The XRCC4 Gene Product Is a Target for and Interacts with the DNA-dependent Protein Kinase*

The gene product of XRCC4 has been implicated in both V(D)J recombination and the more general process of double strand break repair (DSBR). To date its role in these processes is unknown. Here, we describe biochemical characteristics of the murine XRCC4 protein. XRCC4 expressed in insect cells exists primarily as a disulfide-linked homodimer, although it can also form large multimers. Recombinant XRCC4 is phosphorylated during expression in insect cells. XRCC4 phosphorylation in Sf9 cells occurs on serine, threonine, and tyrosine residues. We also investigated whether XRCC4 interacts with the other factor known to be requisite for both V(D)J recombination and DSBR, the DNA-dependent protein kinase. We report that XRCC4 is an efficient in vitro substrate of DNA-PK and another unidentified serine/threonine protein kinase(s). Both DNA-PK dependent and independent phosphorylation of XRCC4 in vitro occurs only on serine and threonine residues within the COOH-terminal 130 amino acids, a region of the molecule that is not absolutely required for XRCC4’s DSBR function. Finally, recombinant XRCC4 facilitates Ku binding to DNA, promoting assembly of DNA-PK and complexing with DNA-PK bound to DNA. These data are consistent with the hypothesis that XRCC4 functions as an alignment factor in the DNA-PK complex.

1 and RAG 2) (3,4) are directly responsible for initiation of the V(D)J recombination reaction (5)(6)(7)(8) together directing cleavage at the recombination signal sequence-coding juncture; whereas resolution of recombination intermediates also requires several ubiquitously expressed DNA repair factors. Evidence linking DNA repair activities with V(D)J recombination arose from the characterization of cells from homozygous severe combined immune deficient (SCID) mice (9 -11). C.B-17 SCID mice which are immunodeficient because of defective V(D)J recombination, were shown to also be defective in repairing DSBs in lymphoid and non-lymphoid cells (12,13). Further evidence for an overlap in the activities of V(D)J recombination and DSB repair came from the observation that radiosensitive rodent cell lines which were defective in DSB repair could not support V(D)J recombination induced by co-transfecting RAG 1 and RAG 2; whereas, radiosensitive mutants proficient in DSB rejoining were normal in this capacity (14 -16). Four factors have been delineated which are required for both V(D)J recombination and DSBR; three of these encode components of the DNA-dependent protein kinase (17)(18)(19)(20)(21)(22)(23)(24)(25). DNA-PK is a nuclear serine/ threonine protein kinase whose catalytic activity is dependent upon its association with DNA discontinuities such as DSBs (26,27). The catalytic subunit of DNA-PK is an ϳ460-kDa protein termed DNA-PK CS and the DNA-binding component is a heterodimer, Ku which is comprised of two polypeptides (Ku70 and Ku86). Because of its unique dependence on DNA ends and other DNA discontinuities for activation, it has been proposed that the role of DNA-PK in DNA repair is to modulate, via phosphorylation, the function and/or expression of other factors in response to DNA damage (18). This hypothesis has not been tested. The fourth factor required for both V(D)J recombination and DSBR is the product of the XRCC4 gene. This factor has no homology to known proteins and its function is unknown (28) although it has recently been reported that XRCC4 interacts with DNA ligase IV (29,30). Although the functions of DNA-PK and XRCC4 in V(D)J recombination have not been elucidated, it is intuitive that their role is in the resolution of coding and signal ends generated by RAG-mediated cleavage.
In both DNA-PK-and XRCC4-defective cells, low levels of rearrangements (both coding and signal joints) can be isolated (i.e. "leaky" SCID phenotype) (31,32). Fine structure of these joints has provided insight into mechanistic details of V(D)J recombination. In SCID mice, rare coding joints often have large deletions and/or long P elements (palindromic sequences at the point of ligation which apparently result from internal nicking of the hairpinned coding intermediate) (31,32). Thus, it has been proposed that these long P elements reflect a diminution of hairpin resolution consistent with the observation that coding ends with hairpinned termini accumulate in SCID lymphocytes (33,34). Rare coding joints isolated from XRCC4 defective cells are distinct from those found in SCID mice in that there is no evidence of extensive P nucleotides. Instead, it appears that short sequence homologies at the DNA termini contribute to the resolution of both coding and signal ends in XRCC4-deficient cells (28). Although homology mediated joining is often observed in coding joints, it is not observed in signal ligations (35)(36)(37). Thus, Li et al. (28) have postulated that one function of XRCC4 may be to align free DNA ends, and in the absence of XRCC4 short sequence homologies might serve to facilitate DNA end joining.
Here we report that recombinant XRCC4 and DNA-PK interact in vitro. The multimeric nature of XRCC4 suggests a model where XRCC4 functions as an alignment factor of DNA-PK bound to DNA serving to align DNA ends during DNA repair and V(D)J recombination. These data, in conjunction with the recent reports of XRCC4's interaction with DNA ligase IV (29,30), suggest a model whereby XRCC4 targets cellular ligase activities to damaged DNA bound by the DNA-dependent protein kinase.
pAcGP67B provides a leader sequence from the viral acidic glycoprotein gp67 and should yield secreted versions of nuclear proteins. We have observed, however, that recombinant proteins expressed from this vector are usually not secreted and proteins must be purified from cell lysates using Ni-NTA resin (see "Protein Purification"). Still, this vector was used as it results in consistently higher levels of recombinant protein than can be obtained with other transfer vectors. 2 An additional recombinant virus encoding full-length XRCC4 but without the gP67 leader sequence was generated using the pH␥1360x transfer vector (38). The resulting plasmids were co-transfected with linearized AcM-NPV viral DNA via liposomes (Invitrogen, San Diego, CA) into Sf9 insect cells. Recombinant viruses were then isolated using standard techniques. Baculovirus expression resulted in a fusion protein consisting of 10 amino acids from the gp67 leader peptide, nine histidine residues, and the murine XRCC4 polypeptide. The Ku86 and Ku70 baculoviruses were the generous gifts of Dr. J. Donald Capra (University of Texas Southwestern Medical Center, Dallas, TX) and have been described previously (39).

Expression of XRCC4 in XR-1 Cells
EcoRV/NotI fragments from the baculovirus transfer vectors encoding XRCC4 fusion proteins were subcloned into the NotI and SmaI sites in the cloning cassette of plasmid pcDNAI. Resulting plasmids were co-transfected with the pNeo plasmid into the XR-1 Chinese hamster ovary mutant cell line which is defective in XRCC4 with LipofectAMINE according to the manufacturer's conditions (Life Technologies, Inc., Gaithersburg, MD). Stable lines were established by G418 selection.

Assessment of Irradiation Sensitivity
2 ϫ 10 3 cells suspended in 2 ml of complete media were exposed to various amounts of ionizing radiation using a 137 Ce source calibrated at 558.8 R/min. Immediately after irradiation, cells were seeded in 20 ml of complete media containing 10% fetal calf serum in 150-mm 2 tissue culture dishes. After 7 days, cell colonies were fixed with 2% formaldehyde followed by 100% methanol. Subsequently, the colonies were stained with trypan blue and colony numbers were assessed. Irradiation sensitivities were assessed in triplicate and are presented as percent surviving fraction based on plating efficiencies of non-irradiated controls.

Protein Purification
For protein purification, Sf9 cells which had been infected ϳ60 h earlier were collected and washed with phosphate-buffered saline. Typically, 2 ϫ 10 8 infected cells were prepared at a time. Cell pellets were frozen and thawed three times in liquid nitrogen in 10 ml of the following buffer: 20 mM Tris (pH 7.9), 0.5 M NaCl, 5 mM imidazole, and 0.5 mM phenylmethylsulfonyl fluoride. The lysates were centrifuged for 30 min at 8,000 ϫ g. Lysates were incubated for 1 h at 4°C with 0.5 ml of Ni-NTA resin (Qiagen, Chatsworth, CA) and the resin washed 3 times in the same buffer with the addition of 50 mM imidazole. Soluble recombinant XRCC4 protein was eluted from the resin in 0.5 ml of the same buffer with 500 mM imidazole three times. The three eluted fractions were pooled and then dialyzed against 25 mM Tris (pH 8.0), 150 mM KCl, 2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 10% glycerol at 4°C. Dialyzed material was aliquotted and stored at Ϫ70°C. 10 l of lysates, beads, and eluate were analyzed by SDS-PAGE (Fig. 1). The results depicted were typical for virus XRCC4 HIS and for ⌬C2. Purification of the other recombinant proteins described here consistently resulted in much lower protein yields. Because of the low yields from viruses ⌬N, ⌬C1, and ⌬N⌬C1, contaminating Sf9 proteins comprised a larger percentage of the total protein than with viruses XRCC4 HIS or ⌬C2. For this reason, control Ni ϩ preparations of wild type-infected Sf9 cells were also prepared and used as controls in kinase assays (Fig. 5) and EMSA assays (not shown).

In Vitro Kinase Assays
XRCC4 Assays-Typically, 1 g of purified XRCC4 was incubated with 300 ng of DNA-PK (Promega Corp., Madison, WI), and with or without 100 ng of linearized pTZ19R plasmid DNA (Pharmacia Biotech, Piscataway, NJ) in a total volume of 20 l of ZЈ0.05 (24). The mixtures were incubated with 37.5 Ci of [␥-32 P]ATP for 1 h at room temperature, then mixed with an equal volume of 2 ϫ SDS-PAGE buffer without 2-mercaptoethanol. In assays utilizing truncated forms of XRCC4, 400 ng of each recombinant protein was incubated with 300 ng of DNA-PK. Samples were loaded onto SDS-polyacrylamide gels and run at 200 volts for 45 min. Following electrophoresis, gels were dried and then exposed to Kodak X-Omat AR film. In some assays, wortmannin was included at a final concentration of 500 nM.
c-Jun assay-Approximately 300 ng of purified DNA-PK (Promega, Madison, WI) was incubated in the presence of 10 g of bovine serum albumin and 2 l of DNA-PK reaction premix (13 mM spermidine and 4 mM MgCl 2 dissolved in 0.05 ϫ phosphate-buffered saline) with 1 g of purified rhAP1 (c-Jun, Promega, Madison, WI), 2 l of 2.5 mM ATP, 0.5 l (6000 Ci/mmol; DuPont) [␥-32 P]ATP, with or without 1 g of purified XRCC4 and with or without 100 ng of linearized plasmid DNA. Samples were incubated for 10 min at 30°C, then mixed with equal volumes of 2 ϫ SDS-PAGE buffer. Samples were loaded onto an SDS-8% polyacrylamide gel and run at 200 volts for 45 min. Following electrophoresis, gels were dried and exposed to Kodak X-Omat AR film.
DNA-PK Pulldown Assay-The "DNA-PK pulldown assay" was performed essentially as described by Finnie et al. (24) using the 0176 normal equine fibroblast and 1863 SCID equine fibroblast cell lines (40). Briefly, cell pellets (1 ϫ 10 7 ) were resuspended in 100 l of extraction buffer and then frozen in liquid nitrogen and thawed at 37°C three times. The extract was microcentrifuged at 15,000 rpm for 7 min at 4°C and the supernatant was frozen at Ϫ70°C. Extracts containing ϳ7 mg of protein were absorbed onto DNA-cellulose beads and washed. 400 ng of XRCC4 was added to the mixture and incubated for 1 h at room temperature. Following incubation, an equal volume of 2 ϫ SDS-PAGE sample buffer was added and the phosphorylated proteins were analyzed on SDS-8% polyacrylamide gels.
Phosphoamino Acid Analysis-Kinase assays were performed as above. Following polyacrylamide gel electrophoresis, the gel was transferred to polyvinylidene difluoride paper and exposed to film. Small strips of polyvinylidene difluoride paper which corresponded to bands of interest on the autoradiograph were excised and wet in methanol for 30 s and 1 min in water. The samples were incubated in 200 l of 6 N HCl at 110°C for 1-2 h and dried overnight. Pellets were counted in a scintillation counter and resuspended in acid (2.2% formic acid, 7.8% acetic acid) to a concentration of 2000 cpm/l. 2 l each of phosphoserine, threonine, and tyrosine standards (1 mg/ml) were mixed with 1 l of sample and spotted onto a thin layer chromatography plate. The plate was then electrophoresed at 1200 volts constant voltage at 4°C for 2 K. Meek, unpublished data. The plate was dried, sprayed with 0.3% ninhydrin in acetone, and exposed to film.

Metabolic Labeling of Baculovirus-infected Cells
Sf9 cells were infected with either wild type baculovirus or recombinant baculoviruses in SF900 serum-free media (Life Technologies, Inc., Gaithersburg, MD). After 24 h, the media was replaced with Grace's insect media containing [ 32 P]orthophosphate (200 mCi/ml). Cells were harvested 24 h later. Metabolically labeled XRCC4 was affinity purified by Ni ϩ chromatography as described above.

Electrophoretic Mobility Shift Assays (EMSAs)
EMSAs were done by the method of Landolfi et al. (41) with the following modifications. Binding reactions were done in 0.5 ϫ buffer C (20 mM HEPES, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 ) instead of buffer D, no nonspecific competitor was added, and bovine serum albumin was included at a concentration of 3 mg/ml. The probe was a double strand oligonucleotide that included a 12-base pair spacer V(D)J recombination signal sequence and had the following sequence: 5Ј-GATCTGGCCTGTCTTACACAGTGCTACAGACTGGAACAAAAAC-CCTGCAG-3Ј (8). Completely analogous results were obtained with a second unrelated oligonucleotide probe derived from a T cell receptor V␥ promoter. Binding reactions were assembled on ice and then incubated at room temperature for 30 min. For EMSAs examining the ability of XRCC4 to bind DNA directly or to affect Ku binding to DNA, 5 g of purified XRCC4 was used without or with 1, 5, or 10 g of baculovirusexpressed Ku purified by absorption to DNA-cellulose (39). Rabbit anti-Ku antisera was the generous gift of Dr. J. Donald Capra (University of Texas Southwestern Medical Center, Dallas, TX). For EMSAs examining DNA-PK assembly, 4, 2, 1, 0.5, or 0.25 g of purified full-length XRCC4 fusion protein or purified ⌬C2 were incubated with 60 ng of DNA-PK (Promega Corp., Madison, WI). Protein-DNA complexes were resolved on 4% native polyacrylamide gels in 0.5 ϫ TBE.

Characterization of Baculovirus Expressed Murine
XRCC4 -To assess the biological role of XRCC4, expression of this protein in the baculovirus system was undertaken. To that end, several different 9X-HIS tagged versions of the XRCC4 cDNA (28) were generated and inserted in the pAcGP67B or pH␥1360x transfer vectors. Recombinant viruses were isolated using standard techniques (see "Materials and Methods"). These are diagrammatically depicted in Fig. 1A. Constructs XRCC4 HIS and ⌬C2 expressed relatively high levels of recombinant protein. Constructs ⌬C1, ⌬N, and ⌬N⌬C1 consistently expressed considerably lower levels of recombinant protein.
Ni ϩ -agarose affinity chromatography was used to purify recombinant XRCC4 expressed in insect cells. When expressing fulllength XRCC4, a prominent ϳ57-kDa band was consistently apparent in lysates of Sf9-infected cells under reducing conditions (Fig. 1B, pre-absorption). Ni ϩ -agarose absorption dramatically reduced the ϳ57-kDa band in cell lysates (Fig. 1B, postabsorption), but the band was apparent in the population of proteins bound to the Ni ϩ -agarose (Fig. 1B, bound). XRCC4 HIS was efficiently eluted from the Ni ϩ -agarose using imidazole (Fig.  1B, eluted). Identification of the ϳ57-kDa protein as the XRCC4 HIS fusion protein was confirmed by amino acid sequence analysis. In addition, mass spectrometric analysis indicated that the mass of the protein was actually ϳ39 kDa, consistent with the predicted mass of the XRCC4 HIS fusion protein (39.5 kDa). The anomalous mobility of recombinant XRCC4 HIS in SDS-PAGE is likely due to its low isoelectric point (5.6).
We next addressed whether the addition of the short leader sequence from gP67 and the 9X-HIS tag or truncations of XRCC4 altered the normal function of XRCC4. Thus, mammalian expression vectors encoding fusion proteins XRCC4 HIS , ⌬N, ⌬C1, ⌬N⌬C1, and ⌬C2 ("Materials and Methods") were prepared. These plasmids were stably transfected into the XR-1 cell line and tested for their capacity to complement the XR-1 defect by testing relative hypersensitivity to ionizing radiation as compared with the parental XR-1 cell line and normal Chinese hamster ovary cells. As can be seen in Fig. 1C, the full-length XRCC4 expressed as a gp67/9X-HIS fusion protein substantially reversed the radiation sensitivity of the XR-1 cell line. Similarly, the plasmid encoding the ⌬C1 fusion protein also reversed the radiation hypersensitivity of XR-1. The plasmids encoding the ⌬N, ⌬N⌬C1, and ⌬C2 fusion proteins did not reverse the defect in the XR-1 cell line to a significant degree. These data demonstrate a functional core of murine XRCC4 spanning amino acids 1-204.
Purified XRCC4 HIS Is a Phosphoprotein That Exists Primarily as a Homodimer but Which Can Form Large Multimers-In our initial analyses of the recombinant XRCC4 HIS fusion protein, it was apparent that the protein: 1) existed as dimers and multimers, and 2) was constitutively phosphorylated during expression in Sf9 cells. To illustrate these points, Fig. 2 presents analyses in both native polyacrylamide gels (right panel) and SDS-PAGE (left panel), of Ni ϩ -agarose purified XRCC4 HIS isolated from [ 32 P]orthophosphate metabolically labeled Sf9 cells. As a control, Ni ϩ -agarose fractions of metabolically labeled wild type-infected cells are also shown. Under nonreducing conditions in SDS-PAGE, XRCC4 HIS migrates in several heterogeneous forms (Fig. 2, top left panel, XRCC4 Ϫ2ME). ϳ130 kDa, although a small amount of the protein was consistently present with an apparent mass of ϳ57 kDa. This suggests that XRCC4 HIS exists primarily as a disulfide-linked homodimer. In agreement with this interpretation, mass spectrometric analysis of nonreduced XRCC4 HIS in SDS buffer showed that the major species had an actual mass of ϳ78 kDa (data not shown). There are also three additional bands with electrophoretic mobilities slower than the 207-kDa marker. It is unlikely that these larger bands represent contaminants because under reducing conditions, the protein migrates as a single homogeneous species (Fig. 2, top left panel, XRCC4  ϩ2ME). Furthermore, the Ni ϩ fractions of wild type-infected cells include no proteins with similar electrophoretic mobilities. We considered that addition of the short leader sequence from gp67 might alter normal trafficking of XRCC4 HIS resulting in inappropriate disulfide bonding of the recombinant protein; thus an additional virus was generated that did not include the leader sequence from gp67. Although XRCC4 expression was considerably lower from this virus, mobility of this protein on nonreducing SDS-PAGE was analogous to that observed with the initial recombinant XRCC4 HIS protein (data not shown). Thus, XRCC4 HIS exists primarily as a homodimer, but it can also form covalent multimers. Furthermore, both dimeric and multimeric forms of the XRCC4 HIS fusion protein are constitutively phosphorylated in Sf9 cells (Fig. 2, lower left panel, XRCC4 Ϫ2ME). We also analyzed the metabolically labeled protein in native polyacrylamide gels. As can be seen (Fig. 2,  right panel, XRCC4), nondenatured XRCC4 HIS migrates as two distinct species: one which migrates between the 101 and 207 kDa markers, and one which migrates just above the 207-kDa marker. There is considerably less heterogeneity in native XRCC4 HIS as compared with the nonreduced, denatured protein; it is possible that this heterogeneity may be due to incomplete disulfide bonding of dimers and/or multimers. In any case, the mobilities of nondenatured XRCC4 HIS in native polyacrylamide gels are consistent with homodimers and homotetramers.
We considered that the protein might also be modified by glycosylation. The glycosylation status of XRCC4 HIS was assessed first by including tunicamycin (which specifically inhibits N-linked glycosylation) in the media during virus infection; addition of tunicamycin in the expression culture had no effect on the mobility of XRCC4 HIS in SDS-PAGE (not shown). Certain proteins that have the capacity to multimerize are modified by O-linked glucosamine residues (42). Two techniques were used to assess O-linked sugar moieties: affinity of XRCC4 HIS for wheat germ agglutinin and metabolic labeling of XRCC4 HIS virus-infected Sf9 cells with [ 3 H]glucosamine. XRCC4 had no specific affinity for wheat germ agglutinin and was not labeled by [ 3 H]glucosamine; in summary, we found no evidence of O-linked glycosylation (data not shown). We therefore conclude that XRCC4 HIS expressed in insect cells is not modified by N-linked sugars, and is probably not modified by O-linked sugars.
DNA-PK and Other Kinases Phosphorylate XRCC4 in Vitro-In the initial analysis of the XRCC4 gene, a potential DNA-PK phosphorylation motif was noted (28). Because of the link between DNA-PK and XRCC4, we investigated whether purified XRCC4 HIS was a substrate for DNA-PK. XRCC4 HIS was phosphorylated when it was incubated in vitro with purified DNA-PK and linearized plasmid DNA (Fig. 3A). Both dimeric and multimeric forms of XRCC4 HIS are phosphorylated. Phosphorylation was largely, but not entirely, dependent on the presence of linear DNA (Fig. 3A, ϩDNA versus ϪDNA). This is consistent with authentic DNA-PK mediated phosphorylation. However, the DNA-PK pulldown assay using SCID extracts, which lack DNA-PK CS , demonstrate that other kinases are also able to phosphorylate purified XRCC4 HIS in vitro as well.
The DNA-PK pulldown assay involves fractionating cellular extracts by binding them to DNA-cellulose and then determining the ability of the bound cellular fraction to phosphorylate a suitable substrate (24). Extracts from normal and SCID equine fibroblast cell lines were used in this assay with purified XRCC4 HIS as the potential phosphorylation substrate (Fig.  3B). Purified XRCC4 HIS was phosphorylated by extracts from SCID fibroblasts albeit to a lesser degree than by extracts from normal fibroblasts. In contrast, another DNA-PK substrate, the p53 peptide spanning the known DNA-PK site, was only phosphorylated by the normal equine SCID fibroblast extracts (Ref. 40, data not shown). Completely analogous results were obtained using extracts derived from normal and SCID mouse fibroblasts, as well as unfractionated equine and murine SCID fibroblast extracts (data not shown). Thus, it is clear that in vitro, XRCC4 HIS is a substrate for DNA-PK as well as another unidentified protein kinase(s).
To further support this conclusion, we next assessed DNA-PK dependent and independent phosphorylation of XRCC4 HIS in the presence of wortmannin (Fig. 3C). (Wortmannin specifically inhibits phosphorylation by phosphatidylinositol 3-kinase family members but not phosphorylation by other protein kinases (43).) As can be seen, DNA-dependent phosphorylation of XRCC4 HIS is completely inhibitable by wortmannin (compare lanes 1 and 2) whereas DNA independent phosphorylation of XRCC4 HIS is not diminished in the presence of the inhibitor (compare lanes 3 and 4). These data demonstrate that XRCC4 HIS is an in vitro substrate for two (or more) protein kinases, one which is inhibitable by wortmannin (DNA-PK) and another which is wortmannin resistant.
Finally, we assessed XRCC4 HIS phosphorylation by DNA-PK compared with phosphorylation of c-jun, an established DNA-PK substrate (Fig. 3D). Recombinant c-jun is efficiently phosphorylated in vitro by DNA-PK in the presence (lane 3) but not in the absence of DNA (lane 1). When equal amounts (1 g) of both XRCC4 HIS and c-jun are provided as substrates in the presence of DNA, phosphorylation of both proteins is approximately equivalent.
Phosphoamino Acid Analysis-We next performed phosphoamino acid analysis of XRCC4 HIS which had been phosphorylated in vitro or in Sf9 cells by metabolic labeling with [ 32 P]orthophosphate. In vitro XRCC4 HIS was phosphorylated in the presence of DNA-PK, with or without linearized DNA. The phosphorylated proteins were hydrolyzed and then analyzed by thin layer chromatography (Fig. 4). As can be seen, phosphorylation of XRCC4 HIS in vivo occurs on serine, threonine, and tyrosine residues. In contrast, both DNA-PK dependent and independent phosphorylation of XRCC4 HIS in vitro occurs entirely on serine and threonine residues. Phosphotyrosine is consistently undetectable from XRCC4 HIS phosphorylated in vitro. The ratio of serine to threonine phosphorylation is 3:1 in vitro in the presence or absence of DNA as well as in Sf9 cells. The ratio of serine to tyrosine phosphorylation in Sf9 cells is 1:1.
Localization of XRCC4 Phosphorylation Sites-We next assessed phosphorylation of the truncated forms of XRCC4 (illustrated in Fig. 1A). Recombinant proteins were isolated via Ni ϩ -Sepharose chromatography; purified proteins (400 ng) were then assessed as DNA-PK substrates (Fig. 5). Since expression levels of the ⌬N, ⌬C1, and ⌬N⌬C1 fusion proteins were relatively low, Ni ϩ affinity purified proteins were less pure than with full-length XRCC4 HIS . Thus, a Ni ϩ fractionation of wild type baculovirus-infected cells was also included as a negative control (Fig. 5, lanes 1 and 2). Of note, the only "SQ" potential DNA-PK site in both human and murine XRCC4 is at amino acid 53 (illustrated in Fig. 1A); there is a second SQ site in murine XRCC4 at amino acid 92. As can be seen, only full-length XRCC4 HIS and the ⌬N fusion protein were phosphorylated in vitro. Phosphorylation of ⌬N occurred in the absence of linearized DNA but was augmented in the presence of linearized DNA. In contrast, neither ⌬C1 nor ⌬C2 were phosphorylated in the presence or absence of DNA when incubated in vitro with purified DNA-PK (Fig. 5) even though each spans the consensus SQ DNA-PK sites at amino acids 53 and 92. In addition, ⌬N/⌬C1 was also not phosphorylated in vitro.
Finally, metabolic labeling of Sf9 cells infected with each of the recombinant viruses revealed that only full-length XRCC4 HIS and ⌬N were phosphorylated in Sf9 cells (data not shown). As with the in vitro kinase experiments, the ⌬C1, ⌬C2, and ⌬N⌬C1 fusion proteins were not detectably phosphorylated during expression in Sf9 cells. Phosphoamino acid analysis of ⌬N revealed that phosphorylation occurred on serine, threonine, and tyrosine residues in Sf9 cells. The most straightforward interpretation of these data is that DNA-PK independent and dependent serine/threonine phosphorylation of XRCC4 HIS occurs between amino acids 204 and 334 and that DNA-PK dependent phosphorylation occurs on a non-SQ site as has been reported for several other DNA-PK substrates (44). (Several potential serine and threonine phosphorylation sites for casein kinases were noted previously between amino acids 204 and 334, illustrated in Fig. 1A.) Similarly, tyrosine phosphorylation of XRCC4 HIS likely occurs in the same region. The only tyrosine between amino acids 204 and 326 is at position 227 (the tyrosine previously noted as a potential target for cytosolic tyrosine kinases, illustrated in Fig. 1A (28)). Alternatively, it is possible that phosphorylation occurs between amino acids 65 and 204, but is dependent on sequences between 204 and 326. In any case, since the ⌬C1 truncated form of XRCC4 substantially reverses the DSBR defect in XR-1 cells but is not phosphorylated, it is likely that phosphorylation of XRCC4 is  ⌬N (lanes 5 and 6), ⌬C1 (lanes 7 and 8), ⌬C2 (lanes 9 and 10), and ⌬N⌬C1, (lanes 11 and 12)) and full-length versions (indicated XRCC4 HIS (lanes 3 and 4)) of purified XRCC4 HIS were incubated in the presence of [␥-32 P]ATP and 300 ng of purified DNA-PK with (ϩ) and without (Ϫ) linearized plasmid DNA as described under "Materials and Methods." A Ni ϩ fractionation of wild type-infected Sf9 cells (indicated WT (lanes 1 and 2)) is included as a control for proteins contaminating the recombinant XRCC4 HIS preparations. Samples were analyzed by 8% SDS-PAGE (lanes 1-6) or 12% SDS-PAGE (lanes 7-12). Arrows indicate electrophoretic mobilities of the various deletion mutants. not requisite for the DSBR function of XRCC4.
XRCC4 Facilitates Ku DNA End Binding Activity as Well as Association of DNA-PK CS to Ku-The ability of purified XRCC4 HIS to bind DNA or affect the affinity of Ku for DNA ends was examined using EMSAs with baculovirus expressed Ku purified by absorption to DNA-cellulose as described previously (39). Under these conditions, XRCC4 HIS had no apparent affinity for the double-stranded oligonucleotide probe used in this assay (Fig. 6A). XRCC4 HIS did, however, facilitate Ku binding to linear DNA. When a constant amount of XRCC4 HIS was added to increasing amounts of recombinant Ku in the presence of the probe, higher levels of the migration retarded species were observed (Fig. 6A). Poly(dI-dC) and rabbit anti-Ku antisera inhibited formation of the migration retarded species, but plasmid DNA and preimmune rabbit sera did not (data not shown). Addition of increasing amounts of an irrelevant protein, had no effect. The effect of XRCC4 HIS on Ku binding is not entirely specific in that a similar but less dramatic effect was also observed when XRCC4 HIS was included in EMSAs examining recombinant c-jun binding to AP1 sites (data not shown).
The effect of purified XRCC4 HIS on DNA-PK assembly was also examined by EMSAs using commercially available DNA-PK (including Ku70, Ku86, and DNA-PK CS ; Fig. 6B). At limiting concentrations of DNA-PK, one major migration retarded species is apparent (Fig. 6B); this species exactly comigrates with the protein-DNA complex formed with recombinant Ku and the same DNA probe (data not shown). In addition, a second slower migrating complex is observed which represents Ku⅐DNA-PK CS ⅐DNA. When XRCC4 HIS is added to the reaction (lanes 3-7), the level of the Ku⅐DNA-PK CS ⅐DNA complex increases, and the level of Ku⅐DNA complexes decreases. Thus, XRCC4 HIS appears to promote association of DNA-PK CS to Ku⅐DNA. In addition, at higher concentrations of XRCC4 HIS , the Ku⅐DNA-PK CS ⅐DNA complex diminishes and is replaced by several very slowly migrating complexes suggesting that XRCC4 HIS forms stable complexes with DNA-PK in vitro. In contrast, addition of equivalent levels of the truncated protein, ⌬C2 does not significantly alter the mobility of the DNA-PK complexes (lanes 10 -14) although there is a slight augmentation of the Ku⅐DNA complex and a slight change in the mobility of the Ku⅐DNA-PK CS ⅐DNA complex at the highest concentrations of ⌬C2.

DISCUSSION
In this report, we have biochemically characterized the murine XRCC4 gene product. XRCC4 produced in insect cells is extensively post-translationally modified both by phosphorylation and cross-linking. The protein is not glycosylated via asparagine linkages and preliminary evidence suggests that XRCC4 lacks O-linked sugar modifications as well. XRCC4 exists primarily as homodimers, but it can also multimerize into multimers via disulfide linkages. Examination of XRCC4 in native polyacrylamide gels suggests that the protein exists primarily as dimers and tetramers. XRCC4 expressed in insect cells is constitutively phosphorylated on serine, threonine, and tyrosine residues.
XRCC4 is an efficient in vitro substrate for the DNA-dependent kinase. However, extracts from several cell lines which are completely deficient in the catalytic subunit of DNA-PK are also capable of phosphorylating XRCC4 in vitro. Thus, in vitro, XRCC4 can be modified via phosphorylation on serine and threonine by at least two different kinases. Phosphorylation by DNA-PK does not occur on a SQ consensus DNA-PK site, and is most likely located within the carboxyl-terminal 130 amino acids. Finally, it is unlikely that phosphorylation is requisite for the DSBR activity of XRCC4 in that the XRCC4 truncation spanning amino acids 1-204, which is not phosphorylated in vitro or in Sf9 cells, significantly reverses the XR-1 DSBR defect.
XRCC4 facilitates binding of Ku to free DNA ends without altering mobility of Ku⅐DNA complexes. This is reminiscent of the recently defined JAB1 factor which increases c-jun's avidity for AP1 sites without forming a stable complex with c-jun and its target sequence (45). More interestingly, the association of DNA-PK CS with DNA-bound Ku is facilitated by XRCC4. This may be because of enhanced binding of Ku to DNA but could also be a direct effect of XRCC4 on DNA-PK CS . Additionally, XRCC4 can apparently form stable complexes with DNA-PK bound to DNA in vitro. Finally, a truncated form of XRCC4 which cannot functionally complement the XRCC4 defect in XR-1 cells, essentially lacks the capacity to enhance DNA-PK assembly and to form complexes with DNA-PK and DNA in vitro. Thus, the ability of XRCC4 to directly interact with FIG. 6. XRCC4 HIS facilitates both the binding of recombinant Ku to DNA ends, and DNA-PK assembly. A, EMSA analysis of a 32 P-labeled double-stranded oligonucleotide incubated without (lanes 1-4) or with (lanes 5-8) 5 g of purified XRCC4 HIS as indicated. 1, 5, and 10 g of DNA-cellulose purified baculovirus expressed Ku was added in lanes 2-4 and 5-7, respectively. B, EMSA analysis of a 32 Plabeled double-stranded oligonucleotide incubated with 60 ng of DNA-PK (lanes 2-7 and lanes 9 -14). 250 ng, 500 ng, 1 g, 2 g, or 4 g of purified full-length XRCC4 HIS (lanes 3-7) or ⌬C2 (lanes 10 -14) were added. Migration retarded species corresponding to Ku, DNA-PK, and DNA-PK/XRCC4 HIS are indicated. DNA-PK may be functionally germane.
It has been demonstrated that maximal DNA-PK activity requires approximately equal ratios of Ku and DNA-PK CS (46). Albeit it is speculation at this point, the dimeric nature of XRCC4 might suggest that XRCC4 could interact with two separate DNA-PK complexes, possibly stabilizing or aligning the two DNA strands by bridging two DNA-PK complexes prior to ligation. Another possibility is that tetramers of XRCC4 might stabilize 4 units of DNA-PK in the V(D)J recombination complex. The fact that rare coding and signal joints in XR-1 cells likely utilize (and possibly require) short sequence homologies to facilitate ligation, may reflect an inability in these cells to stabilize the two strands. This would also agree with a model invoking XRCC4 as a stabilizing factor for DNA-PK complexes.
The evidence that XRCC4 interacts with DNA-PK in vitro is 2-fold: 1) XRCC4 is phosphorylated by DNA-PK; and 2) XRCC4 facilitates Ku binding to DNA and DNA-PK assembly and alters the electrophoretic mobility of DNA-PK bound to DNA. These findings coupled with the fact that XRCC4 interacts with DNA ligase IV, suggests a role for XRCC4 in targeting cellular ligase activities to damaged DNA which is bound by DNA-PK in a multiprotein complex. Leu and Schatz (47) have shown that RAG 1 and RAG 2 are part of a large multiprotein complex, anchored to the nuclear matrix and that V(D)J recombination may be restricted to certain nuclear compartments. Experimental evidence suggests that V(D)J coding intermediates have extremely short half-lives in vivo, suggesting that factors involved both in initiation and resolution of V(D)J recombination are tightly linked (33,34). In fact, very recently, Agrawal and Schatz (48) demonstrated that DNA-PK is associated with V(D)J recombination intermediates after the cleavage step of V(D)J recombination. Thus, it will be of considerable interest to ascertain whether XRCC4 is a component of this complex.
In summary, this report provides the first biochemical characterization of murine XRCC4. XRCC4 is a disulfide-linked homodimer which can also multimerize. XRCC4 is phosphorylated during expression in insect cells and is also phosphorylated in vitro on serine and threonine residues by the DNA-dependent protein kinase and other unidentified kinases. Recombinant XRCC4 facilitates the association of DNA-PK CS with DNA bound Ku possibly through its ability to facilitate DNA-protein interactions; furthermore, XRCC4 and DNA-PK appear to form stable complexes in vitro. Thus, data presented here are consistent with a model invoking XRCC4 as a stabilizer of the DNA-PK complex.