shKcs) and by a specific inhibitor of DNA-PKcs, we conclude that this modification relies on the kinase activity of DNA-PKcs. This is reminiscent of the DNA-PKcs-dependent phosphorylation of XRCC4 in cells (26). Nej1p, the ortholog of Cernunnos-XLF in yeast, is also phosphorylated by the DNA damage-signaling pathway, and this phosphorylation enhances the activity of NHEJ (36). No functional relevance of XRCC4 modification has so far been uncovered in mammals (37,38). In coimmunoprecipitation experiments, we found that the Cernunnos-XLF/ XRCC4/ligase IV interactions were not significantly affected by the phosphorylation of both Cernunnos-XLF and XRCC4 proteins after DSBs production. If such a modification of Cernunnos-XLF serves to regulate NHEJ in mammals, it could act by modulating the ligase IV activity or helping to release the complex after ligation completion. Deciphering the function of this protein modification awaits the exhaustive identification of the phosphorylation sites in the cells.
Cernunnos-XLF is indispensable for the repair of DSBs in mammals. Defective cells show a strong sensitivity to IR, a defect in DSB repair including both signal and coding joins in transient transfection assays for V(D)J recombination (17,18,22), and also a complete lack of end-joining activity on linear DNA by cell extracts in vitro (17,23). Given that Cernunnos-XLF is predicted to have a globular N-terminal DNA binding domain related to XRCC4 (18,19) and binds to DNA directly (20,21), it was suggested that Cernunnos-XLF could help to recruit components of the NHEJ process to DNA ends. However, this possibility is excluded because it is shown here that both DNA-PK and the XRCC4/ligase IV complexes are recruited normally to damaged chromatin in cells expressing undetectable levels of Cernunnos-XLF. In addition, it has been shown that XRCC4 has normal nuclear localization in cells deficient in Cernunnos-XLF (19). Similarly Nej1p is not required for the nuclear localization and association with chromatin of the XRCC4 ortholog Lif1p in haploid yeast cells (36). Cernunnos-XLF stimulates the end-joining reaction catalyzed by purified XRCC4/ligase IV in vitro (20,21). Moreover, it stimulates ligation of one strand from mismatched ends and biases the choice of the ligated strand for sequence preservation (39). In addition, a defect in Cernunnos-XLF cannot be complemented by XRCC4 overexpression (19). Together with these results, our data support a role for Cernunnos-XLF in changing the conformation of the DNA/XRCC4/ligase IV complex rather than in assisting the loading of the ligation complex onto DNA ends. Thus, Cernunnos-XLF binding to XRCC4/ligase IV could trigger its alteration into a different active state.
We have assessed the protein requirement for Cernunnos-XLF recruitment to DNA ends in the cell after treatment with a double strand-breaking agent. A strong impairment of Cernunnos-XLF recruitment to damaged chromatin was observed in cells also exhibiting a decreased XRCC4/ligase IV recruitment, i.e. HeLa-shL4, -shX4, and -shKcs stable clones. Indeed, we have already established that DNA ligase IV is physically required for optimal recruitment of XRCC4 to the NHEJ repair complex on damaged chromatin (26). The direct contact between XRCC4 and DNA-PKcs inferred from experiments in vitro (40) may not be sufficient in vivo, and the interaction reported between Ku and ligase IV (41) may be necessary to stabilize XRCC4/ligase IV on the assembled DNA-PK/DNA end complex. Thus, although Cernunnos-XLF binds to DNA in vitro (20,21), we conclude from our data that this intrinsic affinity for DNA is not sufficient in the cell and that its recruitment to damaged chromatin relies on that of XRCC4/ligase IV to which it is associated.
We show additionally that DNA ligase IV is necessary for the optimal interaction between XRCC4 and Cernunnos-XLF. Our coimmunoprecipitation experiments were carried out under physiological salt conditions and with native proteins in crude extracts from both ligase IV-defective and isogenic parental cells. Under these conditions, we detected a weak interaction between XRCC4 and Cernunnos-XLF in the absence of ligase IV that was markedly enhanced in the presence of ligase IV. Hence, the interaction between Cernunnos-XLF and XRCC4 is FIGURE 6. Analysis of Cernunnos-XLF and XRCC4/ligase IV coimmunoprecipitation in extracts from ligase IV-deficient and -proficient cells. Nalm6 (Lig4 ϩ ) and N114P2 (Lig4 Ϫ ) cells were treated or not with Cali (10 nM) for 30 min. Protein extracts obtained as described under "Materials and Methods" were diluted in immunoprecipitation buffer in the presence or not of ethidium bromide (100 g/ml) and incubated with Immunobeads coated with anti-XLF primary antibodies. Then the supernatant extract was removed, and the beads were washed. Proteins in the immunoprecipitates were heated in SDS sample buffer and separated in 10% SDS-PAGE. Western blotting was performed with antibodies as indicated.
weak and stabilized by the robust interaction between XRCC4 and ligase IV. Interestingly, the yeast Cernunnos orthologue Nej1 interacts with Lif1, the XRCC4 homologue, but not with dnl4 (the yeast DNA ligase IV homologue) (42,43). In a transient expression system of tagged proteins, Lu et al. (21) also reported a modest interaction of Cernunnos-XLF with XRCC4 and no interaction with ligase IV. Likewise, we observed that Cernunnos-XLF expression does not counteract the decrease of ligase IV upon XRCC4 knockdown by shRNA in HeLa cells or MRC5-SV2 cells (data not shown). This implies that Cernunnos-XLF does not contribute to ligase IV stability in the absence of XRCC4 and suggests that no significant amount of stable Cernunnos-XLF/ligase IV complex actually exists in cells. In contrast, Cernunnos-XLF is stable in the absence of XRCC4, and so XRCC4 does not help to stabilize Cernunnos-XLF as it does for DNA ligase IV (35).
In conclusion, Cernunnos-XLF associated with XRCC4/ligase IV is probably the only form of this protein that is mobilized to DSBs in the cell. Ligase IV is needed to stabilize both the association of Cernunnos-XLF/XRCC4 and its recruitment to the sites of DNA breaks. Additionally, Cernunnos-XLF is likely necessary to modulate the efficiency and/or the specificity of the XRCC4/ligase IV ligation activity. Finally, Cernunnos-XLF and XRCC4 are phosphorylated by DNA-PKcs during the cellular response to DSBs, but the cellular function of these modifications still waits to be deciphered.