Specific interaction of DNA polymerase beta and DNA ligase I in a multiprotein base excision repair complex from bovine testis.

Base excision repair (BER) is a cellular defense mechanism repairing modified bases in DNA. Recently, a G:U repair reaction has been reconstituted with several purified enzymes from Escherichia coli (Dianov, G., and Lindahl, T. (1994) Curr. Biol. 4, 1069-1076). Using bovine testis crude nuclear extract, we have shown that G:U is repaired efficiently in vitro, and DNA polymerase β (β-pol) is responsible for the single nucleotide gap-filling synthesis (Singhal, R. K., Prasad, R., and Wilson, S. H. (1995) J. Biol. Chem. 270, 949–957). To investigate potential interaction of β-pol with other BER protein(s), we developed affinity chromatography matrices by cross-linking purified rat β-pol or antibody against β-pol to solid supports. Crude nuclear extract from bovine testis was applied to these affinity columns, which were then extensively washed. Proteins that bound specifically to the affinity columns were co-eluted in a complex with β-pol. This complex had a molecular mass of approximately 180 kDa and was able to conduct the complete uracil-initiated BER reaction. The BER complex contained both β-pol and DNA ligase I. An antibody to β-pol was able to shift the complex in sucrose gradients to a much larger molecular mass (>300 kDa) that again contained both β-pol and DNA ligase I. Furthermore, DNA ligase I and β-pol were co-immunoprecipitated from the testis nuclear extract with anti β-pol IgG. Thus, we conclude that β-pol and DNA ligase I are components of a multiprotein complex that performs BER.

Base excision repair (BER) 1 is initiated by enzymatic removal of a damaged or inappropriate base residue in DNA by a DNA glycosylase. This class of enzymes recognizes and removes a variety of single base lesions in DNA (1). Uracil-DNA glycosylase (UDG), which catalyzes the removal of uracil from a G:U mismatch to initiate BER, is the most extensively studied of these enzymes (2)(3)(4). The apurinic/apyrimidinic (AP) site is generated by DNA glycosylase (5) and also by spontaneous cleavage of the glycosidic bond, in particular those involving purines (6). Both procaryotes and eucaryotes contain AP endonucleases (APE) that incise the phosphodiester backbone of DNA at the 5Ј side of an AP site leaving a 3Ј-OH and 5Јdeoxyribose phosphate (5,7). Subsequently, the 5Ј-deoxyribose phosphate is removed either by a deoxyribose phosphodiesterase (dRpase) or by a 5Ј33Ј exonuclease, generating a onenucleotide gap with 3Ј-OH and 5Ј-phosphate termini (8 -11). The single nucleotide gap is then filled by DNA polymerase I in Escherichia coli or by DNA polymerase ␤ (␤-pol) in mammalian cells (12)(13)(14). Finally, the nick is sealed by DNA ligase. Dianov and Lindahl (12) reconstituted the uracil-initiated BER reaction with five purified enzymes from E. coli. Recently, Singhal et al. (13) have demonstrated ␤-pol-dependent BER reaction in mammalian cells with partially purified bovine proteins and T4 DNA ligase. Since mammalian cells contain multiple species of DNA ligase (15), one or more of these enzymes may complete the BER reaction (16). Studies with mutant mammalian cell lines have implicated both DNA ligase I and DNA ligase III in repair of single-strand breaks generated by BER enzymes (16,17).
Mammalian ␤-pols have been cloned and overexpressed in E. coli (31)(32)(33)(34). The recombinant proteins are fully active in DNA synthesis and have been exploited for structure-function studies (35)(36)(37)(38). We recently generated monoclonal antibodies and several neutralizing polyclonal antibodies specific for mammalian ␤-pols (13,39). Using these reagents, we have investigated potential interactions among mammalian BER proteins. Our results indicate that BER may be mediated, at least in part, by a multiprotein complex that contains ␤-pol and DNA ligase I.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP and [␣-32 P]dCTP were purchased from ICN Radiochemicals. High performance liquid chromatography-purified synthetic oligonucleotide primers were obtained from Genosys Biotechnologies Inc. Epoxy-activated Sepharose 6B and protein A-Sepharose 4B were from Pharmacia Biotech, Inc. Rat ␤-pol was purified as described (33,34). Recombinant human DNA ligase I was purified from baculovirus-infected insect cells (40). The rabbit polyclonal antisera raised against bovine DNA ligase I and against a conserved peptide sequence that is present in all eucaryotic DNA ligases have been de-scribed previously (41). The peptide antiserum recognizes DNA ligase I, II, III (42) and presumably will also recognize DNA ligase IV (43). Antiserum specific to intact ␤-pol was raised in rabbits as described (13). Frozen bovine testis was purchased from J. Schmidt Co. (Baltimore, MD) and stored at Ϫ80°C until use.
Nuclear Extract Preparation-Nuclear extract was prepared from bovine testis as described (44). Briefly, 500 g of bovine testis was minced in 3 volumes of buffer A and homogenized with a blender. The homogenate was pelleted by centrifugation at 10,000 ϫ g for 20 min. Pellet was resuspended in 1.5 volumes of buffer B and blended for 30 s. The homogenate was centrifuged at 100,000 ϫ g for 1 h. The clear supernatant fraction was brought to 50% saturation by adding solid (NH 4 ) 2 SO 4 with stirring. The precipitate was recovered by centrifugation for 20 min at 15,000 ϫ g. The pellet was resuspended in buffer D and dialyzed against the same buffer for 14 h. Nuclear extract was then clarified by centrifugation at 10,000 ϫ g for 20 min. This clear supernatant was referred to as the crude nuclear extract.
Protein Affinity Column Chromatography-Purified rat ␤-pol (14 mg) or bovine serum albumin (BSA) (14 mg) was coupled to approximately 10 ml of epoxy-activated Sepharose 6B according to the manufacturer's instructions (Pharmacia). After equilibration with buffer D, bovine testis nuclear extract (5 ml containing approximately 50 mg of protein) was loaded onto each column repeatedly. The columns were then washed extensively (Ͼ20 column volumes) with buffer D. The bound proteins were eluted from these columns by stepwise increasing NaCl concentrations. Column fractions were dialyzed against buffer D, concentrated to 1 ml with Centriprep-10 (Amicon), and stored at Ϫ80°C until use.
Immunoaffinity Column Chromatography-Affinity purified anti-␤pol IgG (2 mg) and purified preimmune IgG (2 mg) from the same rabbit were covalently cross-linked to epoxy-activated Sepharose 6B according to the manufacturer's suggested protocol. Equal volumes (5 ml) of bovine testis nuclear extract were applied to each column. After extensive washing with buffer D (Ͼ20 column volumes), bound proteins were eluted from the columns as described above.
Sucrose Density Gradient Sedimentation-The nuclear extract preparations were dialyzed against buffer C without glycerol but containing 50 mM NaCl and then layered on a 15-40% sucrose gradient. Centrifugation was performed in a Beckman SW55 rotor at 50,000 rpm at 4°C for 16 h. Fractions, approximately 250 l each, were collected from the bottom of the tube. A parallel gradient was calibrated with 50 g each of ␤-pol (39 kDa), bovine serum albumin (70 kDa), aldolase (158 kDa), catalase (240 kDa), and ferritin (440 kDa). The identity and position of these protein markers were determined by Coomassie Blue staining after separating by SDS-PAGE. The refractive index of each fraction was monitored to obtain an accurate assignment of molecular mass of the proteins in the different gradients.
SDS-PAGE and Immunoblotting Analyses-SDS-polyacrylamide gel electrophoresis was performed according to the method of Laemmli (45) using 12.5% discontinuous slab gels. The proteins were electrophoretically transferred onto a nitrocellulose membrane (Bio-Rad) in a transblotting apparatus (Bio-Rad) according to the manufacturer's instructions. Subsequently, the blot was probed either with monoclonal antibody 18 S (monoclonal antibody to rat ␤-pol) or rabbit polyclonal anti-DNA ligase I/DNA ligase II and III. Anti-mouse immunoglobulin G (IgG) or goat anti-rabbit IgG conjugated to horseradish peroxidase at 1:5,000 dilution was used as the secondary antibody. Antigen-antibody complexes were detected by enhanced chemiluminescence (Amersham Corp.). The amount of ␤-pol was determined in terms of integrated optical density (IOD) which was measured by Millipore BioImage Visage Software on a Sun Sparc Station II Workstation (BioImage Products, Ann Arbor, MI).
Immunoprecipitation-Immunoprecipitation was performed as described (46). Briefly, purified ␤-pol was mixed with recombinant DNA ligase I or BSA in 50 l of buffer E, as indicated in Fig. 5. The mixture was incubated on ice for 1 h. The complex formed between ␤-pol and DNA ligase I was immunoprecipitated with affinity purified anti-␤-pol IgG or preimmune IgG. The immunocomplex was adsorbed on protein A-Sepharose 4B beads. The beads were washed three times with buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1.0% Nonidet P-40, dissolved in SDS-sample buffer, heated for 3 min, and the soluble proteins separated by SDS-PAGE. The proteins were electroblotted, and the blot was developed with DNA ligase I immune serum. For immunoprecipitation of ␤-pol and DNA ligase I complex from the crude nuclear extract, 500 l of nuclear extract was incubated with 10 l of affinity purified antibody to ␤-pol or preimmune IgG, and processed as above. Blot was developed with either immune serum of ␤-pol or DNA ligase I.
AP Endonuclease Assay-A 49-bp oligodeoxynucleotide containing a single uracil at position 21 was obtained from Operon Technologies, Inc. and had the sequence 5Ј-AGCTACCATGCCTGCACGAAUTAAG-CAATTCGTAATCATGGTCATAGCT-3Ј. The oligodeoxynucleotide containing uracil was 5Ј-end labeled with [␥-32 P]ATP and annealed to its complementary strand by heating the solution at 90°C for 3 min, followed by slow cooling to 25°C. 32 P-Labeled duplex oligonucleotide was separated from unincorporated [␥-32 P]ATP using a Nensorb-20 column according to the manufacturer's suggested protocol. The radiolabeled oligonucleotide was resuspended in 70 mM HEPES, pH 8.0, 1 mM EDTA, and 1 mM DTT. The DNA (50 ng) solution was treated with 1 unit of E. coli uracil-DNA glycosylase (Boehringer Mannheim) for 45 min at 37°C to create double-stranded DNA containing a site-specific AP site. AP endonuclease reaction mixture (10 l) contained 50 mM HEPES, pH 7.9, 3 mM MgCl 2 , 2 mM DTT, and 13.5 M 32 P-labeled double-stranded oligonucleotide containing an abasic site at position 21. The reaction was initiated by adding 1 l of the sucrose gradient fractions. Incubation was at 37°C for 5 min. The reaction was terminated by adding 5 l of gel loading buffer (95% formamide, 20 mM EDTA, 0.02% bromphenol blue, and 0.02% xylene cyanol). After incubation at 95°C for 2 min, the reaction products were separated by electrophoresis in a 15% polyacrylamide gel containing 7 M urea in 89 mM Tris-HCl, 89 mM boric acid, and 2 mM EDTA, pH 8.8, and visualized by autoradiography. The amount of the AP endonuclease product was measured in terms of IOD of the radioactive bands and determined as above.
Base Excision Repair Assay-In vitro BER was performed essentially as described (13). Briefly, the standard reaction mixture (50 l) contained 100 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA, 2 mM ATP, 0.5 mM NAD, 5 mM diTris phosphocreatine, 10 units of creatine phosphokinase, 40 nM 51-bp duplex oligodeoxynucleotide substrate containing a uracil residue at position 22, and 0.3 M [␣-32 P]dCTP (specific activity 6.6 ϫ 10 6 dpm/pmol). The BER reaction was initiated by addition of crude nuclear extract (50 g) or appropriate amounts of SDG fractions. The reaction mixture was incubated at 37°C for 10 min, and the reaction was stopped by adding an equal volume of 40 mM EDTA and 80% formamide. After incubation at 95°C for 2 min, the DNA was separated by electrophoresis in a 12% polyacrylamide gel containing 7 M urea in 89 mM Tris-HCl, 89 mM boric acid, and 2 mM EDTA, pH 8.8. Gels were fixed, dried, and autoradiographed to visualize the reaction products.

RESULTS
A complete BER reaction with E. coli proteins has been reconstituted using uracil-DNA glycosylase, AP endonuclease IV, RecJ protein which excises 5Ј-terminal deoxyribose phos-2 R. W. Sobol and S. H. Wilson, unpublished observations. phate, DNA polymerase I, and DNA ligase (12). We have recently demonstrated that a bovine testis crude nuclear extract conducts G:U-initiated base excision repair in vitro inserting dCMP opposite G (13). The DNA synthesis stage of this in vitro repair reaction is blocked by antibody to ␤-pol. Furthermore, purified ␤-pol restores full repair activity to a ␤-pol-depleted system, whereas purified DNA polymerases ␣, ␦, or ⑀ do not (13). These results established that ␤-pol is responsible for the single nucleotide gap-filling DNA synthesis reaction observed in this in vitro base excision repair reaction. In agreement with these biochemical results, studies with a mouse cell line containing a homozygous ␤-pol gene deletion have provided genetic evidence that ␤-pol is responsible for BER gap-filling synthesis in vivo (14).
Analysis of Testis Crude Nuclear Extract for BER Complex(es)-To determine whether the BER reaction involving ␤-pol is mediated by a multiprotein complex, testis crude nuclear extract was sedimented through a linear sucrose density gradient (SDG). After centrifugation, polypeptides in fractions from the gradient were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with monoclonal antibody against ␤-pol (39). Although ␤-pol was distributed throughout the gradient (Fig. 1a), two peaks of ␤-pol with apparent molecular masses of ϳ180 and ϳ40 kDa were observed in fractions 6 -7 and 12-13, respectively. Note that the sucrose gradients used here were calibrated with purified marker proteins (see "Experimental Procedures"), and as expected, the markers sedimented in discrete peaks. The identity and position of the protein markers in different gradients was confirmed by SDS-PAGE and Coomassie Blue staining (data not shown), and the refractive index of each fraction was measured to obtain an accurate assignment of molecular mass of proteins in the different gradients.
To address whether the high molecular weight form of ␤-pol was due to aggregation of ␤-pol monomers and/or nonspecific aggregation of ␤-pol with other polypeptides, we mixed purified ␤-pol with various standard proteins of known molecular mass and subjected the mixture to SDG sedimentation under the same conditions as for the crude nuclear extract. These fractions were then examined for ␤-pol by immunoblotting. Under these conditions, ␤-pol was detected in a single peak with a molecular mass of ϳ40 kDa (Fig. 1b). Thus, the size distribution of ␤-pol in the crude nuclear extract, as revealed by velocity sedimentation, indicated the presence of a ϳ180-kDa complex. Furthermore, fractions 6 and 7 were tested for ability to conduct the G:U-initiated BER reaction using a 51-bp oligonucleotide containing a single uracil at position 22 (Fig. 1c) in one strand. A fully repaired 51-residue molecule with C at position 22 and/or unligated product corresponding to the 22-residue intermediate molecule are the expected products of this in vitro BER system (39). As revealed by the formation of the 51-mer base replacement and ligation product, proteins in fractions 6 and 7 were able to conduct BER. Monomeric forms of ␤-pol, UDG, APE, and DNA ligase in fractions 10 -13 were also able to conduct BER (data not shown). However, these fractions were limiting in DNA ligase activity (see below).
Next, we also evaluated whether ␤-pol was interacting with nucleic acid during the SDG centrifugation analysis (Fig. 1a). The crude nuclear extract was applied to a Q-Sepharose column in a buffer containing 50 mM NaCl. Under these conditions, ␤-pol and other BER proteins including UDG, APE, and DNA ligases were not retained by the column, whereas contaminating nucleic acids were adsorbed. The flow-through fraction was subjected to SDG centrifugation and yielded the same size distribution of ␤-pol as that seen with the crude nuclear extract (data not shown). Taken together, these results initially suggested a complex with a molecular mass of ϳ180 kDa containing ␤-pol and other enzymatic activities required for BER. The slower sedimenting ␤-pol in fractions 11-13 can be attributed to the monomeric form of the enzyme (Fig. 1a).
Interaction of Base Excision Repair Proteins with Affinity Columns-To study interactions between ␤-pol and other BER proteins in the bovine testis crude nuclear extract, we constructed two types of affinity chromatography columns using, on the one hand, purified ␤-pol and, on the other hand, purified anti ␤-pol IgG. Bovine testis nuclear extract, capable of BER in vitro, was applied to the affinity columns.
For the ␤-pol column affinity chromatography, the crude extract was applied to the column and then washed extensively (Ͼ20 column volumes) with buffer containing 100 mM NaCl. Bound proteins were eluted with stepwise additions of buffer containing increasing [NaCl]. A column with bovine serum albumin was used as a negative control. Column fractions were assayed for G:U-initiated BER activity. The results (Fig. 2a) indicated that the proteins bound to the ␤-pol column and eluted with NaCl were able to confer full BER activity, whereas the proteins bound to the BSA column could not. These results also indicated that the various enzymes required for BER were immobilized on the ␤-pol affinity column in the presence of 100 mM NaCl but were eluted with 200 mM NaCl. The putative complex(es) isolated by this column chromatography was, therefore, sensitive to release from ␤-pol by a relatively low NaCl concentration. However, it is noteworthy that the proteins eluted from the ␤-pol affinity column confer full BER activity indicating that the complex also included ␤-pol (see "Discussion").
To further explore complex(es) of ␤-pol and other BER proteins in the crude nuclear extract, we made use of immunoaffinity column chromatography with a ␤-pol antibody. We had shown earlier that our polyclonal antibody to ␤-pol is specific and can completely inhibit both purified ␤-pol and the nuclear extract-based BER activity in vitro (13). Therefore, we used IgG from this antiserum to conduct immunoaffinity chromatography to probe for a ␤-pol-containing protein complex. Purified anti ␤-pol IgG or IgG isolated from preimmune serum of the same rabbit were covalently cross-linked to epoxy-activated Sepharose 6B. Equal volumes of the bovine testis nuclear extract were applied to the columns. After extensive washing (Ͼ20 column volumes) of the columns with buffer containing 100 mM NaCl, bound proteins were eluted with buffer containing increasing concentrations of NaCl. Proteins capable of full BER eluted with 200 mM NaCl (Fig. 2b) from the anti-␤-pol column. In contrast, proteins eluting from the preimmune IgG column failed to show BER activity. The bound proteins from these columns were separated on SDS-PAGE, transferred to a nitrocellulose membrane, and probed with antibody to either ␤-pol, DNA ligase I, or AP endonuclease. The anti-␤-pol column, but not the preimmune IgG column, retained the 39-kDa ␤-pol protein and the 125-kDa DNA ligase I protein (Fig. 2c). In contrast, AP endonuclease was not detected in the fraction from the anti-␤-pol column (Fig. 2c). As we will show below, the BER-proficient complex eluted from the anti-␤-pol column also contains UDG activity. Thus, whereas a combination of individually purified proteins can perform the BER reaction in vitro (13), our results indicate an interaction between BER proteins and ␤-pol, suggesting that at least a portion of the BER enzymes in the testis nuclear extract exist as a complex.
Sedimentation of Bovine Testis BER Complex-To examine the physical properties of the BER complex obtained by immunoaffinity chromatography, the active fraction from the immunoaffinity column was sedimented through a 15-40% sucrose gradient, and the gradient fractions were assayed for BER activity (Fig. 3, a and b) and UDG activity (Fig. 3c). The results of these analyses revealed two peaks of BER and UDG activities, one centered in fractions 6 -7 and the other at fraction 12 corresponding to molecular masses of approximately 180 and 40 kDa, respectively. The presence of the 22-mer intermediate product, which was not detected in assays of the immunoaffinity column eluate (Fig. 2b), suggested that some of the BER proteins that had been in a complex in the original sample had dissociated during the experimental protocol of dialysis, concentration, and/or centrifugation. The profile of the base excision repair activity of these SDG fractions also showed that fractions corresponding to a molecular mass of 100 kDa or less were limiting in DNA ligase activity (Fig. 3, fractions  10-14), since the 22-residue intermediate product molecule accumulated.
Properties of BER Complex-The immunoaffinity purified complex presumably must contain UDG, APE, dRpase, ␤-pol, and DNA ligase activities necessary to complete the BER reaction. The uracil-DNA glycosylase activity appears to be contributed by an enzyme that is present in the immunopurified complex as UDG activity co-sedimented with the complex (Fig.  3c). ␤-Pol presumably contributes both dRpase and DNA synthesis activities (19). 3 To determine the identity of the DNA ligase in the BER complex, the crude nuclear extract was subjected to SDG centrifugation, and the gradient fractions were examined for DNA ligase polypeptides by immunoblotting (Fig. 4, b and c). As expected from experiments described above, ␤-pol was dispersed in fractions 6-14 (Fig. 4a), but it was quantitatively shifted by anti-␤-pol IgG to fractions 2-4 ( Fig.  4a) near the bottom of the gradient. When these fractions were tested by immunoblotting with DNA ligase I antibody, a similar shift of a portion of the DNA ligase I was observed, just as for ␤-pol (Fig. 4b). However, we did not observe any shift in DNA ligase II (Fig. 4c). In these experiments, DNA ligase III was detected at low levels in fractions containing DNA ligase II. These results suggest that ␤-pol was complexed with a portion of the DNA ligase I molecules in the nuclear extract but not with DNA ligase II.
To evaluate the idea that ␤-pol and DNA ligase I can form a complex, purified ␤-pol and DNA ligase I were mixed in 1:1 molar ratio and then immunoprecipitated with antibody to ␤-pol. Antibody-bound material was adsorbed on protein A-Sepharose 4B beads, resolved by SDS-PAGE, transferred to a nitrocellulose membrane, and then blotted with anti-DNA ligase I serum. The results indicate that anti-␤-pol IgG, but not preimmune IgG, co-immunoprecipitated DNA ligase I (Fig. 5a,  lanes 1 and 2, respectively). Control experiments where ␤-pol was replaced with bovine serum albumin and immunoprecipitated with either anti-␤-pol IgG or preimmune IgG failed to show any immunoprecipitation of DNA ligase I (lanes 3 and 4). These results indicate that the 180-kDa BER complex observed in sucrose gradients of crude extract or in proteins isolated from the affinity columns (Figs. 1 and 3) probably represents an association between ␤-pol and DNA ligase I. To further test this idea, bovine nuclear extract was mixed with anti-␤-pol IgG or preimmune IgG to examine co-immunoprecipitation of ␤-pol and DNA ligase I. The immunocomplex was collected on protein A-Sepharose 4B beads, separated by SDS-PAGE, and transferred to a nitrocellulose membrane. The membrane was then probed with antiserum to DNA ligase I (Fig. 5b, lane 2) or anti-␤-pol (Fig. 5c, lane 2). The results show that antibody to ␤-pol co-immunoprecipitated DNA ligase I and ␤-pol, whereas preimmune IgG did not (Fig. 5b, lane 1). Taken together, this result, as well as in vitro co-immunoprecipitation (Fig. 5a), demonstrates that a portion of the ␤-pol and DNA ligase I molecules in bovine testis nuclear extract is present in a complex.

DISCUSSION
Base excision repair is a sequential, multistep DNA repair pathway where a damaged or incorrect base is excised and replaced by a complementary nucleotide, through synthesis by a DNA polymerase. Singhal et al. (13) had demonstrated that uracil-initiated BER can be studied in vitro with purified and partially purified bovine proteins and that the DNA synthesis step is conducted by DNA polymerase ␤. From these results, it was clear that a robust BER system can be reconstituted in vitro with eucaryotic proteins and that at least four enzyme activities are required, UDG, APE, ␤-pol, and DNA ligase. Interactions between individual BER proteins may be important for the regulation of individual steps involved in the BER reaction.
In this paper, we provide evidence of protein-protein interactions among BER proteins, including direct association between ␤-pol and DNA ligase I. To demonstrate a multiprotein complex with bovine testis nuclear proteins, affinity chromatography was employed using ␤-pol or ␤-pol antibody as the immobilized ligand. A protein fraction purified by both types of affinity chromatography was capable of conducting complete uracil-initiated BER in vitro. Thus, these ␤-pol affinity columns were able to retain activities for UDG, APE, ␤-pol/dRpase, and DNA ligase. Biochemical and immunological characterization of the fractions eluted from the affinity columns demonstrated the presence of ␤-pol and DNA ligase. Furthermore, synthesis of the 51-mer product of the in vitro BER system indicated that these columns also retained UDG and AP endonuclease activities, in addition to ␤-pol and DNA ligase. However, the accumulation of the 22-mer DNA ligase substrate, in some cases, indicated that our affinity purified fractions were limiting in DNA ligase activity. It is noted that the affinity purified fraction from the ␤-pol affinity columns contained ␤-pol itself. A similar phenomenon has been observed recently with XPA protein, where Park and Sancar (47) provided two possible explanations. First, XPA might dimerize at the concentrations provided on the XPA column or, alternatively, the ERCC1/ ERCC4 (XPF) complex, when applied to the affinity column, might alternate between complexes with free XPA and XPA on the column, resulting in retardation of free XPA protein and its subsequent co-elution with the ERCC1/ERCC4 (XPF) complex. In any case, ␤-pol appears to behave in a similar fashion as XPA, in the affinity column chromatography.
Having established that the multiple activities required for BER bind to affinity columns, it was interesting to examine the hydrodynamic properties of the BER complex eluted from the affinity columns and compare it with a complex in the initial bovine testis crude nuclear extract. Using velocity sedimentation in linear sucrose gradients, we found that complexes capable of BER were heterodisperse and included species in the ϳ180-kDa range (Figs. 1 and 3). The ϳ180-kDa complex also appeared to contain UDG and AP endonuclease activities in addition to ␤-pol and DNA ligase I. The presence of these activities in a complex of ϳ180 kDa can generally be accommodated. Thus, stoichiometric levels of ␤-pol (39 kDa), UDG (32 kDa), and possibly AP endonuclease (34 kDa) could be present in the 180-kDa complex, since sedimentation of DNA ligase I is consistent with a ϳ80-kDa molecule (41). In the in vitro BER system, synthesis of the 51-mer product molecule requires DNA ligase activity. Sucrose gradient analysis of the immunoaffinity purified BER complex revealed that the fractions corresponding to Ͻ160 kDa showed very little synthesis of the 51-mer product, as compared with synthesis of the 22-mer product (see Fig. 3). This indicated the presence of BER proteins in a complex but with little or no DNA ligase I.
Our data clearly show that BER is mediated at least in part by a multiprotein complex. Furthermore, biochemical and immunological properties of the BER complex indicate that DNA ligase I is a component of the ϳ180-kDa complex. The involvement of DNA ligase I in BER is consistent with the phenotype of the DNA ligase I mutant cell line 46BR and the observed in vitro defect in BER in this cell line (16). We note that DNA ligase I also has been purified as a component of a 21 S DNA replication complex (48,49). The phenotype of the Chinese hamster ovary cell line EM9 suggests a defect in BER at a post-incision stage (17,50). DNA ligase III activity is reduced in this cell line because of low levels of its interacting protein Xrcc1. It appears that there may be two pathways for repairing DNA single-strand breaks, one of which involves the BER complex described here with DNA ligase I and the other which is mediated by XrccI/DNA ligase III and probably poly(ADP-ribose) polymerase (51). Results in Fig. 4 are particularly noteworthy as they demonstrate a linkage between ␤-pol-mediated BER and DNA ligase I but not DNA ligase II. Recent studies indicate that DNA ligase II is encoded by the Lig3 gene and is probably derived from DNA ligase III by proteolysis (40,32,43). The BER complex described here may be mainly responsible for uracil-DNA glycosylase-initiated repair, whereas the Xrcc1/DNA ligase III/poly-(ADP-ribose) polymerase pathway may deal with DNA singlestrand breaks. One compelling reason to have BER carried out by a multiprotein complex is to protect the DNA strand break intermediates from poly(ADP-ribose) polymerase, as this protein binds avidly to strand breaks; there is recent evidence that the cell uses this binding to activate checkpoints that may decide between cell cycle arrest and apoptosis (52). Therefore, it makes biological sense to have the BER reactions as concerted as possible, so that the mechanism of repair does not interfere with the surveillance mechanisms that monitor the genome for DNA damage.
In summary, if the BER enzymes noted here were in an equimolar macromolecular complex, the cumulative molecular mass is 196 kDa (i.e. UDG, 32 kDa; ␤-pol/dRpase, 39 kDa; and DNA ligase I, 125 kDa). This is consistent with the molecular mass of the BER-proficient complex detected here by sucrose gradient centrifugation (ϳ180 kDa), because DNA ligase I sediments as a Ͻ125-kDa molecule (41). A significant portion of the ϳ180-kDa complex obtained by immunoaffinity chromatography appeared to dissociate in the elution and subsequent sucrose gradient centrifugation steps, pointing to the idea that proteins in this BER complex are only weakly associated with each other. The significance of such a BER complex in the cell may be that the lifetime of each DNA intermediate during the sequential BER pathway is reduced, thus enhancing the efficiency of the overall pathway.