3-Methyladenine-DNA glycosylase (MPG protein) interacts with human RAD23 proteins.

Human 3-methyladenine-DNA glycosylase (MPG protein) initiates base excision repair by severing the glycosylic bond of numerous damaged bases. In comparison, homologues of the Rad23 proteins (hHR23) and the hXPC protein are involved in the recognition of damaged bases in global genome repair, a subset of nucleotide excision repair. In this report, we show that the hHR23A and -B also interact with the MPG protein and can serve as accessory proteins for DNA damage recognition in base excision repair. Furthermore, the MPG.hHR23 protein complex elevates the rate of MPG protein-catalyzed excision from hypoxanthine-containing substrates. This increased excision rate is correlated with a greater binding affinity of the MPG protein-hHR23 protein complex for damaged DNA. These data suggest that the hHR23 proteins function as universal DNA damage recognition accessory proteins in both of these major excision repair pathways.

Human 3-methyladenine-DNA glycosylase (MPG protein) initiates base excision repair by severing the glycosylic bond of numerous damaged bases. In comparison, homologues of the Rad23 proteins (hHR23) and the hXPC protein are involved in the recognition of damaged bases in global genome repair, a subset of nucleotide excision repair. In this report, we show that the hHR23A and -B also interact with the MPG protein and can serve as accessory proteins for DNA damage recognition in base excision repair. Furthermore, the MPG⅐hHR23 protein complex elevates the rate of MPG protein-catalyzed excision from hypoxanthine-containing substrates. This increased excision rate is correlated with a greater binding affinity of the MPG protein-hHR23 protein complex for damaged DNA. These data suggest that the hHR23 proteins function as universal DNA damage recognition accessory proteins in both of these major excision repair pathways.
Excision repair of DNA, which protects cells against numerous insults inflicted by both exogenous and endogenous agents, is subdivided into the NER 1 and BER pathways. Extensive investigation of the NER pathway has revealed that more than 30 proteins are involved (1)(2)(3)(4). To effect repair both in vivo and in vitro, NER requires that these proteins function as complexes. In contrast to NER, BER requires fewer proteins to complete repair, and the BER enzymes can accomplish in vitro repair reactions in the absence of protein-protein interactions (5)(6)(7).
Despite the fact that BER enzymes generally do not require complex formation to catalyze in vitro reactions, such interplay can facilitate in vivo DNA repair. One way to expedite repair is to increase the velocity of a rate-limiting reaction in the pathway. An example of this for BER is the acceleration of the rate of deoxyribose phosphate removal by DNA polymerase ␤ in the presence of human AP-endonuclease (HAP1 of APE1) (8). Another example is XPG protein functioning in the BER pathway (9 -11). XPG protein increases the reaction rate from 2-7-fold for in vitro excision of Tg by the hNTH protein (12,13). However, complex formation does not necessarily increase reaction rates. Examples of complex formation without activation of BER enzymatic activity in vitro include X-ray cross-complementing 1-poly(ADP-ribose) polymerase-DNA ligase III or uracil-DNA glycosylase-RPA protein (6, 12, 14 -19).
The coupling of repair to transcription is now well documented for NER and BER (9, 12, 13, 20 -24). In both pathways this coupling of repair to transcription occurs via complex formation. Nonetheless, not all adducts repaired by NER and BER are repaired using transcription-coupling. In NER, 4-nitroquinoline 1-oxide and aminofluorene adducts are not repaired by a transcription coupled mechanism (25,26). Similarly for BER, although transcription-coupled BER is observed for the repair of certain adducts, including Tg, not all BER occurs via such a mechanism. Repair of N-methylpurines is not coupled to transcription, suggesting that the role of the XPG protein in BER is not universal (27)(28)(29)(30)(31).
Global genome repair (GGR) is a subset of the NER pathway responsible for the removal of lesions that are not repaired by transcription coupling (32)(33)(34). One of the first steps in global genome repair is the recognition of damage by the XPC⅐hHR23B complex (35). Two human homologues of the Saccharomyces cerevisiae RAD23 protein, hHR23A and B (36 -38), stimulate the rate of in vitro NER in concert with XPC protein (20,(37)(38)(39)(40)(41). To date, however, the hHR23 proteins have not been implicated in other repair processes (42).
hHR23 proteins are also targets for ubiquitination by human papilloma virus E6-associated protein (43). Exposure of U2OS cells to ultraviolet light prevented the degradation of hHR23A, showing that ubiquination could be prevented in a situation that required hHR23 proteins to function in a repair complex. Previously the yeast homologue of the hHR23 proteins, Rad23, was linked to DNA repair and ubiquitination (44). In a recent study, the hHR23 proteins were found to interact with the S5a subunit of the 26 S proteasome (45). Thus, the hHR23 proteins are associated with ubiquitin metabolic pathways in addition to NER.
The MPG protein is a DNA glycosylase excising numerous  1 The abbreviations used are: NER, nucleotide excision repair; BER, base excision repair; DDRAP, DNA damage recognition accessory protein; MPG, methylpurine-DNA glycosylase; Hx, hypoxanthine; Hx-oligodeoxyribonucleotide, a double-stranded oligodeoxyribonucleotide with a hypoxanthine at a unique position; hNTH, human homologue of the E. coli Nth protein, endonuclease III; hHR23, human homologue of the S. cerevisiae RAD23; Tg, thymine glycol; XPC, xeroderma pigmentosum complementation group C; XPG, xeroderma pigmentosum complementation group G; BSA, bovine serum albumin; PCR, polymerase chain reaction; bp, base pair; Ade, adenine; AAG, alkyladenine-DNA glycosylase; NMPG, N-methylpurine-DNA glycosylase; ANPG, alkyl-N-purine-DNA glycosylase; 7-meG, 7-methylguanine; AP, abasic site specific. modified DNA bases, including N-methylpurines (46 -50). Recently, the three-dimensional structure of DNA-MPG protein interaction was solved (51), showing that MPG protein binds to a DNA adduct without additional protein requirements. In addition to crystallographic data, in vitro and in vivo kinetic results showed that 7-meG excision by the MPG protein is not altered by complex formation (31). Therefore, any complex formation of the MPG protein will not affect the relative rates of sequence context-dependent excision of 7-meG. Nonetheless, complex formation may have a role in accelerating the rate of the in vivo reaction.
In this report, we demonstrate that the hHR23 proteins specifically interact with the MPG protein as DNA damage recognition accessory proteins (DDRAP). We also identify the binding domains of the MPG protein and hHR23 proteins critical for this interaction. Moreover, we show that the interaction of the MPG protein with hHR23 proteins stimulates the DNA glycosylase activity and is correlated to increased binding affinity of the MPG⅐hHR23 protein complex for the substrate. These data establish a link between the NER and BER pathways subject to regulation by the hHR23 proteins.

MATERIALS AND METHODS
Plasmids, Chemicals, Strains, Oligonucleotides, Enzymes-The pANPG40 plasmid hosting the truncated MPG cDNA, pANPG70, phHR23A, phHR23B, pACT-1, pAS2-1, pAS-1-SV40Tag, and pACT1-p53 were from CLONTECH or from laboratory stocks (42,52). The yeast strain, PJ69 -4A, used in the two-hybrid screen, was grown in rich medium, YPD (1% yeast extract, 2% peptone, 2% dextrose) (53). Transformed strains were grown in completed minimal drop-out medium with the appropriate supplements. Yeast strains were grown at 30°C, and all manipulations involved standard yeast methods. Standard molecular biology methods followed protocols from Ausubel et al. (55) or Sambrook et al. (54). The truncated MPG protein, hHR23A protein, and hHR23B protein were obtained from laboratory stocks. Restriction enzymes and molecular biology grade BSA were obtained from Roche Molecular Biochemicals. The DNA sequencing and oligodeoxyribonucleotides were provided by Comprehensive Cancer Center Core facilities of the City of Hope National Medical Center. Oligodeoxyribonucleotides were gel-purified, phosphorylated with T4 polynucleotide kinase using [␥-32 P]ATP, and annealed with a 2-fold excess of the unlabeled complementary strand.
Yeast Interaction Trap Screening-To identify proteins that interact with MPG protein, the complete coding sequence of MPG was amplified by PCR and ligated into the EcoRI and BamHI sites of yeast Gal4DB plasmid pAS2-1 (CLONTECH) to create the plasmid pAS2-1-MPG. The sequence of the forward or sense oligodeoxyribonucleotide used for PCR amplification was 5Ј-CCG GAA TTC CCC GCG CGC AGC GG-3Ј. The sequence of the reverse or antisense primer used for PCR amplification was 5Ј-GCG GAT CCT CAG GCC TGT GTG TCC-3Ј. The sequence of the MPG insert was verified to ensure that the Gal4DB-MPG fusion was in the correct reading frame. The yeast strain PJ69 -4A was co-transfected with pAS2-1-MPG and a human keratinocyte Matchmaker cDNA library in the pACT-1 plasmid (CLONTECH). Transformants were plated on synthetic drop-out medium lacking adenine, histidine, leucine, and tryptophan and incubated at 30°C for 3-7 days.
Tranfection of the pAS-1-MPG plasmid alone did not yield any Ade ϩ His ϩ Leu ϩ Trp ϩ -positive clones. Ade ϩ His ϩ Leu ϩ Trp ϩ colonies were assayed for ␤-galactosidase activity. Briefly, Ade ϩ His ϩ Leu ϩ Trp ϩ colonies were lifted onto nitrocellulose membranes, frozen briefly (20 s) in liquid nitrogen, thawed at room temperature, and overlaid onto Whatman No. 3MM filters saturated with ␤-galactosidase buffer in Petri dishes. Filters were incubated at 30°C for 4 -6 h until a dark blue color indicating the transformation of the X-gal (5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside) substrate was observed in the control pAS-1-SV40Tag and pACT1-p53 plasmids. Plasmid DNA was recovered from positive yeast colonies by electroporation into Escherichia coli DH5␣, followed by isolation using standard miniprep procedures. Sequences of the cDNA inserts were determined by using a primer derived from Gal4AD (5Ј-ATACCACTACAATGGATGA-3Ј).
His 6 -hHR23 Protein Affinity Chromatography-100 ng of His 6 -hHR23A or His 6 -hHR23B was mixed with 50 ng of MPG protein and 20 l of Ni 2ϩ -NTA resin (Qiagen) in 10 mM imidazole, 300 mM NaCl, 50 mM NaH 2 PO 4 (Buffer A) at 4°C for 60 min. After extensive washes with the same buffer, the proteins retained on the beads were eluted using 200 l of the Buffer A containing 250 mM imidazole. The MPG protein in the eluates was analyzed by Western blot analysis using a polyclonal anti-MPG antibody.
MPG Protein Deletion Mutants-Deletion mutants of MPG cDNA were obtained by ligating the PCR-amplified MPG protein fragments from the pANPG70 sequence into the EcoRI and BamHI sites of the pAS2-1 vector. The oligodeoxyribonucleotides used for PCR to construct N-terminal and C-terminal deletion mutants of the MPG protein are indicated in Table I. DNA Glycosylase Activity Assays-The double-stranded 39-bp Hxoligodeoxyribonucleotide 5Ј end-labeled with 32 P-GGTTCAGGGTCAG-GCHxGGCAATGGCTGCGAGAAACCAGC was incubated with the MPG protein and various amounts of hHR23A or hHR23B to perform activity assays as described previously (56). The complementary strand had a T opposite the Hx base. Briefly, the reaction buffer was 100 mM KCl, 70 mM HEPES-KOH, pH 7.5, 0.5 mM EDTA, 5 mM ␤-mercaptoethanol, and 5% glycerol in a total volume of 20 l, and the concentration of the oligodeoxyribonucleotide varied from 0 to 40 nM. The reactions were incubated at 37°C for 2 min after adding the diluted MPG protein (230 pM final concentration) and stopped by the addition of 20 l of 2 M piperidine. The stopped reaction mix was incubated at 37°C for 15 min, and the oligodeoxyribonucleotide was precipitated after the addition of 3 M sodium acetate (final concentration 0.3 M) and the addition of cold ethanol. After washing with 80% ethanol and drying, the sample was resuspended in 10 l of formamide loading buffer and separated by electrophoresis on a 20% denaturing polyacrylamide (7 M urea) minigel to resolve reaction products. The dried gels were subjected to Phospho-rImager (Molecular Dynamics) analysis for quantification of band intensities.
A second method for determination of DNA glycosylase activity was performed using a DNA substrate with Ade and guanine bases methylated using [ 3 H]dimethyl sulfate. Reaction conditions were used such that greater than 90% of the radioactivity released was 3-methyladenine (less than 10% of the total radioactivity was released during the assay). The frequency of modification was approximately 1 per 100 bases. Production of the substrate has been described previously (57). The specific activity of the substrate was 550 cpm/pmol of labeled base. The reaction mixture contained (25 l) 9000 cpm of [ 3 H]dimethyl sulfate-treated calf thymus DNA in the same buffer as used for the oligodeoxyribonucleotide assay. The reactions for determination of the CGGGAATTCATGCCCGCGCGCAGC GCGGATCCTTAGGTAAGGTGGCCCTT effect of the hHR23 proteins on the MPG protein activity used 2 pmol of MPG protein. For the assays, 4 pmol of BSA or 4 pmol of one of the hHR23 proteins were also added. The reaction mixtures were incubated at 37°C for from 1-4 min and stopped using a cold DNA-BSA mixture (58). The DNA was precipitated by the addition of 150 l of ethanol (Ϫ20°C). The sample was centrifuged in an Eppendorf centrifuge for 10 min. A 150-l aliquot of the ethanol supernatant was removed and subjected to liquid scintillation spectrometry for quantification. DNA Binding Assays-The 39-bp oligodeoxyribonucleotide with Hx (40 fmol) was incubated with the MPG protein (50 nM) in a total volume of 6 l in a buffer containing 70 mM Hepes-KOH, pH 7.5, 0.5 mM EDTA, 5 mM ␤-mercaptoethanol, 5% glycerol. No KCl was added to the binding buffer, since higher salt concentrations catalyze the reaction of the MPG protein and removal of Hx from the DNA. The mixture was incubated at 0°C for 10 min, 3 l of agarose gel loading buffer (3ϫ loading buffer: 10% glycerol, 0.5% xylene cyanol) was added, and the mixture was loaded onto a 10% non-denaturing polyacrylamide gel (acrylamide:bisacrylamide, 20:0.25) using Tris-Borate-EDTA buffer. The products were separated by electrophoresis at 100 V for 4 -5 h at 4°C. After electrophoresis, gels were dried, quantified using PhosphorImager analysis, and autoradiographed. Analysis of each point was performed in triplicate, and the error bars in the figures indicate the S.D. (56).
Supershift Assay-DNA binding assays were performed in 50 l of total volume at 4°C for 30 min in a mixture containing 4% Ficoll, 20 mM Tris-HCl, pH 7.5, 13 mM Hepes-KOH, pH 7.9, 1 mM EDTA, 10 mM MgCl 2 , and 0.5 mM dithiothreitol. The binding assay mixtures also contained MPG protein (40 nM) and hHR23 protein (200 nM) or MPG protein (40 nM) and BSA (200 nM). The mixtures were incubated in the presence of a 10-fold excess of specific unlabeled competitor 39-bp Hx-oligodeoxyribonucleotide (1 g). The 32 P-labeled, 39-bp Hx-oligodeoxyribonucleotide (100 ng) was then added and incubation continued for an additional 30 min on ice. DNA-protein complexes were resolved by electrophoresis through low ionic strength 4% polyacrylamide gels at 4°C in TAE buffer (6.7 mM Tris-HCl, pH 7.9, 1 mM EDTA, 3.3 mM sodium acetate). The gels were dried and imaged with a PhosphorImager (Molecular Dynamics).
Western Blotting Analysis and Antibody-Anti-MPG protein antibody was prepared in the animal facility of the City of Hope using a standard protocol, and the cleared serum was used as a polyclonal antibody reagent (54). Proteins transferred to polyvinylidene difluoride membrane were incubated with a primary rabbit anti-MPG protein antibody diluted 1:1000 and a secondary horseradish peroxidase-conjugated anti-rabbit immunoglobulin (Amersham Pharmacia Biotech). Antibody binding was revealed using ECL Western blotting detection reagents (Amersham Pharmacia Biotech).

RESULTS
hHR23 Proteins Interact with MPG Protein-The full-length MPG protein was used as a bait protein fused to the DNA binding domain of the GAL4 transcription factor. As a control, the pAS2-1-MPG construct was transfected into PJ69 -4A to demonstrate that the construct did not activate transcription from the GAL promoters. A human keratinocyte library fused to the pACT1 plasmid with the GAL4 transcriptional activation domain (1 ϫ 10 6 clones) was co-transfected with the pAS-1-MPG construct. An initial screen for selective growth on Ade Ϫ His Ϫ selection media yielded approximately 100 positive colonies. Clones identified as prey and the pAS2-1-MPG were screened for their ability to produce ␤-galactosidase. The positive colonies were transformed into bacteria, and the plasmid DNA obtained was analyzed using restriction enzyme digestion. Unique clones positive for all three screens were sequenced and subjected to a search for Expressed Sequence Tag sequences in GenBank. One of the positive sequences obtained in 12 clones coded for the C-terminal end of the hHR23A protein. Since the hHR23A and B sequences share regions of up to 93% identity (37), we reasoned that both proteins could interact with the MPG protein. The complete coding sequences for the hHR23A and hHR23B proteins placed into the pACT-1 vector demonstrated MPG protein-hHR23A and MPG protein-hHR23B protein interactions in both the Ade Ϫ His Ϫ selection and ␤-galactosidase production. Both pACT-1-hHR23A and pACT-1-hHR23B required co-transfection of the pAS2-1-MPG plasmid to activate transcription. Therefore, yeast interaction trap results demonstrate an interaction between the MPG protein and both hHR23 proteins.
Affinity Chromatography Using Tethered hHR23 Proteins-Although the yeast interaction trap assay alone is insufficient to prove that the MPG protein interacts with the hHR23 proteins, affinity chromatography provides such proof. The hHR23A and -B proteins were labeled with a His 6 tag sequence that permits the attachment of the hHR23 proteins to a Ni 2ϩ -NTA resin. In control experiments, the hHR23 proteins without a His 6 tag do not bind to the Ni 2ϩ -NTA resin. The His 6 -hHR23A or -B proteins were incubated with MPG protein in the presence of the Ni 2ϩ -NTA resin. After thorough washing, no proteins were released from the matrix of the column (Fig.  1). Bound proteins were eluted using high imidazole concentrations and analyzed using Western blotting with an anti-MPG protein antibody. Fig. 1 shows that the MPG protein is released specifically from the Ni 2ϩ -NTA resin and is specifically associated with the hHR23 proteins.
N-and C-terminal Regions of the MPG Protein Can Interact with the hHR23 Proteins-To elucidate regions of the MPG protein that are crucial to maintaining the hHR23 protein interaction, a series of deletions of the MPG protein was constructed at either the N or C terminus of the MPG protein in the pAS2-1 vector. These MPG protein deletion mutants were co-transfected into the PJ69A strain with the pACT1 GAL4 activation domain fused to either hHR23A or hHR23B. Cotransfectants that grew on Ade Ϫ His Ϫ selection media were further tested for ␤-galactosidase production at least 3-fold higher than background levels. Fig. 2a shows that the region of interaction of the MPG protein with hHR23A or B encompasses a 40-amino acid portion of the N terminus, as indicated by a black region. Deletion of a region at the C terminus of the MPG protein, however, also abrogates the interaction of the MPG protein with the hHR23 proteins (Fig. 2a). Thus, regions at both ends of the MPG protein are potentially critical for the maintenance of the hHR23 protein interactions.
The C-terminal Region of hHR23 Proteins Interacts with the MPG Protein-Initial screening of the yeast interaction trap human keratinocyte library identified a GAL4 activation domain-truncated hHR23A fusion protein with only 68 amino acids from the C-terminal end of the hHR23A protein. This region is distinct from the region attributed to the binding of the XPC protein. Similar regions of more than 90% identity exist in both the hHR23A and B proteins (Fig. 2b). Fig. 2c   FIG. 1. Affinity chromatography demonstrating the interaction of MPG protein and hHR23 proteins. His-tagged hHR23A or -B was incubated with the MPG protein in the presence of Ni 2ϩ -NTA beads as described under "Materials and Methods." Rinsing with 1 ml of loading buffer revealed that no MPG protein was released followed by 100 l of Buffer A (lanes marked Wash). Elution was performed using 200 l of Buffer A plus 250 mM imidazole (lanes marked Elution). The MPG protein was detected by Western blotting. Lanes (left to right), MPG-MPG protein positive control, MPG hHR23A-MPG protein hHR23A negative control, MPG His-hHR23A-MPG protein hHR23A protein bound to Ni 2ϩ -NTA beads, and MPG His-hHR23B-MPG protein hHR23B protein bound to Ni 2ϩ -NTA beads.
shows the regions implicated in the interaction of the MPG and hHR23 proteins. Construction of the full-length hHR23A or -B in fusion with the transcriptional activation domain of GAL4 does not significantly increase the ␤-galactosidase activity, suggesting that other regions of the hHR23 proteins are not essential for the preservation of the interaction. Therefore, the C-terminal ends of the hHR23 proteins are the major binding sites for the MPG protein.
The hHR23A/-B⅐MPG Protein Complex Has an Increased Activity for Release of Hx from DNA Compared with the MPG Protein Alone-Interaction of a protein with an enzyme does not necessarily alter the enzymatic activity. To examine the effect of hHR23 proteins on MPG protein excision, we assayed the activity of MPG protein on a 39-bp Hx-oligodeoxyribonucle-otide as a function of the hHR23 proteins. Fig. 3 shows the MPG protein-catalyzed excision of Hx from the substrate as a function of the hHR23 proteins or a BSA control. In the absence of hHR23A or B proteins, the MPG excises only a small fraction of Hx from the substrate. The excision of Hx leaves an abasic site that is revealed by severing the phosphodiester backbone of the oligodeoxyribonucleotide using a mild piperidine treatment. Incubation of the reaction mix with increasing amounts of hHR23 proteins increases excision of Hx up to approximately 1.5-fold that of MPG protein excision alone. In a control experiment, BSA only slightly increases the excision rate of MPG protein, suggesting that the increases observed with hHR23A or -B are not the result of nonspecific interactions. This increase in the activity removing Hx by the MPG⅐hHR23 complex is reproducible. A [ 3 H]dimethyl sulfate-treated calf thymus DNA substrate was also assayed for the ability of the MPG protein to stimulate excision of 3-methyladenine. Table II shows that there is a slight stimulation in the excision rate of 3-methyladenine noted for the hHR23B protein, but excision by the hHR23A protein is almost identical to that of the BSA control. Thus, depending on the substrate, hHR23A or -B modestly stimulates the DNA glycosylase activity of MPG protein.
hHR23 Proteins Participate in a Complex with MPG Protein Binding to an Hx-oligodeoxyribonucleotide Substrate-Increased MPG protein DNA glycosylase activity could indicate the indirect involvement of the hHR23 proteins. Alternatively, the hHR23 proteins could increase activity by the formation of a complex that binds to DNA. We obtained evidence for the FIG. 2. Interaction of MPG protein deletion mutants with hHR23A or B proteins in the yeast interaction trap assay and binding domains. a, deletion mutants of the MPG protein assayed using the yeast interaction trap assay. The full-length MPG protein is 293 amino acids in length. The length of each of the deletion mutants is indicated on the left-hand side of the figure. Deletion mutants were selected on Ade-and His-deficient media and then subjected to ␤-galactosidase liquid assays. ϩ indicates that the assay was positive, Ϫ indicates that the assay was negative, and Ϯ indicates that there was a weak interaction, approximately 2-fold above background. The arrow indicates the position of the truncated MPG protein used in the biochemical assays. Numbers 1-4 indicate regions as defined from the three-dimensional structure of the MPG protein complexed with a DNA substrate: 1-␣A␤B, 2-␤1␤2, 3-␣C␤3␤4, 4-␤5␣D␣E␤6␤7␣F␤8␣G (51). The position of the active-site Glu is Glu120. The black regions denote regions critical for interaction of the MPG and hHR23 proteins. The length of the MPG deletion mutants in amino acids (aa) is on the left-hand side of the figure. b, hHR23 proteins binding domains. Structural regions of the hHR23 proteins adapted from Masutani et al. (39). There are four regions with high identity that are conserved between the two proteins. The region with the highest identity is the 93% conserved C-terminal region of the MPG protein binding site. The other regions are ubiquitin-ubiquitin-like region, SPTA-region rich in the amino acids Ser, Pro, Thr, and Ala, UBA-ubiquitin-associated domain, and XPC-XPC protein binding domain. c, amino acid sequences of the binding domains of the MPG protein and hHR23 proteins. The bold sequences of the hHR23 proteins represent sequences that are in the 93% identity regions. The underlined sequence in the MPG protein is part of a direct repeat at the protein sequence level. formation of a complex involving hHR23 proteins using EMSA. Fig. 4 demonstrates that the MPG protein, as previously reported, binds to the Hx-oligodeoxyribonucleotide, resulting in decreased electrophoretic mobility. On the other hand, hHR23A or -B proteins alone do not alter the electrophoretic mobility of the substrate (not shown). If the Hxoligodeoxyribonucleotide is incubated in the presence of both the MPG protein and one of the hHR23 proteins, however, the electrophoretic mobility is further reduced, indicating a supershift. Such an EMSA supershift is evidence that the hHR23 proteins form a complex between the MPG and the Hx-oligodeoxyribonucleotide.
hHR23 Proteins Increase the Binding Affinity of MPG Protein for an Hx-containing Substrate-To determine if the hHR23 proteins changed the binding affinity of the MPG protein for an Hx substrate, we used the same substrate as in the previous section but limited the reaction to binding. Fig. 5 shows binding of hHR23 proteins to the MPG protein as a function of hHR23 concentration. Less than 25% of the MPG protein bound to the Hx-oligodeoxyribonucleotide substrate without the hHR23 proteins. The addition of the hHR23A or -B proteins to the MPG protein increases binding of the MPG protein up to 3-fold to the Hx-oligodeoxyribonucleotide. Moreover, in a control experiment, the nonspecific interaction of the MPG protein with BSA increased minimally the bound fraction of the oligodeoxyribonucleotide. No binding was observed to undamaged DNA (data not shown). Therefore, the stimulation of the DNA glycosylase activity is dependent on increased MPG protein-DNA interactions mediated by either of the hHR23 proteins. DISCUSSION Until recently, general belief was that excision of damaged bases by BER glycosylases was independent of other proteins and that protein-protein interactions were more significant only after the initial removal of a damaged base during BER. We have now shown that two DDRAPs, hHR23A or -B, can increase the rate of excision of a damaged base by a DNA glycosylase.
In BER, the rate-limiting step for repair is variable (31, 59 -61) and depends on the adduct type. Previous investigations show that the rate-limiting step for BER could occur at the DNA glycosylase, AP-endonuclease, or DNA polymerase ␤/AP lyase steps. hNTH protein-catalyzed BER of Tg is linked to transcription and depends on hXPG, since the two proteins interact and Tg excision in vitro is rate-limiting for BER (9 -13). Excision of damaged bases by the MPG protein, however, is not coupled to transcription and is linked to hHR23 proteins (27,28,31). Therefore, different protein-protein interactions can determine the rate of DNA glycosylase-mediated excision. Since DNA glycosylases were previously envisaged to act independently, their substrate ranges and kinetic parameters were defined in vitro using this assumption. The discovery of different DDRAPs requires that these substrate ranges and kinetic parameters be re-examined in light of protein-protein interactions.
hHR23 proteins increase the affinity of DNA repair proteins for their substrates. An example of a complex with a similar role is the XPE complex of DDB1 and DDB2 (62,63). The XPE complex presumably increases the rate of 6-4 photoproduct recognition by the XPC⅐hHR23B complex (33). The uniqueness of the hHR23 protein interaction with the MPG protein or the   XPC protein is that these represent early steps in both BER and NER. Therefore, it is possible that the hHR23 protein interactions serve as a gate between BER and NER pathways. Fig. 7 is a scheme representing one-way hHR23 proteins could control excision repair. Adducts can be repaired by both BER and NER, and the recognition step could direct the pathway. If the MPG protein-hHR23 complex were more efficient at recognition, repair would proceed via BER, whereas efficient recognition by XPC-hHR23 would shift the equilibrium to NER. Therefore, the hHR23 proteins can connect both pathways. The model presented in Fig. 6 suggests several hypotheses to test.
In cells that are deficient in XPC protein, competition between BER and NER should be reduced, and the MPG protein-mediated repair of dimethyl sulfate damage should increase. This could be manifested as either an increased repair rate or an increased survival rate. On the other hand, if MPG is overexpressed and hHR23 proteins are bound, XPC-mediated repair of ultraviolet radiation should be decreased if there is a competition between XPC and MPG proteins for hHR23 binding. This could decrease survival, increase mutation rate, or reduce the repair rate of cyclobutane pyrimidine dimers. Other questions also arise in light of these results. These include the generalization of DDRAPs for reactions of other DNA glycosylases. There is the possibility that DNA glycosylases and DNA glycosylase/AP lyases use different accessory proteins to effect repair. The hNTH protein, a DNA glycosylase/AP lyase is activated by the XPG protein, and MPG is activated by hHR23 proteins (12,13). It is possible that there are many accessory proteins or that only a few of these proteins function in damage recognition for excision repair. It is also necessary to examine the rate of excision of different lesions by these complexes. The hNTH protein excision of Tg is enhanced in the presence of the XPG, and the MPG protein excision of Hx is enhanced in the presence of the hHR23 proteins. The role of the accessory proteins in stimulating the excision of other substrates, for example dihydrothymine for hNTH protein or 3-methyladenine for MPG protein, however, is unknown. Future experiments will also include the investigation of substrates in longer DNA fragments to determine if there is a greater stimulation of the MPG protein activity compared with that observed in oligodeoxyribonucleotides. The results presented in this study provide a basis for pursuing such studies and provide a connection between the two separate branches of excision repair.