Characterization of a Ku86 variant protein that results in altered DNA binding and diminished DNA-dependent protein kinase activity.

Three proteins known to play a critical role in mammalian DNA double-strand break repair and lymphoid V(D)J recombination are the autoantigens Ku86 and Ku70 and a 465-kDa serine/threonine protein kinase catalytic subunit (DNA-PKcs). These proteins physically associate to form a complex (DNA.PK) with DNA-dependent protein kinase activity. In this study, we demonstrate using electrophoretic mobility shift assays (EMSAs) that the nuclear DNA end-binding activity of Ku is altered in the human promyelocytic leukemic HL-60 cell line. Western blot and EMSA supershift analyses revealed that HL-60 cells expressed both full-length and variant Ku86 proteins. However, a combined EMSA and immunoanalysis revealed that the Ku heterodimers complexed with DNA in HL-60 cells contained only the variant Ku86 proteins. Finally, UV cross-linking experiments and DNA.PK assays demonstrated that the Ku complexes containing variant Ku86 had a greatly reduced ability to interact with DNA-PKcs and that consequently HL-60 cells had severely diminished DNA.K activity. These data provide important insights into the interaction between Ku and DNA-PKcs and into the role of DNA.PK in DNA double-strand break repair.

The exposure of mammalian cells to ionizing radiation (IR) 1 induces lesions in chromosomal DNA such as strand scissions, single-stranded breaks, double-stranded breaks (DSBs), and base cross-links (1,2). In particular, DNA DSBs appear to be the predominant cytotoxic lesions as even a single unrepaired DNA DSB can be a lethal event (3,4). IR-sensitive (IR s ) mutants have been isolated from hamster ovary (CHO) or lung (V79) cell lines, and in approximately half of these cell lines, IR sensitivity correlated with a greatly decreased ability to repair DNA DSBs (reviewed in Ref. 5). Thus, the DSB repair capacity of a cell appears to be a critical, although not the sole, factor in determining cellular IR sensitivity.
Lymphoid V(D)J recombination is a site-specific reaction that involves the assembly of noncontiguous genomic segments that encode the variable (V), diversity (D), and joining (J) elements of immunoglobulin and T cell-receptor genes (for a recent review, see Ref. 6). Elegant analyses of V(D)J recombination products in vivo and in vitro strongly suggests that DNA DSBs are an essential intermediate in the V(D)J reaction mechanism (7)(8)(9)(10)(11). Thus, it is plausible that DNA DSB repair and mammalian V(D)J recombination may share some common factors. Consistent with this hypothesis, mutants have been identified that have defects in both pathways. Mice homozygous for the murine severe combined immune deficiency (scid) mutation, exhibit a profound immune deficiency in vivo, that is caused by a defect in V(D)J recombination coding junction formation, and they are also severely IR s due to a DNA DSB repair defect (reviewed in Refs. 6, 12, and 13). In addition, several Chinese hamster IR s mutants (V3, xrs-6, XR-1, sxi-1, sxi-2, sxi-3, and sxi-4) that are impaired in DSB repair were also shown to be defective for V(D)J recombination (14 -19). Therefore, the genes defective in these mutant cell lines are strong candidates to encode factors which are involved both in DNA DSB repair and V(D)J recombination.
While the exact enzymology of mammalian DNA DSB repair is still unknown, a complex, which has DNA-dependent serinethreonine protein kinase activity (DNA⅐PK) (20,21), and which consists of at least two components, the 465-kDa catalytic subunit (DNA-PK cs ) and Ku protein (22)(23)(24), has been shown to be intimately involved in DNA DSB repair (reviewed in Refs. 13,[25][26][27][28]. DNA-PK cs has been shown recently to very likely be the product of the scid gene (29 -33), and it has long been known that animals homozygously defective at this locus were profoundly IR s and defective in DNA DSB repair (34 -36). Ku is a heterodimeric protein of 70-and 86-kDa subunits which binds tightly to the ends of a variety of double-stranded DNA (37)(38)(39)(40). Ku is thus thought to provide the DNA binding component for the DNA⅐PK holoenzyme. Recently, it was shown that members of the fifth x-ray cross-complementation group (XRCC5) (5,41), which are very IR s and DNA DSB repairdefective, lack Ku DNA end-binding activity (19,(42)(43)(44). Since the Ku86 gene maps to human chromosome 2q33-35 (45) and XRCC5 group cells could be rescued by the same region (46), Ku86 was a strong candidate for the XRCC5 gene. Consistent with this hypothesis, several groups independently reported that transfection of a Ku86 cDNA was able to rescue the defects of XRCC5 mutants (18,19,(47)(48)(49), and molecular defects in the Ku86 gene in a XRCC5 group cell line have been identified (49). Thus, DNA⅐PK has been unequivocally identified as an impor-tant mammalian DNA repair complex and mutations in either the DNA-PK cs or the 86-kDa subunit of Ku result in severe IR s and V(D)J recombination and DNA DSB repair defects.
We have recently demonstrated that the human promyelocytic leukemic HL-60 cell line was IR s , but DNA DSB repairproficient (50) and thus, in this latter regard, differed significantly from scid or XRCC5 mutant cell lines. In this study, we demonstrate that the nuclear DNA-binding activity of Ku is altered in HL-60 cells. Using monoclonal antibodies directed against various epitopes of the Ku proteins, we demonstrate that, although HL-60 cells expressed both a full-length and a variant truncated version of Ku86, only the heterodimer of Ku70-variant Ku86 was complexed with DNA. This altered Ku complex had a greatly reduced ability to interact with DNA-PK cs and this resulted in severely diminished DNA⅐PK activity. These data provide important insights into the interaction between Ku and DNA-PK cs and into the role of DNA⅐PK in DNA double-strand break repair.
Cells-HL-60 cells were purchased from American Type Culture Collection (Rockville, MD). Isolation of the HCW-2 cell mutant from HL-60 cells has been described (54). All cells were cultured in RPMI 1640 medium (Life Technologies, Inc.) supplemented with fetal bovine serum (20% for HL-60 and 10% for HCW-2 cells), 100 units/ml penicillin, and 50 units/ml streptomycin. Cell cultures were kept in a humidified incubator with 5% CO 2 at 37°C.
Preparation of Nuclear and Cytoplasmic Extracts-Cells were washed three times in phosphate-buffered saline. The cells were lysed in 10 ml of ice-cold lysis buffer A (10 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 0.25 mM spermine, 0.5 mM spermidine, 0.25% Nonidet P-40) on ice, and then briefly vortexed. The lysate was layered on top of 30 ml of ice-cold sucrose buffer (buffer A with 0.25 M sucrose) in a 50-ml centrifuge tube and centrifuged at 1500 relative centrifugal force in an MSE MISTRAL 3000i centrifuge at 4°C for 5 min. The supernatant was removed by aspiration. The pellet was resuspended in lysis buffer A, and the nuclei were purified through sucrose medium twice. Nuclei were then lysed in NaCl extraction buffer (10 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride), and the lysate was incubated on ice for 30 min. An equal volume of 0.5 M NaCl, 18% polyethylene glycol 8000 buffer was mixed with the lysate, and the incubation continued for another 10 min. The lysate was then centrifuged at 12,500 ϫ g for 10 min at 4°C. The clarified supernatant was used as the native nuclear extract. To prepare cytoplasmic extract, cells were lysed in lysis buffer A in the presence of 1 mM phenylmethylsulfonyl fluoride. The lysate was immediately centrifuged at 12,500 ϫ g for 10 min at 4°C. The clarified supernatant was used as native cytoplasmic extract. Protein content in the extract was determined using Bio-Rad protein assay reagents. Aliquots of the extract was stored at Ϫ85°C.
Preparation of DNA Probes-EcoRI and XbaI DNA fragments (ϳ250 bp) were end-labeled with Klenow DNA polymerase in the presence of [␣-32 P]dATP. To prepare the double-stranded 25-mer DNA, two oligonucleotides (5Ј-ACTTGATTAGTTACGTAACGTTATG-3Ј and 5Ј-CATA-ACGTTACGTAACTAATCAAGT-3Ј) were end-labeled with T4 polynucleotide kinase in the presence of [␥-32 P]ATP and subsequently annealed together. All radiolabeled fragments were electrophoresed through 5% polyacrylamide gels and subsequently purified.
Electrophoretic Mobility Shift Assay-A gel mobility shift assay for the determination of Ku DNA end binding activity was adapted for this study (18,19). Briefly, radiolabeled DNA (4 ng, 100,000 cpm) was incubated with nuclear or cytoplasmic extract in 15 l of binding buffer (10 mM Tric-HCl, pH 8.0, 1 mM EDTA, 10% glycerol, 200 mM NaCl, 1 g of circular plasmid pRSV-neo DNA) on ice for 5 min. The samples were electrophoresed in a 5% polyacrylamide gel at room temperature for 3 h at 130 V. The gel was dried on Whatman 3M paper and exposed to Fuji x-ray films at Ϫ85°C. In the supershift experiments, monoclonal antibodies (1 l of ascites preparations) were added to the binding mixture following the 5-min incubation on ice and incubated for an additional 10 min prior to gel electrophoresis.
Digestion of DNA-Protein Complexes by DNase I-A radiolabeled ϳ150-bp DNA fragment (4 ng, 100,000 cpm) and nuclear extracts (3 g) were incubated on ice for 5 min as described except that the circular plasmid pRSV-neo DNA was omitted. Then, 5 l of DNase I solutions (containing various amounts of DNase I and 30 mM MgCl 2 and 30 mM CaCl 2 ) were added, and the mixtures were incubated at 37°C for 3 min. The samples were immediately subjected to electrophoresis in a 5% polyacrylamide gel under the conditions described for the gel mobility shift assay. The gel was subsequently dried on Whatman 3M paper and exposed to x-ray film.
Western Analyses of Ku Proteins-For the immunoblot detection of Ku70 and Ku86, proteins were subjected to electrophoresis on a 10% SDS-polyacrylamide gel under reducing conditions. The proteins were then electroblotted onto a nitrocellulose filter in a Bio-Rad Trans-Blot chamber and detected as described previously (50,54).
Combined Electrophoretic Mobility Shift and Western Analyses-Approximately 500,000 cpm of a 32 P-radiolabeled 25-mer doublestranded DNA (dsDNA) probe was mixed with 5 g of nuclear extract and incubated on ice for 5 min under conditions described in the electrophoretic mobility shift assay. Each sample was prepared in duplicate and an electrophoretic mobility shift assay was performed. One half of the gel containing one set of the samples was dried on a piece of Whatman paper and exposed for autoradiography. The duplicate samples in the other half of the gel were electroblotted onto a nitrocellulose filter using SDS-containing transfer buffer (20 mM Tris, 150 mM glycine, 20% methanol, 0.5% SDS, pH 8.0). The filter was washed in running tap water for 1 min before specific proteins on the filter were localized by standard Western analyses. By comparing the positions of the bands on the autoradiogram and Western blot, the presence of a particular protein in the DNA-protein complex can be determined.
UV Cross-linking-A protocol used by Gottlieb and Jackson (23) to UV cross-link Ku70 and DNA-PK cs proteins to DNA was slightly modified. Briefly, approximately (10 ng) 500,000 cpm of a 32 P-radiolabeled double-stranded 25-mer DNA probe was mixed with 25 g of nuclear extract in the presence or absence of 500 ng of unlabeled DNA in a final volume of 50 l. The final mixture also contained 1 g of pRSV-neo plasmid DNA, 200 mM NaCl, 2 mM EDTA, and 10 mM Tris-HCl, pH 8.0. The mixture was incubated on ice for 10 min, followed by UV irradiation at 254 nm in a UV Stratalinker (Stratagene) for 5 min at room temperature. Then, 15 l of a DNase I solution (500 units of DNase I, 50 mM MgCl 2 , 10 mM CaCl 2 ) was added, and the DNA was digested at 37°C for 20 min. Five volumes of acetone were added to each sample, and proteins were precipitated at Ϫ20°C for 15 h. The precipitated proteins were pelleted by centrifugation at 12,500 ϫ g, dissolved in a ureaglycerol-SDS buffer, and subjected to electrophoresis in a 5% polyacrylamide gel. The gel was electroblotted onto a nitrocellulose filter and exposed to x-ray film. DNA-PK cs on the same filter was identified subsequently by Western analysis.
DNA⅐PK "Pull Down" Kinase Assays-Briefly, 100 g of nuclear extract protein was incubated with 50 l of dsDNA cellulose (30 mg/ml) in 1 ml K buffer (25 mM Tris-HCl, pH 7.9, 10 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 2.5% glycerol) containing 60 mM NaCl for 30 min at 4°C (55). The DNA cellulose was then washed by centrifugation 3 times in 1 ml of ZЈ0.05 buffer (55). The DNA cellulose was then resuspended in 50 l of potassium buffer. Using synthetic DNA⅐PK peptide substrates derived from the N-terminal transcriptional activation domain of murine p53 (wild-type peptide, EPPLSQEAFADLLKK; mutated peptide, EPPLSEQAFADLLKK), DNA⅐PK activity in an aliquot (13 l) of the DNA celluloses was assayed as described elsewhere (55). To determine the status of Ku and DNA-PK cs in the kinase reaction mixture, aliquots of the reaction mixture were electrophoresed in 6% (for DNA-PK cs ) or 10% SDS-polyacrylamide gel (for Ku) under reducing conditions, proteins were transferred onto nitrocellulose filters which were subjected to Western blot analysis for the presence of DNA-PK cs , Ku70 and Ku86 proteins.

IR Responses of HL-60 and a Clonal Variant, HCW-2-We
have recently demonstrated that the human promyelocytic leukemic HL-60 cell line was hypersensitive to x-irradiation, whereas a stable clonal variant of the HL-60 cell line, HCW-2, was IR-resistant (50). Interestingly, both cell lines were proficient for DNA DSB repair and appeared to repair DSB lesions with identical kinetics (50). In addition, both HL-60 and HCW-2 cells progressed to G 2 following x-irradiation; however, HL-60 cells, unlike HCW-2 cells which eventually resumed cell cycling, arrested at G 2 for at least 48 h before they underwent apoptosis (50). Since much of the mammalian cellular response to x-irradiation appears to be mediated by the DNA⅐PK complex (reviewed in Refs. 13, 25-28), we investigated the status of this activity in HL-60 and HCW-2 cells. These results are presented below.
Ku⅐DNA Complexes of HL-60 and HCW-2 Cells Have Different Electrophoretic Mobilities-The DNA binding component of the DNA⅐PK complex is provided by the Ku70 and Ku86 autoantigens and this binding activity can easily be detected by using dsDNA fragments in an electrophoretic mobility shift assay (EMSA) (43,44). Thus, nuclear extracts were prepared from HL-60 and HCW-2 cells, mixed with a radiolabeled 250-bp dsDNA fragment, and then analyzed for Ku⅐DNA complex formation by an EMSA (Fig. 1A). The results consistently showed two features. First, as the concentration of nuclear extract in the binding assay was increased, the mobility of the protein-DNA complexes became slower. At least four complexes of different mobility were observed for both HL-60 (L1-L4) and HCW-2 (H1-H4). The slowest moving complex for both cells appeared when the protein concentration in the binding assay was approximately 2.0 g. This mobility change in association with increased protein concentration has been observed in many laboratories and has been interpreted as multiple Ku heterodimers binding to the same DNA probe (37,40,56). This is presumed to occur because, once bound to the end of a dsDNA fragment, Ku has the ability to translocate internally on that fragment, thus freeing up the end for another Ku heterodimer to bind (37,40,57). More interestingly, however, was the observation that the mobility of HL-60 complexes was always faster than the corresponding HCW-2 complexes (Fig. 1A, compare L1 with H1, L2 with H2, etc.). These results suggested that the Ku⅐DNA complexes of HL-60 were different from those of HCW-2.
It has been demonstrated that a 25-30-bp dsDNA fragment is the minimum length required for the binding of a single Ku heterodimer (38). Thus, a 25-mer dsDNA probe was prepared and used for EMSA analysis with HL-60 and HCW-2 nuclear extracts. The resulting HL-60 complex (L1) still migrated faster than the corresponding HCW-2 complex (H1) (Fig. 1B). Thus the size differences observed on a larger fragment (Fig.  1A) could be observed with a minimal DNA binding site (Fig.  1B), and this was suggestive that the difference in DNA binding between the two cell lines was an inherent property of the Ku heterodimer and not due to additional factors.
We next surveyed a large number of human cell lines for their Ku DEB activity. Jurkat and HT-H9 T cell lymphoma cell lines showed the higher HCW-2 pattern (Fig. 1B) as did a variety of transformed and nontransformed cell lines derived from different tissues (data not shown). Thus, it appeared as if it was the lower (HL-60) Ku DEB pattern which was abnormal.
Altered Immunoreactivity of the HL-60 and HCW-2 Ku⅐DNA Complexes-We sought to identify the nature of the alteration in Ku⅐DNA complexes in HL-60 cells. As a first approach we attempted to confirm the presence or absence of Ku70 and Ku86 proteins in the HL-60 Ku⅐DNA complexes. This was achieved by using mAbs in EMSA "supershift" experiments. Thus, nuclear extracts were prepared from HL-60 and HCW-2 cells and incubated with a ϳ250-bp radiolabeled probe in the presence or absence of mAbs specific for Ku polypeptides before being subjected to electrophoresis. Ab 162 (52,53), was recently shown to recognize an epitope defined by the heterodimer of Ku70/Ku86. In the presence of this antibody, both the HL-60 (L1, L2, and L3) complexes and the HCW-2 (H1 and H2) complexes were completely supershifted and retarded in the gel (Fig. 2). A different result, however, was obtained using mAb 111, which recognizes an epitope on the C terminus (defined by amino acids 610 -705) of the Ku86 protein (51). This antibody clearly supershifted the HCW-2 Ku⅐DNA complexes although not as completely as mAb 162. The lower reactivity of mAb 111 in supershift assays has been previously observed in other laboratories and is probably an inherent property of this mAb (18). In contrast to HCW-2 extracts, however, mAb 111 did not affect the migration of the HL-60 complexes at all. A control monoclonal antibody specific for the human bcl-x proteins similarly did not affect the migration of the Ku complexes (Fig. 2). From these results we concluded that although both Ku70 and Ku86 were present in the HL-60 Ku⅐DNA complexes, the Ku86 C terminus was either altered or missing, such that mAb 111 could no longer react with it.
Detection of a Variant Truncated Ku86 in Nuclei of HL-60 Cells-To pursue the above observation, we next performed Western blot analyses of nuclear Ku proteins from HL-60 and HCW-2 cells using additional monoclonal antibodies that react against different epitopes of Ku. mAb N3H10 is specific for an epitope (defined by amino acids 506 -541) in the C terminus of Ku70 (51). This mAb detected equal amounts of Ku70 proteins in nuclear extracts from HL-60 and HCW-2 (Fig. 3A). Not surprisingly, mAb 111 (specific for the C terminus of Ku86) which had failed to supershift HL-60 Ku⅐DNA complexes (Fig.  2), also failed to detect full-length Ku86 protein in standard HL-60 nuclear extracts (Fig. 3B, lane 1). Only by overloading the gel was Ku86 detected (Fig. 3B, lane 2). This was in sharp contrast to nuclear extracts prepared from HCW-2 cells, which had easily detectable levels of full-length Ku86 (Fig. 3B, lanes  3 and 4). mAb S10B1, which recognizes an epitope (defined by amino acids 8 -221) on the N terminus of Ku86 (51), detected an equivalent amount of Ku86 in HCW-2 cells as had mAb 111 (compare Fig. 3C, lanes 3 and 4 with Fig. 3B, lanes 3 and 4). Similarly, this mAb detected only low levels of full-length Ku86 in HL-60 nuclear extracts (Fig. 3C, lane 2). However, mAb S10B1 also detected two additional proteins of 69 and 71 kDa (hereafter referred to as variant Ku86) that were not detected in the nuclear extract from HCW-2 cells (Fig. 3C, lane 2). An identical result was obtained with a different monoclonal antibody, N9C1 (52,53), which also recognizes an epitope (defined by amino acids 1-374) on the N terminus of Ku86 (data not shown). From these experiments we conclude that 1) nuclear extract from HCW-2 cells contains wild-type levels of fulllength Ku70 and Ku86 proteins, 2) HL-60 nuclear extract contains wild-type levels of full-length Ku70 protein, but greatly diminished levels of full-length Ku86 protein, and 3) the majority of the immunoreactive Ku86 protein in HL- 60  This hypothesis was experimentally tested in an experiment that combined an EMSA with Western blot analyses. Samples were prepared in duplicate and a standard EMSA was performed as described. One half of the gel was dried and used for the development of an autoradiogram to localize the positions of Ku⅐DNA complexes. Proteins in the other half of the gel were electroblotted onto a nitrocellulose filter for Western analyses using mAbs N3H10, 111, S10B1, or N9C1. The autoradiogram demonstrated that the Ku complexes were in their standard L (HL-60) and H (HCW-2) positions on the gel (Fig. 4A). When the Western blot was analyzed using mAb N3H10 (Fig. 4B) Ku70 was detected at both of these positions. When mAb S10B1 (Fig. 4D) or mAb N9C1 (data not shown) was used, Ku86 was also detected at both positions. However, when mAb 111 was used, Ku86 was identified only at the H position but not at the L position (Fig. 4C). These results clearly demonstrated that the Ku⅐DNA complexes of HL-60 cells contained only variant Ku86. Variant Ku-86 Is Localized in the Nucleus of HL-60 Cells-To investigate the subcellular distribution of the variant Ku86, extracts were prepared from the nuclear and cytoplasmic compartments of HL-60 and HCW-2 cells and used for EMSAs. Unlike HL-60 nuclear extract, the HL-60 cytoplasmic extract contained mostly Ku⅐DNA complexes (H) with the same mobility as those observed with HCW-2 extracts (Fig. 5). This observation suggested that significant amounts of full-length Ku86 can be (and are) synthesized in HL-60 cells, but that inside the nucleus the variant Ku86 form predominates.
HL-60 Ku⅐DNA Complexes Have Altered DNase I Sensitivity-To determine if there were any functional differences associated with the altered HL-60 Ku⅐DNA complexes we next tested the sensitivity of the bound DNA to nuclease digestion. A radiolabeled 150-bp DNA probe was preincubated with an excess amount of HL-60 or HCW-2 nuclear extract to first obtain maximal Ku binding. The Ku⅐DNA complexes were then digested with increasing amounts of DNase I and analyzed by EMSA (Fig. 6). The DNA of HL-60 Ku⅐DNA complexes was susceptible to even 5 units of DNase I and was completely converted to the L1 form with 25 units. In striking contrast, the DNA in the HCW-2 Ku⅐DNA complexes was very resistant to digestion, such that even at 50 units of DNase I approximately a third of the DNA remained undigested (Fig. 6). Thus, the Ku⅐DNA complexes formed with HL-60 nuclear extracts ap- FIG. 2. EMSA supershift experiments demonstrate that a mAb specific for the C terminus of Ku86 will not retard HL-60 Ku⅐DNA complexes. A ϳ250-bp radiolabeled DNA probe was incubated with 1.2 g of HL-60 or HCW-2 nuclear extract and used in an EMSA. Either no antibody (control) or monoclonal antibodies specific for amino acids 610 -705 of Ku86 (Ab 111), an uncharacterized epitope defined by the heterodimer of Ku70 and Ku86 (Ab 162) or the bcl-x proteins (Ab bcl-x), were added to the binding mixture prior to gel electrophoresis. The samples were then electrophoresed on a 5% polyacrylamide gel, the gel was dried and subsequently exposed to x-ray film. The position of free probe (f.p.) and some of the complexes formed with either HL-60 extracts (L1, L2, and L3) or HCW-2 extracts (H1 and H2) are shown as well as the region where the supershifted complexes migrate. nuclear extracts and analyzed with an EMSA. One half of the gel containing one set of the samples was dried and exposed for autoradiography. The positions of the complexes formed with HL-60 extract (L) or HCW-2 extract (H) are shown with the arrows. B, C, and D, the duplicate samples in the other half of the gel were electroblotted onto a nitrocellulose filter before specific proteins on the filter were sequentially localized by Western analyses. B, Western analysis with mAb N3H10 (Ku70); C, Western analysis with mAb 111 (Ku86, C terminus); D, Western analysis with mAb S10B1 (Ku86, N terminus). peared to be 5-10-fold more sensitive to DNase I, and this suggested that the spatial arrangement of Ku heterodimers on DNA was significantly different between HL-60 and HCW-2 Ku⅐DNA complexes.
HL-60 Ku Is Less Capable of Interacting with DNA-PK cs -The known role of Ku is to provide the DNA binding activity for the DNA⅐PK holoenzyme. In this capacity, Ku associates with the catalytic subunit (DNA-PK cs ) of the DNA⅐PK holoenzyme. This association, which can be observed by immunoprecipitation and UV cross-linking (21,23,24,29), is not detected in the salt conditions used for standard EMSAs (24). Therefore, to test if the HL-60 Ku complex was altered in its interaction with DNA-PK cs , UV cross-linking experiments were performed. A radiolabeled dsDNA oligonucleotide probe was incubated with equal amounts of HL-60 and HCW-2 nuclear extracts and exposed to UV light. The nuclear extracts were then digested with DNase I to remove excess probe, and the proteins were subsequently separated by polyacrylamide gel electrophoresis, transferred to a nitrocellulose filter, and exposed for autoradiography. Under these conditions, other laboratories have shown that significant cross-linking of the DNA probe occurs to Ku70 and DNA-PK cs , but not to Ku86 (23,24,29). Similarly, we observed significant cross-linking of proteins of 70 and Ͼ205 kDa (the approximate size of DNA-PK cs ) (20,21) in HCW-2 nuclear extracts (Fig. 7A, HCW-2, Ϫ lane). The cross-linking of both proteins could be diminished, but not abolished, by preincubating the nuclear extracts with an excess of unlabeled DNA probe (Fig. 7A, HCW-2, ϩ lane). A level of cross-linked Ku70 similar to that observed in HCW-2 extracts was detected in the nuclear extracts of HL-60 cells (Fig. 7A, compare HL-60, Ϫ lane with HCW-2, Ϫ lane), but this could be completely competed away with excess probe (Fig. 7A, HL-60, ϩ lane). Most impressively, however, was that only a minute amount of the Ͼ205-kDa species was cross-linked in HL-60 nuclear extracts. This cross-linking could also be completely competed away with excess unlabeled probe (Fig. 7A, HL-60, ϩ lane). When the same filter, from which the autoradiogram shown in Fig. 7A was derived, was subjected to a Western analysis using a monoclonal antibody directed against DNA-PK cs (20) a signal in all lanes was detected that precisely corresponded to the Ͼ2050-kDa species (Fig. 7B). From these results we conclude several things. First, the Ͼ205-kDa species observed in the UV-cross-linking experiment is indeed DNA-PK cs . Second, that the Ku70 in HL-60 cells is not bound as tightly to DNA as the Ku70 in HCW-2 cells, since the former could be competed by excess DNA, whereas the latter could not. This conclusion is consistent with the altered DNase I specificity we previously observed (Fig. 6). Last, and most importantly, we conclude that the HL-60 Ku complex is greatly diminished in its ability to interact with DNA-PK cs .
Expression of Variant Ku86 in HL-60 Cells Results in Diminished DNA⅐PK Activity-Since the available evidence suggests that interaction of DNA-PK cs with Ku activates the protein kinase (29,55,58), we anticipated that nuclear extract derived from HL-60 cells would be deficient in DNA⅐PK activity. To experimentally test this hypothesis, DNA-cellulose was added to HL-60 and HCW-2 nuclear extracts and the DNA⅐PK complex was "pulled down" (55) by centrifugation. The resulting pellets were then incubated with a synthetic peptide (wildtype) derived from the N-terminal transcriptional activation domain of murine p53 (EPPLSQEAFADLLKK) (55), in the presence of [␥-32 P]ATP. As negative controls, either no peptide or a peptide (mutant; EPPLSEQAFADLLKK), which is a poor DNA⅐PK substrate were also used to ensure the specificity of DNA⅐PK-catalyzed phosphorylation. As expected, the nuclear pull-down preparation of HCW-2 cells exhibited a significant amount of DNA⅐PK activity, whereas there was no detectable DNA⅐PK activity in the nuclear pull-down of HL-60 cells (Fig.  8A). To verify the status of Ku and DNA-PK cs , the proteins associated with the DNA cellulose were eluted and subjected to FIG. 5. Wild-type Ku⅐DNA binding activity is found in the cytoplasm of HL-60 cells. Nuclear (N) and cytoplasmic (C) extracts were prepared from HL-60 and HCW-2 cells and analyzed for Ku DNA binding activity using standard EMSA conditions. The positions of the lower (L) and higher (H) complexes normally formed with nuclear extracts are shown with the arrows.
FIG. 6. HL-60 Ku⅐DNA complexes are more susceptible to DNase I digestion than HCW-2 Ku⅐DNA complexes. A radiolabeled ϳ150-bp DNA fragment and nuclear extracts were incubated on ice for 5 min. Solutions containing either 0, 1, 5, 10, 25, or 50 units of DNase I were added, and the mixtures were then incubated at 37°C for 3 min before being subjected to electrophoresis in a 5% polyacrylamide gel under the conditions described for the EMSA .   FIG. 7. UV cross-linking experiments demonstrate that Ku70 is associated with DNA in both HL-60 and HCW-2 nuclear extracts but that DNA-PK cs can be cross-linked to DNA only in HCW-2 extracts. A, a 32 P-radiolabeled 25-mer dsDNA probe was mixed with 25 g of nuclear extract in the presence (ϩ) or absence (Ϫ) of 500 ng of unlabeled DNA added as competitor. The mixture was incubated on ice for 10 min, followed by UV irradiation and then DNase I digestion. The proteins were then precipitated with acetone, dissolved in a ureaglycerol-SDS buffer, and subjected to electrophoresis in a 5% polyacrylamide gel. The gel was electroblotted onto a nitrocellulose filter and exposed to x-ray film. The positions of protein markers of 66 and 205 kDa electrophoresed in a neighboring lane are shown. B, HL-60 nuclear extracts contain DNA-PK cs even though it is not complexed with DNA. DNA-PK cs on the same filter shown in part (A) was identified by Western analysis.
Western analyses. HCW-2 cell pull-downs contained full-length Ku86, Ku70, and DNA-PK cs (Fig. 8, B, C, and D, lane 2, respectively). In contrast, the pull-down of HL-60 cells contained full-length Ku70 (Fig. 8C, lane 1), only variant Ku86 (Fig. 8B, lane 1), and lacked DNA-PK cs (Fig. 8D, lane 1). These results demonstrate that DNA-PK cs is incapable of interacting with a DNA-bound Ku heterodimer containing a variant Ku86 subunit and this results in the loss of DNA⅐PK activity detectable by the pull-down assay.

DISCUSSION
In this study, we have shown that the nuclear DNA binding activity of Ku is altered in the human promyelocytic leukemic HL-60 cell line. Western blot and EMSA supershift analyses revealed that although HL-60 cells expressed both a full-length and a variant truncated version of Ku86, only the variant Ku86 was found to be complexed with DNA. The variant Ku86 complexes had a greatly reduced ability to interact with DNA-PK cs which resulted in severely diminished DNA⅐PK activity. The significance of these results is discussed below.
On the Nature of the Variant Ku86 Protein-There are multiple possible explanations for the appearance of the faster migrating species which we have termed variant Ku86. First, it is possible that the variant Ku86 is not Ku86 at all, but an immunologiclly related protein(s). However, the observation that three different monoclonal antibodies, 162 (Fig. 2), S10B1 (Figs. 3 and 4), and N9C1 (data not shown), which have been rigorously characterized as recognizing human Ku86 protein (52,53), cross-react with the faster migrating Ku86 protein(s) argues strongly against this interpretation. Second, it is possible that the variant truncated Ku86 arises from an unusual spliced form of Ku86. While we cannot rule out this possibility, we have seen no evidence for a smaller Ku86 mRNA on Northern blots (data not shown). In addition, the observation that cytoplasmic extracts from HL-60 cells express full-length Ku86 ( Fig. 5 and data not shown) argues strongly against the possibility that HL-60 cells contain an alternatively spliced or mutated Ku86 allele that simply expresses a truncated Ku86 protein. Instead, these results suggest that HL-60 cells express wild-type, full-length Ku86, but that either as the protein enters, or once it is inside, the nucleus it is proteolytically processed to a smaller form(s).
Last, it should pointed out that proteolytic cleavage of the C terminus of Ku86 has been previously observed. Paillard and Strauss (61) have shown that in vitro under certain incubation conditions that preparations of simian Ku86 could be proteolytically cleaved to a single species of 69 kDa. This truncated Ku86 resulted in an altered EMSA Ku⅐DNA binding profile that was similar or identical to the profile we observed with nuclear extracts from HL-60 cells. However, these authors showed that the protease inhibitors leupeptin and chymostatin or buffers with an elevated pH (8.0) would completely abrogate the cleavage event. We have made HL-60 nuclear extracts under all these conditions individually and in combination and we have not detected any inhibition of the cleavage event (data not shown). Resolution of this issue may come with the purification of the protease responsible for the cleavage event.
Multiple Roles for the C Terminus of Ku86 -The observation that HL-60 cells contain a variant truncated Ku86 provides many interesting clues to the role of this protein in DNA repair. First, Ku has been identified to contain DNA-dependent ATPase (64) as well as ATP-dependent helicase (57) activities.
In photoaffinity labeling experiments with [␣-32 P]ATP both Ku subunits were radiolabeled (57). In particular, the putative ATP binding site in Ku86 was postulated (62) to reside within amino acids 677-689, which is in the portion of Ku86 that is apparently missing in the variant form. It can easily be envisioned that the loss of this ATP binding site may significantly affect either ATPase or the helicase activities. Second, we observed that a long (250 bp) dsDNA fragment complexed with variant Ku heterodimers was much more susceptible to DNase I digestion than full-length heterodimers (Fig. 6). This suggests either that the variant Ku complexes do not bind as tightly to DNA or they are less capable of cooperative interaction. The observation that the variant Ku complexes, in comparison to full-length complexes, could be more efficiently competed away with excess competitor DNA in the UV cross-linking experiments suggests that DNA binding may indeed be affected (Fig.  7). Therefore, we favor a model where the presence of the C terminus of Ku86 is necessary for generating cooperative interactions between Ku heterodimers. Since it has long been known that the DNA end-binding activity of Ku resides solely or exclusively in the Ku70 subunit (Fig. 7) (23,39,56,63,64), the predominant role of Ku86 may be in mediating protein: protein interactions. The loss of interaction between Ku heterodimers could easily explain the increased DNase I sensitivity of the variant Ku⅐DNA complexes. Third, the greatly diminished levels of DNA-PK cs that can be UV cross-linked to DNA (Fig. 7) and the greatly reduced DNA⅐PK levels in DNA pull-down assays implies that the truncated Ku complexes fail to interact with DNA-PK cs . A simple interpretation would be that the C terminus of Ku86 directly interacts with DNA-PK cs , though we cannot rule out at this time that the loss of Ku86 sequences is affecting the conformation of the heterodimer and it is some other domain that is actually making contact with DNA-PK cs .
Implications for DNA DSB Repair-Cells defective in DNA-PK cs (scid) and Ku86 (sxi-3) cells are defective in DNA DSB repair (17,34,35), which suggested that a functional DNA⅐PK holoenzyme was essential for DNA DSB repair activity. Thus, it is surprising that HL-60 cells, which have been shown to be proficient for DNA DSB repair (50,59), contain an altered Ku86 and greatly diminished DNA⅐PK activity (Fig. 8). One explanation for this could be that the sxi-3 mutant is completely null for Ku86 expression (18,19), and while the molecular basis of the scid mutation is unknown, scid cells contain undetectable amounts of DNA⅐PK activity (29,30). In HL-60 cells, some nuclear full-length Ku86 is expressed (Fig. 3), and while we could not detect DNA⅐PK activity in DNA pull-down assays (Fig. 8) there was some, albeit greatly reduced, DNA-PK cs detectable in the UV cross-linking experiments (Fig. 7). This agrees with an earlier study which, using a different assay FIG. 8. HL-60 cells are deficient in DNA⅐PK activity. A, DNAcellulose was added to HL-60 and HCW-2 nuclear extracts and DNA⅐PK was "pulled down" by centrifugation (55). This was then incubated with [␥-32 P]ATP in the presence of either no peptide, a mutant (MT) peptide substrate, or a wild-type (WT) peptide and the amount of radioactive incorporation was determined. B, C, and D, Western analyses of the proteins associated with the DNA cellulose following pull-down. Proteins from HL-60 (lane 1) or HCW-2 (lane 2) cells were eluted from the DNA cellulose and electrophoresed in 6% (for DNA-PK cs ) or 10% (for Ku) SDS-polyacrylamide gels under reducing conditions. Proteins were then transferred onto nitrocellulose filters which were subjected to Western blot analysis for the presence of B, Ku86 using mAb S10B1; C, Ku70 using mAb N3H10; and D, DNA-PK cs using mAb 42-27. system, suggested that HL-60 cells had very low, but detectable, DNA⅐PK activity (60). This implies that even low levels of DNA⅐PK may be sufficient to carry out DNA DSB repair. This interpretation is supported by the observation that, in comparison to human cells, rodent cell lines normally contain 50-fold less DNA⅐PK activity (55) and yet are perfectly capable of carrying out DNA DSB repair (17,34,35). Thus, in terms of DNA DSB repair, human cell lines may be more tolerant of alterations in DNA⅐PK activity.
Alternatively, the DNA DSB repair-proficient phenotype of HL-60 cells might also be explained if the presumed proteolysis of Ku86 is DNA-dependent and only occurs after the DNA⅐PK complex forms on DNA. In vivo then, once the DNA⅐PK complex is assembled, there could be a competition between the complex performing its function (i.e. DNA DSB repair) and inactivation of the complex through proteolysis of Ku86. This would contrast with the case of scid or sxi-3 cells, where there would be no opportunity for DNA repair since the complete DNA⅐PK complex cannot form.
A Novel Function for DNA⅐PK?-The original motivation for performing the experiments described above was the observation that following X-irradiation HL-60 cells arrested permanently at the G 2 checkpoint whereas x-irradiated HCW-2 cells eventually resumed cell cycling (50). The subsequent demonstration that HL-60 cells contain reduced DNA⅐PK activity due to an alteration in Ku86 suggests that permanent G 2 arrest may be a common feature of cells defective in DNA⅐PK (65). This observation would suggest that DNA⅐PK may provide an activity, distinct from its activity required for DNA DSB repair, that is necessary for DNA-damaged cells to traverse a G 2 checkpoint. DNA⅐PK is clearly required for the efficient rejoining of broken chromosomes following DNA damage (17,34,35,66,67). Consequently, it could have been argued that DNA-PK cs or Ku mutant cells arrest in G 2 not because DNA⅐PK activity is needed to progress through the G 2 checkpoint, but simply because the cells arrive at that checkpoint with a significant amount of DNA damage. This, however, is unlikely to be the case for HL-60 cells because we (50) and others (59) have shown that HL-60 cells are proficient for DNA DSB repair. Together, these observations argue that DNA PK may have (at least) two roles in mammalian cells: a function required for DNA DSB repair and a second function required for G 2 checkpoint transition following DNA damage.