If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by Grant BCRP 1RB-0381A (to T. K.-S.) from the Breast Cancer Research Program, University of California, Sankyo Co. Ltd., Japan, and the United States Department of Energy Contract DE-A C03-76SF00098 (to T. K.-S.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Recipient of BCRP Postdoctoral Fellowship 3FB-0053.
Genomic sequences with a cluster of ATC sequence stretches where one strand consists exclusively of well mixed As, Ts, and Cs confer high base unpairing propensity under negative superhelical strain. Such base unpairing regions (BURs) are typically found in scaffold or matrix attachment regions (SARs/MARs) that are thought to contribute to the formation of the loop domain structure of chromatin. Several proteins, including cell type-specific proteins, have been identified that bind specifically to double-stranded BURs either in vitro or in vivo. By using BUR-affinity chromatography to isolate BUR-binding proteins from breast cancer SK-BR-3 cells, we almost exclusively obtained a complex of poly(ADP-ribose) polymerase (PARP) and DNA-dependent protein kinase (DNA-PK). Both PARP and DNA-PK are activated by DNA strand breaks and are implicated in DNA repair, recombination, DNA replication, and transcription. In contrast to the previous notion that PARP and Ku autoantigen, the DNA-binding subunit of DNA-PK, mainly bind to free ends of DNA, here we show that both proteins individually bind BURs with high affinity and specificity in an end-independent manner using closed circular BUR-containing DNA substrates. We further demonstrate that PARP and Ku autoantigen form a molecular complexin vivo and in vitro in the absence of DNA, and as a functional consequence, their affinity to the BURs are synergistically enhanced. ADP-ribosylation of the nuclear extract abrogated the BUR binding activity of this complex. These results provide a mechanistic link toward understanding the functional overlap of PARP and DNA-PK and suggest a novel role for these proteins in the regulation of chromatin structure and function.
Specific regions of genomic DNA that exhibit high affinity to the nuclear matrix in vitro have been identified from various eukaryotic species and are called scaffold/matrix attachment regions (SARs/MARs)
). In particular, studies on MARs flanking the immunoglobulin μ heavy chain (IgH) enhancer showed that these sequences are essential for the B lymphocyte-specific transcription of a rearranged μ gene (
). These MARs have also been shown to be required for extending the domain of chromatin that is accessible to transcription factors and also confer factor access to enhancer-distal positions. The IgH MARs together with the μ enhancer can also induce, in a transcription-independent manner, extensive demethylation across the chromatin domain harboring these elements (
). Several lines of evidence have demonstrated that various genomic segments containing MARs are often found at the boundaries of transcription units and near regulatory elements such as enhancers (see Refs.
), and the choice of the replication origin within the amplified dihydrofolate reductase domain may be dictated by the alteration in the local chromatin structure mediated by attachment of DNA to the nuclear matrix (
). The yeast H4 autonomous replication sequence contains a DNA sequence element with high unwinding potential and is located adjacent to the core replication sequence. This unwinding element is essential for initiation of replication at the H4 origin (
J. D. Alvarez, D. H. Yasui, H. Niida, T. Joh, D. Y. Loh, and T. Kohwi-Shigematsu, submitted for publication.
2J. D. Alvarez, D. H. Yasui, H. Niida, T. Joh, D. Y. Loh, and T. Kohwi-Shigematsu, submitted for publication.
These results demonstrate the significance of BURs and BUR-binding proteins in gene regulation.
Previously a protein with an apparent molecular mass of 114 kDa was isolated from breast cancer cells as a BUR-binding protein. The BUR binding activity of this protein, called p114, correlated with progression of breast tumorigenesis (
). In this study, we used BUR affinity chromatography and obtained, almost exclusively, a mixture of PARP and DNA-PK from the breast cancer cell line, SK-BR-3. PARP has been implicated in DNA repair in response to DNA damage (
). However, we show that either PARP alone or the p70 and p86 subunits of Ku autoantigen as a heterodimeric complex specifically bind with high affinity to IgH MARs recognizing internal sequences. We also show that PARP and the Ku70/86 directly interact and form a protein complex and, as a consequence, synergistically enhance their highly specific binding to these MARs.
Human breast adenocarcinoma cell line SK-BR-3 (ATCC) was used for the studies described here. The cells were grown in McCoy's 5A medium (Life Technologies Inc.) supplemented with 10% fetal bovine serum (Tissue Culture Biologicals) and 1× antibiotic/antimycotic solution (Life Technologies, Inc.).
BUR Affinity Chromatography
BUR affinity columns were prepared by coupling either a 24-bp mutated BUR-containing duplex oligonucleotide (upper strand, 5′-TCTTTAATTTCTACTGCTTTAGAAttc-3′) or a 25-bp wild type BUR-containing duplex oligonucleotide (upper strand, 5′-TCTTTAATTTCTAATATATTTAGAAttc-3′), ligated to an average length of 25, to freshly prepared CNBr-activated Sepharose CL-4B (Amersham Pharmacia Biotech) at 200 μg/ml bed volume as described (
). The differences between the wild type and mutated BUR-containing oligonucleotides are underlined. The affinity matrices were equilibrated with 20 mm Hepes, pH 7.9, 20% glycerol, 100 mm KCl, 0.2 mm EDTA, 0.5 mmdithiothreitol, 0.1 mm phenylmethylsulfonyl fluoride, and 0.1% Nonidet P-40. Nuclear extract from 5 × 108SK-BR-3 cells was prepared as described previously (
) and is referred to as the “0.4 m NaCl extract.” This extract was diluted with equilibration buffer to reduce the salt concentration to 0.2 m. The diluted extract (final volume 15 ml) was mixed with 10 μg/ml sheared salmon sperm DNA and directly loaded onto a MUT MAR column (0.5 × 4 cm). The flow-through of the MUT MAR column was loaded onto WT MAR column (0.5 × 4 cm), washed first with equilibration buffer, and then the same buffer containing 0.3m KCl. Bound proteins were then eluted in a stepwise manner with buffers containing increasing amounts of KCl. Four 0.5-ml fractions were collected for each concentration of KCl used. The fractions containing most of the eluted protein (0.4–0.8m) were combined, dialyzed against 20 mm Tris, pH 7.4, 50 mm NaCl, 10% glycerol, 0.5 mmdithiothreitol, and 0.1 mm phenylmethylsulfonyl fluoride, and loaded onto single-stranded DNA cellulose (Amersham Pharmacia Biotech) column (0.5 ml) equilibrated in same buffer. Bound proteins were eluted with a step gradient of NaCl up to 1 m.
Proteins were estimated using QUANTIGOLD reagent (Diversified Biotech) as described by the manufacturer. SDS-polyacrylamide gel electrophoresis was performed using standard procedures, and the proteins were visualized by staining with Silver Stain Plus kit (Bio-Rad). The identity of BUR affinity column purified p70 was obtained by analyzing the high pressure liquid chromatography-purified peptides derived from the tryptic digest of SDS-polyacrylamide gel-purified p70 using an Applied Biosystems 470 peptide sequencer. The N-terminal sequence of p70 tryptic peptides was LYRETNEP and ELVYPPDYN, identical to amino acids 288–295 and 527–535 of human Ku-70, respectively (
). Nonspecific sites on the membrane were blocked by incubation with 5% bovine serum albumin (fraction V, Sigma) in TST (20 mm Tris-Cl, pH 7.4, 0.5 m NaCl, and 0.05% Tween 20). The blots were then incubated with either anti-PARP (1:2500 dilution of rabbit polyclonal antibody H-250 from Santa Cruz Biotechnology Inc.) or anti-Ku70 and anti-Ku86 (1:750 dilution of monoclonal antibody clones N3H10 and 111, respectively, from NeoMarkers). The blots were then incubated with appropriate peroxidase-conjugated secondary antibodies (1:10,000 dilution in TST) and detected by using SuperSignal chemiluminescence kit (Pierce). The quantitation of signals was performed by densitometric analysis of the x-ray films using Molecular Dynamics Personal Densitometer SI and ImageQuant (version 1.11) software.
ADP-ribosylation of Nuclear Extract
In vitroADP-ribosylation of the nuclear extract was performed using modification of a previously published procedure (
). Briefly, 5 ml of nuclear extract containing 25 mg of protein from SK-BR-3 cells was mixed with an equal volume of a buffer containing 200 mmTris-HCl, pH 8, 20 mm MgCl2, and 3 mm dithiothreitol. ADP-ribosylation was induced by the addition of 10 μg of DNase I (Worthington)-activated salmon sperm DNA (Sigma) and 10 μm NAD (Calbiochem) to one-half of the extract. To the other half of the extract (control), only activated DNA was added. Both batches of extracts were incubated for 5 min at 37 °C, followed by addition of 1 mm 3-aminobenzamide (Calbiochem) and chilling on ice. BUR affinity chromatography was performed separately for the ribosylated and the control extract as described above except that the elution was carried out in a single step with 1 m KCl in equilibration buffer. Southwestern analysis was performed as described previously (
Fifty μg of 0.4 m NaCl nuclear extract from SK-BR-3 cells was diluted with an equal volume of PBS containing 0.2% Nonidet P-40 and precleared with mouse IgG and protein A/G plus beads (Pierce). Briefly, 1 μl of 0.1 mg/ml mouse IgG 1 and IgG 2a each (Sigma) were added to the extract and incubated at 4 °C for 1 h, followed by the addition of 10 μl of protein A/G plus beads and incubation at 4 °C for 1 h on a rotating shaker. The precleared extract was incubated with 2 μl of either anti-PARP (clone CII-10, PharMingen) or anti-Ku (clone 162, NeoMarkers) for 8 h at 4 °C, followed by addition of 10 μl of protein A/G plus beads and further incubation at 4 °C for 8 h. The beads were then washed four times with 500 μl of PBS each containing 0.1% Nonidet P-40. The washed beads were resuspended in 20 μl of 2× SDS-PAGE sample buffer and incubated at 95 °C for 5 min. Solubilized immunoprecipitates were obtained by centrifugation and analyzed by 7.5% SDS-PAGE. Immunoprecipitates obtained from an equal amount of nuclear extract were subjected to immunoblot analysis, and the entire amount of immunoprecipitated protein was loaded on the gel to compare the results. For GST pull-down assay, the NaCl concentration in the reaction mixtures was adjusted to 0.2 m, and the glutathione-Sepharose beads (Amersham Pharmacia Biotech) were washed four times with 500 μl of PBS each, and the immunoprecipitates were subjected to SDS-PAGE analysis as described above.
DNA Binding Analysis by EMSA
Multimeric wild type and mutant MAR DNA fragments were prepared as described (
). Minicircles were prepared by blunt end ligation of these DNAs using 20 units/μl T4 DNA ligase (New England Biolabs) followed by treatment with 5 units/μl exonuclease III (New England Biolabs) to eliminate DNA molecules with free ends. The exonuclease III-resistant minicircles were gel-purified and were also confirmed to be resistant to S1 nuclease. Binding reactions with both linear and circular DNAs contained 50–100-fold excess ofXmnI-ScaI fragment of pBluescript as competitor DNA (
). After incubation at 25 °C for 15 min, the reaction products were loaded on 6% native polyacrylamide gels and electrophoresed at 150 V in 0.5× Tris borate/EDTA buffer and visualized by autoradiography of the dried gel. Data were quantified by densitometric analysis of the autoradiograms as described above.
Isolation of BUR-binding Proteins from SK-BR-3 Cells
For purification of BUR-binding proteins, we used a combination of mutated and wild type BUR columns as described previously (
). For the preparation of BUR affinity columns, we used a 25-bp sequence derived from the MAR located 3′ of the IgH enhancer containing the core unwinding element. The mutated BUR column was prepared using a 24-bp sequence derived from the same region as the 25-bp sequence except that the core unwinding element was mutated (
Nuclear extract from SK-BR-3 breast cancer cells was passed successively through a non-MAR (mutated BUR) and a MAR (wild type BUR) affinity column to select for BUR-specific binding proteins. A 114-kDa MAR-binding protein from breast cancer cells specifically bound to a double-stranded BUR affinity column (
). Interestingly, we found that three other polypeptides of apparent molecular masses of 460, 86, and 70 kDa almost exclusively copurified with p114 from a BUR affinity column (Fig.1A). Peptide microsequence analysis of p70 revealed its identity with the p70 subunit of Ku autoantigen (
). This result suggested that the other proteins that copurified may represent individual polypeptide subunits of DNA-PK. We next performed Western blot analysis of various fractions from either wild type (WT) MAR or mutated (MUT) MAR affinity columns to compare the binding of PARP and Ku and to verify the immunological identity of the purified proteins. Since we used identical amounts (20 μl each) of the column fractions for the immunoblot analysis, the amount of PARP and/or Ku70/86 present in any given fraction could be directly quantitated and compared by the corresponding signal on the immunoblots. We typically observed that less than 10% of the input amount of PARP and Ku70/86 was bound to the MUT MAR column as judged by the amount of protein in the column flow-through (Fig. 1B, compare lanes 1 and 2 in MUTMAR panels), whereas virtually all of PARP and Ku70/86 were retained by the WT MAR column as judged by the weak signals in the column flow-through (Fig. 1B, lane 2 in WTMAR panels). Both columns were then washed with equilibration buffer containing 0.1m KCl to remove loosely bound proteins. Up to 30% of input PARP and Ku were removed at this step from both WT and MUT MAR columns (Fig. 1B, lane 3 in respective panels). The columns were then eluted in two steps of increasing concentration of KCl to collect bound proteins. The bound PARP and Ku70/86 were almost quantitatively eluted from the WT MAR column with 0.3 mKCl- and 1 m KCl-containing buffers (Fig. 1B, lanes 4 and 5, respectively, in WT MAR panels). In contrast, extremely low amounts of PARP and Ku70/86 were left to be eluted from the MUT MAR column under identical conditions (Fig. 1B, lanes 4 and 5 in MUTMAR panels). Quantification of the immunoblot signals corresponding to 0.3 and 1 m KCl eluates from both these columns revealed that the WT MAR column bound at least 10-fold more of PARP and Ku as compared with that of the MUT MAR column. These results suggest that both PARP and Ku70/86 exhibit selective binding to the WT MAR column over the MUT MAR column. The immunoblot analysis of the 1 m KCl eluate from the WT MAR column also confirmed the immunological identity of PARP and both subunits of the Ku autoantigen (Fig. 1B, lane 5 in WTMAR panels). In addition, by using an antibody against the catalytic subunit of DNA-PK, we confirmed the identity of p460 as the catalytic subunit of DNA-PK (data not shown). All three of the proteins that copurified were thus subunits of DNA-PK, suggesting that PARP might be physically associated with DNA-PK by interacting with at least one of its subunits in vivo. To test this hypothesis, we attempted to separate the individual components of this putative complex by employing ion exchange column chromatography. However, we observed that PARP and Ku70/86 always copurified through a variety of matrices such as DEAE-cellulose and phosphocellulose (data not shown). This copurification was observed in the presence of up to 0.6m KCl and 0.1% Nonidet P-40, the conditions that we typically employed for elution of these proteins from the BUR affinity matrix. Separation of these proteins to near-homogeneity was ultimately achieved by taking advantage of the extremely high affinity of PARP to single-stranded DNA (ssDNA). Both Ku70/86 and PARP bound the ssDNA cellulose column. However, as depicted in Fig. 1C, p70 and p86 subunits of Ku autoantigen co-eluted at 0.4 m NaCl (Fig. 1C, lane 2), whereas PARP eluted at 0.7 mNaCl (Fig. 1C, lane 3) from the ssDNA cellulose matrix.
ADP-ribosylation Abolishes the BUR Binding Activity of PARP
Upon autoribosylation, PARP has been shown to lose its DNA end binding activity, and this feedback mechanism has been implicated in base excision repair (
). We therefore tested the effect of autoribosylation of PARP on its double-stranded BUR binding activity. ADP-ribosylation was stimulated in nuclear extract of SK-BR-3 cells by the addition of NAD and nicked DNA, and the ribosylated extract was loaded successively onto mutated and wild type BUR affinity columns. After washing the columns with equilibration buffer, bound proteins were eluted in two steps using 0.3 m KCl- and 1m KCl-containing buffer. Both PARP and Ku70/86 from the ADP-ribosylated extract bound neither the MUT MAR column nor the WT MAR column and were quantitatively recovered in the flow-through of both columns as revealed by immunoblot analysis (Fig.2A, lanes 2 and3,respectively). Washing the columns with the equilibration buffer did not yield either of the two proteins indicating loss of loose binding (Fig. 2A, lanes 4 and 5). Elution with 0.3 m KCl- and 1 m KCl-containing buffers did not elute any of the two proteins (Fig. 2A, lanes 6–9) confirming the lack of binding of both Ku and PARP to either column. Based on these data, we conclude that ADP-ribosylation of SK-BR-3 nuclear extract resulted in the loss of the BUR binding activity of both PARP and Ku.
Next, we examined the effect of varying levels of PARP autoribosylation upon its BUR binding activity by Southwestern analysis. Autoribosylation of column purified preparation of PARP was induced by the addition of DNase I-activated salmon sperm DNA and various amounts of unlabeled NAD. The migration of ADP-ribosylated PARP on denaturing gels was retarded exponentially as a function of NAD concentration (Fig. 2B, lanes 2–9). After resolving the ADP-ribosylated PARP by SDS-PAGE, the protein was electroblotted onto PVDF membrane and renatured in situ. The membrane was then incubated with end-labeled BUR-specific DNA probe (WT(25)7). As depicted in Fig. 2C, PARP abruptly lost its BUR binding activity upon ADP-ribosylation at concentrations of NAD exceeding 10 μm(Fig. 2C, lanes 7–9). Intriguingly, this threshold value is approximately 100-fold lower than the intracellular concentration of NAD (
), and yet both endogenous Ku70/86 and PARP do bind to the BUR column (Fig. 1A). These results suggest that in vivo ADP-ribosylation of PARP may be tightly regulated to modulate its BUR binding activity.
Physical Interaction between PARP and Ku Autoantigen in Vivo and in Vitro
The BUR affinity co-purification profile of PARP and Ku and the loss of their binding capacity to the BUR affinity matrix upon ADP-ribosylation of the nuclear extract strongly suggested that PARP and Ku form a protein complex, either directly or mediated by DNA. Direct physical association between Ku and PARP is unprecedented. To first examine whether PARP and Ku interact physically in vivo, we employed a coimmunoprecipitation strategy using SK-BR-3 nuclear extracts. Immunoprecipitates obtained from an equal amount of nuclear extract were subjected to Western blot analysis for the identification of interacting partner(s). Antibody clone CII-10, which recognizes the N-terminal DNA binding domain of PARP, was able to immunoprecipitate both Ku and PARP (Fig.3A, lane 3). For immunoprecipitating Ku autoantigen, we used an antibody (clone 162) that recognizes the dimer interface of its p70 and p86 subunits (
). This antibody not only immunoprecipitated Ku autoantigen, but also pulled down PARP along with it (Fig. 3A, lane 4) indicating that Ku and PARP are indeed physically associated in vivo. The coimmunoprecipitation signals for both PARP and Ku obtained with either antibody clone CII-10 or clone 162 were significantly greater than that with mouse IgG (Fig. 3A, lane 2). It could be argued that the physical interaction between these proteins is merely an artifact due to their binding to the ends of the same DNA molecule. However, it is most likely not the case because proteins were extracted from the nuclei of SK-BR-3 cells using a mild procedure to avoid genomic DNA contamination. The extraction procedure involved isolation of nuclei at low ionic strength followed by selective extraction of proteins by incubating these nuclei in a buffer containing 0.42m NaCl (
). The absence of trace amounts of genomic DNA within this extract was confirmed by the total failure of ADP-ribosylation by endogenous PARP which strictly requires DNA to activate it. ADP-ribosylation in this extract was induced only after exogenous DNA was added (data not shown).
To demonstrate unequivocally the physical association between Ku and PARP in the absence of DNA and to verify whether these two proteins interact in vitro, we incubated bacterially overexpressed and purified glutathione S-transferase (GST)-fused PARP together with ssDNA cellulose column purified Ku70/86 and captured with glutathione-Sepharose beads. Recombinant PARP associated with Ku70/86 regardless whether purified native Ku70/86 (Fig. 3B, lane 3) or crude 0.4 m nuclear extract (Fig. 3B, lane 2) was used as a source of Ku autoantigen. We performed several controls to ensure the validity of this interaction. First, proteins in the 0.4m extract alone did not display any nonspecific binding to the beads (Fig. 3B, lane 4). A parallel pull-down assay with GST did not capture Ku70/86 (Fig. 3B, lane 5). Western blot with anti-p110 Rb (Fig. 3B, lowermost panel) served as an additional control indicating that the retinoblastoma protein was not associated with GST-PARP or with glutathione-Sepharose beads under identical conditions. Collectively, these experiments demonstrate that the physical interaction between Ku autoantigen and PARP occurs independently of DNA.
Preferential Binding of PARP and Ku to the 3′ MAR within the IgH Enhancer
PARP has been known to bind solely to nicks and ends of DNA (
), PARP has not yet been shown to bind directly to double-stranded DNA in a sequence-specific manner. To elaborate the apparent specificity of PARP and Ku70/86 for the BUR DNA in the affinity matrix, we next performed electrophoretic mobility shift assays (EMSA) with the purified proteins using well characterized MARs surrounding the IgH enhancer region (
) (Fig.4A). Interestingly, Ku-DNA and PARP·DNA complexes were observed with both 5′ (Fig. 4B, top panel) and 3′ (Fig. 4B, bottom panel) linear MAR substrates but not with the enhancer under the concentrations indicated (Fig. 4B, middle panel). It should be noted here that Ku70/86 did bind the enhancer fragment, only when its concentration was raised to at least 20-fold, above the Kd value for the IgH 3′ MAR, most probably reflecting end-related binding (data not shown). Similar analysis using PARP at up to 5-fold of itsKd value for the IgH 3′ MAR indicated complete lack of binding to the enhancer fragment (data not shown), suggesting that end binding by PARP must occur at much higher concentrations. Additionally, Ku70/86 was able to clearly distinguish wild type (
)2 (data not shown). Between the two IgH MARs, the 3′ MAR appeared to be a better substrate for both PARP and Ku70/86 with at least a 2-fold lower dissociation constant (Kd) as compared with that of the 5′ MAR, consistent with the presence of a core unwinding element in the 3′ MAR (
). Surprisingly, the binding affinities of Ku70/86 to these MARs were at least 15-fold higher (Kd of 0.05 and 0.2 nm for the 3′ and 5′ MAR, respectively) than those of PARP (Kd of 1.8 and 3 nm for the 3′ and 5′ MAR, respectively). Intriguingly, this observation is in direct contrast with the BUR affinity elution profile of the two proteins which indicated that PARP required up to 1 m KCl for its complete elution from the matrix, whereas almost all of Ku was eluted with 0.6 m KCl. We are currently investigating the mechanisms underlying the binding of Ku 70/86 and PARP to BURs in vivo.
Direct binding of Ku 70/86 and PARP to the primary BUR sequence was confirmed by using covalently closed DNA minicircles as substrates. We chose the 3′ MAR for circularization since both Ku and PARP showed higher affinity for this substrate in linear form. These circular substrates were predigested with exonuclease III and S1 nucleases to eliminate the possible contribution of any structural features or nicks (data not shown). Interestingly, both Ku70/86 and PARP bound the 300-bp circular IgH 3′ MAR (Fig. 4C, upper panel) with affinities similar to the corresponding linear substrate (Kd values of 0.05 and 2 nm,respectively). Up to three complexes could be identified, indicating that these proteins were able to bind to one or more sites within this substrate perhaps in a manner similar to that of the T cell MAR/SAR-binding protein SATB1 (
). However, both Ku and PARP failed to bind a minicircle comprising the core enhancer (Fig. 4C, lower panel) as well as a control circle prepared by ligation of a 400-bp HinfI-XbaI fragment of pUC19 (data not shown). Furthermore, both Ku and PARP also bound a circular concatemeric BUR containing seven copies of the core unwinding element of the Cμ 3′MAR (
)7), with high affinity (Fig. 4C, lower panel). The binding reached saturation in a protein concentration-dependent manner and occurred in the presence of a 50-fold excess of linear DNA competitor. Up to seven different complexes could be identified, indicating increasing occupancy of the BURs within the minicircle. These data convincingly demonstrate that both Ku70/86 and PARP recognize the primary sequence of the BURs and not the ends of DNA.
Physical Association between PARP and Ku Promotes Their Synergistic Binding to BURs
We next examined the possible effects of physical association between Ku and PARP on their BUR binding activity. We performed EMSA analysis using 300 bp of linear IgH 3′ MAR under conditions wherein Ku70/86 and PARP were added simultaneously at concentrations at which neither of them individually bound more than 50% of the DNA substrate (Fig. 5A, lanes 2 and 4), and the possibility of end-mediated binding is virtually eliminated. Contrary to the previous notion that Ku and PARP would compete for their DNA substrates (
), we found that at least with respect to MARs, they exhibit synergistic binding. When added together, Ku70/86 and PARP bound more than 90% of the labeled DNA probe (Fig. 5A, lane 5), indicating an increase of at least 3-fold over the summation of their individual binding. Increasing the amount of added Ku (Fig. 5A, lane 3) led to further enhancement of binding activity (Fig. 5A, lane 6). It appeared that at 1.8 nm, PARP itself did not bind very well to the MAR substrate (Fig. 5A, lane 4); however, its binding activity was much enhanced by the addition of only as little as 0.05–0.1 nm Ku70/86 (Fig. 5A, lanes 5 and6). A similar enhancement in binding was observed using the circular IgH 3′ MAR DNA template. However, since in the case of this substrate it was difficult to score the intermediate protein-DNA complexes, and it was difficult to measure the synergy of binding (data not shown).
To test whether this increase in the binding activity was due to actual physical association between Ku and PARP versus the proteins independently occupying the binding sites in the substrate, we used a truncated form of PARP that does not interact with Ku.
S. Galande, T. Nishimura, and T. Kohwi-Shigematsu, unpublished observations.
In a GST pull-down assay, the recombinant DNA binding domain (DBD) of PARP fused to GST captured a barely detectable amount of Ku (Fig. 5B, lane 2), indicating that they do not form a complex. EMSA analysis indicated that the recombinant DBD of PARP alone could bind to the MAR substrate in a manner identical to that of native PARP (Fig. 5A, compare lane 7 with 4). However, addition of the same amounts of Ku that led to an increase in the binding activity of native PARP did not significantly affect the binding activity of truncated PARP (Fig. 5A, lanes 8 and 9), suggesting that the synergistic effect that we observed using full-length PARP and Ku reflects a specific effect and not a nonspecific stabilization of binding. Taken together, these results indicated that physical association between Ku and PARP is indeed critical for their synergistic binding to MARs.
) have demonstrated sequence-specific DNA binding by Ku70/86. In the present study, we utilized BUR affinity chromatography, and we showed that PARP, Ku70/86, and the catalytic subunit of DNA-PK were almost exclusively copurified. PARP and DNA-PK together preferentially bound only the wild type BUR affinity matrix and not the mutated BUR affinity matrix. In contrast to the previous notion that PARP only binds to ends and nicks of DNA (
), we show that PARP specifically binds with high affinity to the BURs surrounding the IgH enhancer using circular DNA substrates and does not bind the non-BUR core enhancer. Similar binding specificity was also observed for Ku70/86 using circular DNA substrate. Apparently, under the experimental conditions employed in this study, the affinity of either PARP or Ku70/86 to BURs is much higher than their affinity to the free ends of the linear DNA.
Several lines of evidence implying DNA-PK and PARP interaction have been reported previously. Studies utilizing PARP and DNA-PK-deficient cells have suggested a functional overlap between these two proteins to mitigate genomic damage caused by DSBs (
) demonstrated co-purification of DNA-PK, PARP, and human heterogeneous nuclear ribonucleoprotein-U proteins by employing antibody affinity chromatography using antibody against the p86 subunit of Ku autoantigen. However, they argued that the interaction among these proteins was mediated only by virtue of binding to free ends of DNA in cis. Furthermore, they also demonstrated that ADP-ribosylation of DNA-PK by PARP in vitro stimulated its kinase activity, suggesting their functional interaction in response to DNA damage. We have shown that Ku70/86 and PARP directly interact to make a protein complex independently of DNA and synergistically enhance their binding to BURs. This result, to our best knowledge, is the first demonstration of a direct physical association between Ku and PARP, and their synergistic binding to DNA recognizing specific internal DNA sequences.
The DNA binding domains of PARP and Ku differ significantly as follows: that of PARP consists of two zinc fingers and resides in the N-terminal portion of the same 114-kDa polypeptide that harbors the catalytic site (
). The DBD of Ku autoantigen consists of two separate domains each residing at either terminus of its p70 subunit. The N-terminal DBD of p70 binds DNA only upon dimerization with p86, whereas the C-terminal DBD is p86-independent (
). Results of our Southwestern analysis to monitor the BUR binding activity of individual Ku subunits indicated that Ku70 or Ku86 alone do not bind BURs (data not shown), whereas PARP (a 114-kDa homodimeric protein) does exhibit BUR binding activity (this study). The BUR binding activity of Ku70/86 therefore must be dependent on heterodimerization of p70 and p86. Despite the obvious differences in their DNA binding domains, PARP and Ku70/86 remarkably seem to bind to similar targets. It is noteworthy, however, that in the in vitro binding experiments, we have observed that the binding affinity of Ku70/86 for BURs is at least 10-fold higher than that of PARP. Surprisingly, the BUR affinity profile indicated that PARP is bound much more strongly than Ku70/86. Since extremely low amounts of Ku70/86 are sufficient to increase the binding affinity of PARPin vitro, it is conceivable that Ku70/86 may assist PARP in binding to BURs; however, once PARP is bound it does not dissociate easily from BURs. The effects of post-translational modification(s) of PARP other than ADP-ribosylation on its BUR binding activity need to be examined.
The apparent specificity of Ku70/86 and PARP for the BUR sequences and their synergistic binding to these regions may have important biological implications. PARP has been found within the DNA replication (
). In addition, a role for PARP in aging has also been suggested. PARP associates with p53 protein in vivo, and PARP activation leads to accelerated loss of telomeres, activation of p53, and premature senescence (
M. Ramakrishman, P. A. DiCroce, A. Posner, J. Zehng., T. Kohwi-Shigematsu, and T. Krontiris, submitted for publication.
Our finding that PARP and Ku70/86 proteins form a complex suggest that PARP·DNA-PK complex may be commonly found to be involved in some or all of the functions mentioned above. At least some of the seminal roles of PARP and Ku70/86 may be mediated by their association with BURs, which are the key structural elements of MARs. We speculate that the PARP·DNA-PK complex may bridge MARs with these multiprotein machineries.
It is likely that the PARP·DNA-PK complex has a role in chromatin structure. ADP-ribosylation of histones by PARP at the site of DNA damage has been shown to cause relaxation of the local chromatin superstructure (
). Thus, studies on the association between Ku and PARP and their high affinity binding to BURs may provide novel insights into some aspects of the mechanistic link between chromatin structure and various other functions involving DNA.
We thank Drs. Judith Campisi for critical reading of this manuscript, Mark Smulson for kindly providing human PARP cDNA construct, and Guy Poirier for the kind gift of polyclonal anti-PARP antibody that was used in the initial stages of this work.