Dynamic assembly of end-joining complexes requires interaction between Ku70/80 and XRCC4

Abstract

DNA double-strand break (DSB) repair by nonhomologous end joining (NHEJ) requires the assembly of several proteins on DNA ends. Although biochemical studies have elucidated several aspects of the NHEJ reaction mechanism, much less is known about NHEJ in living cells, mainly because of the inability to visualize NHEJ repair proteins at DNA damage. Here we provide evidence that a pulsed near IR laser can produce DSBs without any visible alterations in the nucleus, and we show that NHEJ proteins accumulate in the irradiated areas. The levels of DSBs and Ku accumulation diminished in time, showing that this approach allows us to study DNA repair kinetics in vivo. Remarkably, the Ku heterodimers on DNA ends were in dynamic equilibrium with Ku70/80 in solution, showing that NHEJ complex assembly is reversible. Accumulation of XRCC4/ligase IV on DSBs depended on the presence of Ku70/80, but not DNA-PK_CS. We detected a direct interaction between Ku70 and XRCC4 that could explain these requirements. Our results suggest that this assembly constitutes the core of the NHEJ reaction and that XRCC4 may serve as a flexible tether between Ku70/80 and ligase IV.

Genomic stability in all living organisms is constantly threatened by DNA-damaging agents, such as ionizing radiation, UV light, and genotoxic chemicals. DNA double-strand breaks (DSBs) are particularly dangerous lesions, because they can give rise to loss of heterozygosity or chromosomal translocations, which can eventually lead to cancer or cell death (1). Eukaryotes have developed at least two different repair pathways that can deal with these lesions, homologous recombination and nonhomologous end-joining (NHEJ). Homologous recombination needs a homologous template to guide the repair process, whereas NHEJ uses little or no homology to repair DSBs.

The core NHEJ machinery encompasses the DNA-dependent protein kinase (DNA-PK) complex and ligase IV/XRCC4. Patients and mouse models defective in one of these components show ionizing radiation sensitivity and a defective immune system caused by the inability to carry out V(D)J recombination (2).

DNA-PK consists of the DNA end-binding Ku70/80 heterodimer and the DNA-PK catalytic subunit (DNA-PK_CS), which phosphorylates a large number of substrate proteins on serine and threonine residues after DSB induction (3). Although a large number of DNA-PK substrates have been described, the only phosphorylation event that has been proven to be functionally relevant is DNA-PK_CS autophosphorylation (4). DNA-PK bound to DNA termini can juxtapose these ends in a synaptic complex (5). However, the DNA ends are not accessible for processing factors or DNA ligase unless DNA-PK_CS is first autophosphorylated (6–8). The joining of DNA ends is carried out by ligase IV/XRCC4 (9, 10). This process is probably stimulated by the recently identified XLF/Cernunnos protein, which binds to XRCC4 (11, 12).

Although biochemical studies have elucidated several aspects of the NHEJ reaction mechanism, much less is known about NHEJ in living cells, mainly because of the inability to visualize localization of NHEJ repair proteins at DNA damage. Recently, NHEJ and homologous recombination proteins have been shown to accumulate at DNA damage induced by focused laser beams (13–15). Here we provide evidence that a pulsed near IR (NIR) laser can produce DSBs in the nucleus and that NHEJ proteins accumulate in the irradiated areas. We observed that binding of Ku heterodimers to DSBs is a dynamic process involving transient associations with DNA ends. Recruitment of the ligase IV/XRCC4 complex proceeds even in the absence of DNA-PK_CS, which can be explained by a direct Ku70–XRCC4 interaction.

Results

Characterization of the EGFP-Ku80-Expressing Cells.

Although many attempts have been made to reconstitute the NHEJ reaction in vitro, it is still unclear how well these reactions resemble DSB repair in vivo. Therefore, we studied NHEJ in living cells using biologically active, fluorescently tagged proteins. We generated a Ku80-deficient Chinese hamster cell line expressing EGFP-Ku80 (Fig. 7, which is published as supporting information on the PNAS web site). The EGFP-Ku80 protein heterodimerized with the hamster Ku70, and no signs of EGFP or other degradation products were detected, confirming that the GFP signal could indeed be used to visualize the Ku80 fusion protein. The functionality of EGFP-Ku80 was shown by γ-ray survival, end joining, and V(D)J recombination assays (Fig. 8, which is published as supporting information on the PNAS web site).

We first investigated diffusion of EGFP-Ku80 through the nucleus by fluorescence recovery after photobleaching (FRAP) (16). The fluorescence of nuclear EGFP and EGFP-Ku80 recovered to approximately the same level, which implies that the large majority of Ku70/80 was not bound to immobile structures in the nucleus (Figs. 9 and 10, which are published as supporting information on the PNAS web site). Furthermore, EGFP-Ku80 fluorescence recovered with kinetics comparable to two other proteins of similar molecular weight, ERCC1/XPF and XPG (A. Zotter and W. Vermeulen, personal communication), showing that the majority of the Ku70/80 heterodimer is not part of a larger protein complex (Fig. 9).

Accumulation of Ku70/80 on DNA Ends.

We used a pulsed NIR laser to study accumulation of Ku70/80 after DSB induction in a small nuclear region (17). A pattern of parallel lines was scanned through the cells, generating a “cylinder of damage” of <1 μm in diameter through the nuclei. Almost all cells in the irradiated area developed a stripe of γ-H2AX staining. The presence of DNA ends in this area was also confirmed by TUNEL staining and accumulation of Ku80 protein (Fig. 1A). The accumulation of Ku was maximal in the damaged areas with the highest DNA content, suggesting that Ku can interact efficiently with heterochromatin (Fig. 1B). We also selected cells in prometaphase to anaphase and scanned a single line of damage through the nucleus. After this treatment we observed fast accumulation of Ku70/80 where the laser path had intersected with the condensed chromosomes, showing that even this highly condensed form of chromatin supported binding of the Ku proteins to DSBs (Fig. 1C).

Fig. 1.

Accumulation of NHEJ proteins on laser-induced damage. (A) Accumulation of Ku80, EGFP-Ku80, γ-H2AX, and TUNEL staining in V79B cells after laser irradiation. The leftmost panels show the position of the laser-induced damage as dotted gray lines. (B) EGFP-Ku80 accumulation intensities correlate with DNA-dense heterochromatic regions. The chromatin density is revealed by the DNA dye DRAQ5 introduced after laser irradiation was performed. The image was taken 30 min after irradiation. (C) Accessibility of condensed chromosomes to DNA end binding of EGFP-Ku80. The image was taken 10 min after irradiation. (D) A representative cell division of EGFP-Ku80-expressing XR-V15B cells. The time in hours after irradiation is given in each frame. (Scale bars: 5 μm.)

In contrast to what has been observed with other laser-induced DNA damage protocols (14, 18), no damage to the nucleus after NIR-laser irradiation in either the transmitted light or after staining with DAPI, ToPro3, or Draq5 was seen. Moreover, cell membranes were still intact in all cells 2 h after laser irradiation (trypan blue exclusion), and cells that had been damaged could carry out cell division (Fig. 1D).

For kinetic measurements, the NIR laser irradiation was limited to a small disk, rather than a single line (Fig. 2A). EGFP-Ku80 started to accumulate in this damaged area within a few seconds and reached 95% of its maximum after ≈3 min (Fig. 2B), suggesting that most DNA ends have recruited Ku70/80 within that time range.

Fig. 2.

Ku70/80 interaction with DSBs in vivo. (A) Example of EGFP-Ku80 accumulation after laser irradiation of a disk inside the nucleus of XR-V15B cells expressing EGFP-Ku80. (B) Accumulation curve of EGFP-Ku80 on laser-induced DNA damage. For every cell, the pre-damage fluorescence level was set to 0 and the maximum was set to 1. The average and twice the SEM of a total of at least 10 cells is depicted in the graph. (Inset) Ku accumulation between 0 and 8 min in untreated cells shown in more detail. Filled circles represent untreated cells, and open triangles represent cells treated with Wortmannin. (C) Example of FRAP on a local EGFP-Ku80 accumulation. (D) FRAP on local damage curve for EGFP-Ku80. The data were normalized to the prebleach fluorescence level. The average and twice the SEM of 10 independent FRAP curves are depicted in the graph. The open circles represent the FRAP curve, and the filled circles show loss of fluorescence due to ongoing repair (from B). (Scale bars: 5 μm.)

We compared DSB repair kinetics to the EGFP-Ku80 accumulation levels at laser-induced DNA damage (19). After rapid accumulation, the Ku levels gradually decreased within the next 2 h (Fig. 2B). Approximately 20% of the initial Ku accumulation remained after 2 h, which is similar to the previously reported 20–30% of DSBs remaining at 2 h after ionizing radiation. In the presence of the DNA-PK inhibitor Wortmannin a much slower loss of fluorescence accumulation was observed, confirming that this decrease in Ku accumulation was indeed the result of ongoing DSB repair (Fig. 2B).

To get a better understanding of the interaction between Ku70/80 and DNA ends, we investigated the dynamics of the accumulated protein in the damaged area of the nucleus. Accumulated EGFP-Ku80 was bleached and fluorescence recovery was monitored (Fig. 2C). We found that the GFP signal recovered to ≈75% of the original relative level within 8 min (Fig. 2D), suggesting that Ku70/80 resided at a DNA end with a half-life of ≈2 min. The fraction of EGFP-Ku80 that did not recover within this time frame can be attributed to the decreasing number of available DNA ends because of ongoing DNA repair.

Interdependence of NHEJ Proteins for Accumulation on DSBs.

Subsequently, we extended our analysis to other NHEJ proteins. DNA-PK_CS, ligase IV, and XRCC4 were also found to accumulate at sites of laser-induced damage (Fig. 3), indicating that all core NHEJ proteins could be recruited to DSBs. Accumulation of these proteins was visible in virtually every nucleus of these exponentially growing cells, showing that the attraction to DSBs is essentially cell cycle-independent (data not shown).

Fig. 3.

Accumulation of NHEJ proteins on laser-induced damage. The immunofluorescent staining was done with antibodies to T2609 phosphorylated DNA-PK_CS in HeLa cells or XRCC4 in V79B hamster cells. EGFP-ligase IV was directly imaged in primary human fibroblasts derived from a SCID patient with a ligase IV mutation that renders the endogenous protein unstable. (Scale bars: 5 μm.)

Subsequently, we investigated the interdependence of NHEJ proteins for accumulation on DSBs. Accumulation of Ku in the damaged area does not depend on the presence of DNA-PK_CS or XRCC4, as expected (Fig. 4A and B). The ligase IV/XRCC4 complex acts late in the NHEJ reaction, suggesting that it may be recruited by interaction with DNA-PK. We found that XRCC4 accumulation in damaged areas indeed colocalized with the GFP signal in EGFP-Ku80-expressing XR-V15B cells and that cells with undetectable expression levels of EGFP-Ku80 did not show any visible accumulation of XRCC4 (Fig. 4C). Interestingly, we observed that the DNA-PK_CS protein was not required for XRCC4 accumulation (Fig. 4D).

Fig. 4.

Interdependence of NHEJ protein assembly on DSBs. (A and B) Accumulation of Ku80 (green) in XRCC4-deficient XR-1 cells (A) and in DNA-PK_CS mutant XR-C1 cells (B). (C and D) Accumulation of XRCC4 protein in XR-V15B cells expressing EGFP-Ku80 (C) and in DNA-PK_CS-deficient XR-C1 cells (D). The arrowheads in C point to cells that do not express EGFP-Ku80. (Scale bars: 5 μm in A, B, and D and 20 μm in C.)

Direct Interaction Between Ku70/80 and XRCC4.

We confirmed the previous finding that Ku70/80 interacts with the ligase IV/XRCC4 complex by performing XRCC4 immunoprecipitations of HeLa cell nuclear extracts (20). Approximately 4% of Ku70/80 coimmunoprecipitated with XRCC4 irrespective of ionizing radiation treatment before extract preparation (Fig. 5A), showing that this interaction is not affected by DSB induction.

Fig. 5.

Direct interaction between Ku70 and XRCC4. (A) Immunoprecipitation from HeLa cell nuclear extracts using XRCC4 antisera. (Upper) The input material (10%). (Lower) A Western blot of the immunoprecipitated material. C, unirradiated cells; IR, cells irradiated with 45 Gy of γ-rays. (B) Schematic representation of the trifunctional cross-linker sulfo-SBED. (C) Cross-linked polypeptides after incubation of sulfo-SBED-labeled Ku70/80 with XRCC4, followed by photoactivation of the cross-linker. UV cross-linking was performed with only Ku70/80 (lane 1), only XRCC4 (lane 2), or Ku70/80 and XRCC4 (lanes 3 and 4). The products in lane 3 were treated with DTT to reverse the cross-link. The sizes of the molecular mass markers are depicted on the left, and the nature of the various products in the gel have been confirmed by mass spectrometry (MALDI-TOF). (D) Products that could be attached to the aryl azide group of sulfo-SBED that had been linked to Ku70/80. Mixtures contained Ku70/80, DNA-PK_CS, ligase IV/XRCC4, and DNA. Lane 1 shows the Ku70/80 preparation alone, and lane 2 shows unbound proteins. Lane 4 shows the products that were precipitated by using streptavidin beads after activation of the aryl azide group and treatment with DTT, with lane 3 showing streptavidin-bound products without activation or DTT treatment.

Previous experiments have shown that the Ku–XRCC4 interaction was weak or absent (20). We therefore developed a photo-cross-linking affinity label transfer method, which allows detection of transient interactions. We labeled the Ku70/80 heterodimer with the photo-cross-linking reagent Sulfo-SBED, which contains a sulfo-NHS ester (which can form a covalent linkage to lysine residues), an aryl azide group (which can be photoactivated) and a biotin moiety (Fig. 5B). The Ku70/80/Sulfo-SBED complex was incubated with HeLa cell nuclear extracts, followed by mass spectrometric analysis of cross-linked proteins. Upon photoactivation, a covalent linkage between interacting proteins located within a distance of 20 Å can be formed. We found the preferential retention of XRCC4 (data not shown), suggesting that Ku70/80 directly interacts with XRCC4. Subsequently, recombinant XRCC4 was incubated with the Ku/Sulfo-SBED complex. We found that Ku70/80 interacted with XRCC4 in the absence of DNA-PK_CS, ligase IV and DNA (Fig. 5C). In addition to Ku70/80 complexes, we found a cross-linked product in which both Ku70 and XRCC4 peptides could be identified by mass spectrometry (Fig. 5C, lane 4). We then investigated the cross-linking pattern when DNA-PK_CS and the XRCC4/ligase IV complex were present. In this experiment, the cross-linker arm that was originally attached to Ku70/80 was cleaved by treatment with DTT after photo-cross-linking. The biotin moiety on the third arm of the trifunctional cross-linker was used to purify any attached proteins, which were then analyzed by SDS/PAGE. Even in this case, mainly XRCC4 was cross-linked, although all XRCC4 was in complex with ligase IV (Fig. 5D). In conclusion, our combined in vivo and in vitro data reveal that XRCC4 directly interacts with Ku70, which may be required for all productive NHEJ complexes.

Discussion

We have investigated the assembly of NHEJ complexes under both in vivo and in vitro conditions. We observed that recruitment of Ku70/80 on DSBs does not require the presence of any other core NHEJ factors and starts within seconds after DNA damage induction. The accumulated Ku heterodimers could exchange with the soluble fraction, showing that Ku on DNA ends is in dynamic equilibrium with protein in solution. We found that Ku70/80, but not DNA-PK_CS, is required for XRCC4 accumulation at DSBs, which is explained by our observation of a direct interaction between the XRCC4 and Ku70 polypeptides. Taken together these results support the view that the core of all NHEJ complexes revolves around the Ku–XRCC4 interaction.

Induction of DSBs with a Laser.

Historically, DSBs have been introduced by ionizing radiation or radiomimetic drugs. Recently, various types of high-power lasers have been used to induce DNA damage in cells (21). The typical approach to locally induce DNA damage in cells makes use of a UV-A laser, often in combination with DNA sensitizing agents. These approaches have been valuable to investigate DNA repair. However, presensitization of DNA can potentially influence the DNA repair processes under investigation (21). Furthermore, variable amounts of DNA damage will be produced in the nucleus along the laser beam path (not just at the focal point), with the possibility of creating unwanted structural damage to the cellular membranes. For these reasons, we chose to irradiate cells with a pulsed NIR (800 nm) laser without the use of DNA sensitizing agents. Living tissues absorb very little energy at that wavelength (22), which should keep phototoxicity and heat accumulation to a minimum (17). Moreover, the multiphoton events, which are expected to be necessary for induction of DSBs, will only happen close to the laser beam's focal point (17).

Several of our experiments support the notion that this method of DSB induction can be used to study NHEJ in vivo. First, all NHEJ proteins that we tested were recruited to the damaged area. Second, XRCC4 accumulation depended on the presence of Ku70/80, ruling out aspecific aggregation as the cause for the laser-induced accumulation. Third, Ku70/80 accumulation was dynamically maintained, which also argues against aspecific events. Finally, decrease of Ku accumulation and repair of ionizing radiation-induced DSBs show similar kinetics and can be blocked by the DNA-PK inhibitor Wortmannin (19). Therefore, we conclude that NIR-induced local DNA damage is a bona fide model system to study DSB repair mechanisms in living cells.

Interaction of Ku70/80 with DNA Ends.

We found a homogeneous distribution of EGFP-Ku80 throughout the nucleus (except the nucleolus), and FRAP analysis showed free movement of Ku70/80 through the nucleoplasm. This latter observation is markedly different from a recently published study that invokes association of Ku proteins with the nuclear matrix (23). Although the cause for this discrepancy remains to be resolved, it might be explained by different expression levels (we used stable clones, as opposed to transient expression), a different form of the GFP-fusion protein (N-terminal tagging in this study versus C-terminal tagging) and/or a difference in the method of FRAP analysis (we used strip-FRAP instead of spot-FRAP). Although our study was conducted in hamster cells, we obtained very similar results in HeLa cells (data not shown). Because our study makes use of a tagged protein that is fully active in NHEJ and contains an internally controlled comparison with other EGFP-tagged proteins that have been measured with the same microscope settings, we conclude that Ku70/80 diffuses relatively unhindered in the nucleoplasm until it encounters a DNA end.

Our analysis of EGFP-Ku80 accumulation on laser-induced DSBs indicates that NHEJ complex assembly in vivo is a very dynamic process (see model in Fig. 6). Indeed, photobleaching on locally accumulated Ku70/80 revealed that the bound proteins could be exchanged within 10 min, suggesting that NHEJ complex assembly is reversible. Although we cannot formally exclude the possibility that (a fraction of) Ku70/80 binds indirectly to the DNA ends, several observations argue that our analysis reflects a property of Ku heterodimers bound to DNA ends. First, both accumulation and FRAP curves are compatible with the association/dissociation kinetics one expects when only one binding substrate exists (e.g., monoexponential behavior with no delay in the accumulation). Second, if Ku could not dissociate from DNA ends before ligation has taken place, this would result in an immobile fraction in the FRAP experiments, unless all repair had been completed within the 15 min measuring period. However, such an immobile fraction was not observed and from DSB repair kinetics after ionizing radiation one can conclude that DSB repair takes much more time for completion. However, a fraction of DNA end bound Ku below 10% of the total amount of immobilized Ku heterodimers would go undetected in our analysis. Still, we do not favor the idea that a large fraction of Ku molecules binds indirectly to a single DSB, because accumulation of ≈50–100 GFP-Ku molecules would be visible as nuclear foci (24), which have never been observed.

Fig. 6.

Working model for joining of various types of DNA DSBs. First, DSB recognition by Ku70/80 is fast and reversible. Subsequently, XRCC4/ligase IV can be attracted, forming a reversible complex. Alternatively, DNA-PK_CS and XRCC4/ligase IV can both be recruited and form a more stable complex. We hypothesize that the joining reaction of simple DSBs can be accomplished via either of these two intermediates, whereas complex DSBs can be joined only via the right branch.

Assembly of NHEJ Complexes in Vivo.

Our data support a model in which Ku70/80 directly mediates incorporation of ligase IV/XRCC4 into end-joining complexes. DNA-PK_CS and DNA-processing enzymes can be recruited to these core complexes, as required. The XRCC4 protein most likely functions as a versatile adaptor to attract and position the ligase, as well as other DNA-processing proteins, in a way that is quite similar to the role of XRCC1 in base excision repair (25). An interaction of XRCC4 with poly nucleotide kinase has been shown recently, suggesting that XRCC4 may also serve to attract processing factors, which would place XRCC4 at the very heart of the NHEJ machinery rather than just serving as a cofactor for ligase IV (26). This view is certainly consistent with the severe phenotype of the XRCC4-deficient mice and cells (27, 28), even for joining of simple blunt DSBs (29). The adaptor function of XRCC4 would also explain why the conservation of this protein is so low between yeast and mammals (30), despite its important function: the divergent interactions in yeast and mammals would require a different amino acid composition, even though the adaptor function is highly conserved.

Prior studies on interactions between NHEJ factors have found that Ku70/80 primarily interacted with ligase IV and DNA-PK_CS with XRCC4 (20). It is possible that the interaction between Ku70/80 and XRCC4 is not stable under conditions used by Hsu et al. However, it is clear that XRCC4 is attracted to DSBs in vivo in the absence of DNA-PK_CS. The attraction of the yeast Dnl4/Lif1 complex in the absence of a DNA-PK_CS homolog (31) and the relatively mild phenotype of DNA-PK_CS deficiency is also consistent with a minor role for this protein in attracting the ligase IV/XRCC4 complex (19, 29). However, it is to be expected that interactions of ligase IV with DNA and/or proteins contribute to the stability of NHEJ complexes.

An important issue that has not been resolved completely yet is the order of NHEJ protein assembly at a DSB in vivo. The DNA-PK_CS and XRCC4 proteins are expected to bind to DNA end bound Ku70/80 independently, although a stabilizing effect of DNA-PK_CS on ligase IV/XRCC4 in NHEJ complexes would not be unexpected (32). Thus, NHEJ factors need not be recruited in a fixed order and stable NHEJ complexes may result from different assembly routes.

The use of GFP-tagged proteins provides a means to determine the relative strength of interactions between these factors at DNA ends in vivo. For example, basic kinetic parameters can be estimated by using the accumulation and FRAP on local damage experiments in vivo (unpublished data) and the k_on and k_off values can be compared with results of in vitro binding studies. Furthermore, precise evaluation of in vivo binding affinities in various genetic backgrounds will allow correlation of NHEJ complex assembly and stability with the repair efficiency of the NHEJ machinery.

Materials and Methods

Plasmids and Cell Lines.

Ku80 cDNA was amplified by RT-PCR from HeLa cell mRNA with the primers DG135 (5′-CGGGATCCATGGTGCGGTCGGGGAATAAGG) and DG136 (5′-TCTAGTCGACCTATATCATGTCCAATAAATCGTC). This PCR product was digested with BamHI and SalI and ligated into the BglII and SalI sites of pEGFP-C1 (Clontech, Mountain View, CA). Stable EGFP-Ku80-expressing XR-V15B clones (33) were tested for expression of endogenous (hamster) Ku80 by RT-PCR using the primers HTB31 (5′-GATTGTAGCCTATAAATCGATC) and HTB32 (5′-CAAAATGTGCTGCTGAATTGG). Other cell lines used in this study were DNA-PK_CS-deficient XR-C1 (34), XRCC4-deficient XR-1 (35), HeLa cells, and ligase IV-deficient SC2 human fibroblasts expressing EGFP-ligase IV (36).

Cell Lysis, Immunoprecipitation, and Immunoblotting.

Cells were sonicated in lysis buffer [20 mM sodium acetate, pH 7.4/350 mM NaCl/5 mM MgCl₂/5% glycerol and protease inhibitor mixture (Roche, Basel, Switzerland)] and cleared by centrifugation. Proteins were separated by SDS/PAGE, electroblotted, and detected with primary anti-Ku70 (M-19) or anti-Ku86 (M-20) goat polyclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) or ab290 anti-GFP rabbit polyclonal antibody (Abcam, Cambridge, U.K.). For immunoprecipitations, 100 μg of total protein was diluted in 1 ml of lysis buffer, precleared with 25 μl of protein A beads (Amersham, Piscataway, NJ), incubated with 1 μg of anti-Ku70 Ab-4 (Clone N3H10; Neomarkers, Fremont, CA) for 5 h at 4°C, precipitated with 50 μl of protein A beads (50% slurry in lysis buffer) for 2 h at 4°C, washed three times with lysis buffer, and analyzed by Western blotting.

Functional Assays.

The end joining and V(D)J recombination assays were performed as described previously (37). For the survival assays, cells were seeded onto 6-cm dishes and allowed to attach for 12 h before receiving a single dose of γ-radiation from a ¹³⁷Cs source. The cells were grown for 5 days, and colonies were counted.

DNA Damage Induction in Live Cell Quantitative Imaging Experiments.

Induction of DSBs in cultured cells was obtained by exposure to 75 Gy of γ-radiation (0.75 Gy/min ¹³⁷Cs source). For laser-induced DNA damage, a Coherent Verdi pump laser with a Mira 900 mode locked Ti:Sapphire laser system (Coherent, Santa Clara, CA) was directly coupled to a LSM 510 NLO microscope (Carl Zeiss, Jena, Germany) to obtain an 800-nm pulsed output (200-fs pulse width at 76 MHz, 10-mW output at the sample). To target a large number of nuclei, several adjacent fields were scanned by the NIR laser in a pattern of evenly spaced parallel lines. For single nuclei measurements, an 86-ms exposure restricted to a 2.5-μm circular region was used to induce DNA damage. When indicated, DRAQ5 (Biostatus, Shepshed, U.K.) was added at 10 μM after damage induction. For live cell imaging to detect cell division, 5 mM caffeine was added 2 h after damage induction to override the G₂/M checkpoint.

Immunofluorescence Assays.

Cells were grown on glass coverslips, washed twice with PBS containing 0.05% Triton X-100, and fixed with 2% paraformaldehyde/0.025% Triton X-100 at 37°C. Coverslips were washed four times with PBS containing 0.1% Triton X-100 and once with PBS⁺ (PBS containing 0.15% glycine and 0.5% BSA). Cells were incubated at room temperature with primary antibodies to phosphorylated histone H2AX (Upstate Biotechnology, Charlottesville, VA), Ku80 (C-20; Santa Cruz Biotechnology), DNA-PK_CS phosphorylated on T2609 (38) or XRCC4 (NIH13) (39) in PBS⁺ for 90 min at room temperature. After two washes with PBS with 0.1% Triton X-100 and one with PBS⁺, coverslips were incubated for 1 h with secondary antibodies in PBS⁺. Excess antibody was washed away by two 10-min washes with PBS/0.1% Triton X-100, and cells were mounted with Vectashield including DAPI (Vector Laboratories, Burlingame, CA). TUNEL staining was carried out as described by the manufacturer (In Situ Cell Death Detection Kit; Roche).

Microscopy and Quantitative Fluorescence Measurements.

Imaging and FRAP experiments were performed on a LSM 510 NLO multiphoton confocal microscope (Carl Zeiss, Jena, Germany) The acquired digital images were processed and analyzed with ImageJ (40). Imaging of immunofluorescence assays involved the sequential use of 405-, 488-, and 561-nm lasers for the excitation of DAPI, EGFP, and Alexa Fluor 594 dyes. No apparent “bleed-through” was detected between the different detection channels.

The accumulation kinetics of EGFP-Ku80 on laser-induced DNA damage was obtained by measuring the average fluorescence within the targeted area. The fluorescence before damage induction was set to 0, and the data were normalized to the maximum EGFP-Ku80 accumulation level (see Supporting Text, which is published as supporting information on the PNAS web site, for details). FRAP on local damage was performed after a 10-min accumulation period. The background fluorescence was subtracted and the data were normalized to the prephotobleached fluorescence. When indicated, 20 μM Wortmannin (Sigma, St. Louis, MO) was added 45 min before damage induction.

Protein Expression, Purification, and Cross-Linking.

Ku70/80 was expressed in insect cells and purified by using Ni²⁺ Sepharose and double-stranded DNA cellulose columns as described (41). XRCC4 and ligase IV/XRCC4 complexes were expressed in Escherichia coli and purified as described (39). Chemical cross-linking was carried out by using the trifunctional cross-linker sulfo-SBED (Pierce, Rockford, IL) according to the manufacturer's instructions. Briefly, the Ku70/80 was dialyzed against 20 mM Hepes (pH 7.5), 100 mM KCl, 1 mM DTT, and 10% (vol/vol) glycerol and incubated with 10-fold molar excess of sulfo-SBED for 2 h at 4°C. Excess cross-linker was quenched with 20 mM Tris·HCl (pH 7.5), and proteins were stored at −80°C until use. This protein preparation was incubated with XRCC4 or protein extracts for 30 min on ice followed by cross-linker activation with 365-nm light for 5 min at room temperature. Proteins were separated by SDS/PAGE, stained, and analyzed by MALDI-TOF mass spectrometry, as described (42).

Abbreviations

FRAP

fluorescence recovery after photobleaching

NHEJ

nonhomologous end-joining

DSB

double-strand break

DNA-PK

DNA-dependent protein kinase

NIR

near IR.