Interaction of estrogen receptor alpha with 3-methyladenine DNA glycosylase modulates transcription and DNA repair.

Estrogen receptor alpha (ERalpha) interacts with basal transcription factors, coregulatory proteins, and chromatin modifiers to initiate transcription of the target genes. We have identified a novel interaction between ERalpha and the DNA repair protein 3-methyladenine DNA glycosylase (MPG) thereby providing a functional link between gene expression and DNA repair. Interestingly, the ERalpha-MPG interaction was enhanced by the presence of estrogen response element (ERE)-containing DNA. In vitro pull-down assays indicated that the interaction of ERalpha with MPG was direct and occurred through the DNA- and ligand-binding domains and the hinge region of the receptor. More importantly, endogenously expressed ERalpha and MPG from MCF-7 cells coimmunoprecipitated with ERalpha- and MPG-specific antibodies. The ERalpha-MPG interaction had functional consequences on the activities of both proteins. ERalpha increased MPG acetylation, stabilized the binding of MPG with hypoxanthine-containing oligos, and enhanced MPG-catalyzed removal of hypoxanthine from DNA. In turn, MPG dramatically stabilized the interaction of ERalpha with ERE-containing oligos, decreased p300-mediated acetylation of the receptor, and reduced transcription of simple and complex ERE-containing reporter plasmids in a dose-dependent manner. Our studies suggest that recruitment of MPG to ERE-containing genes influences transcription and plays a role in maintaining integrity of the genome by recruiting DNA repair proteins to actively transcribing DNA.

Estrogen is critical for the growth, development, and homeostasis of neural, skeletal, cardiovascular, and reproductive tissues (1,2). The actions of estrogen are mediated by two members of the nuclear receptor family, estrogen receptors (ERs) 1 ␣ and ␤. ER␣ is, however, generally a more potent transcriptional activator than ER␤ (3)(4)(5). Like other nuclear receptor family members, ER␣ has a modular structure. At the amino terminus is the A/B region with its autonomous activa-tion function 1 (6). Region C encompasses the DNA binding domain and is linked by the hinge domain to region E, which contains the ligand-binding domain (LBD, Ref. 7). The DNAbinding domain has two zinc finger motifs that are involved in DNA binding. The LBD contains a hydrophobic pocket that interacts with estrogens and antiestrogens. The LBD also contains a ligand-dependent activation function 2, which is responsible for interaction of the receptor with coregulatory proteins (8 -10).
ER␣ binds to estrogen response elements (EREs) in target genes to initiate changes in transcription. The consensus ERE is comprised of the palindromic sequence GGTCAnnnTGACC and is found in the Xenopus laevis vitellogenin A2 gene (11). In addition to interacting with EREs, ER␣ modulates transcription through its interaction with components of the basal transcription machinery, regulatory proteins, and chromatin modifiers (12). ER␣ interacts with proteins in the basal transcription complex including TATA-binding protein (13). The coregulatory proteins steroid receptor coactivator 1, transcription intermediary factor 2, and amplified in breast cancer 1, interact with the receptor in a ligand-dependent manner and enhance estrogen responsiveness (reviewed in Refs. 12 and 14 and references therein). ER␣ also interacts with chromatin modifying proteins including proteins with histone acetylase activity such as the cyclic AMP response element-binding protein (CBP/p300, Refs. [15][16][17][18] and BRG1, the human homolog of the SNF2␤ subunit of the SWI2/SNF2 complex (19,20).
Regions of the genome involved in active transcription are more susceptible to DNA-damaging agents, which are present in the environment and as by products of cell metabolism (21,22). DNA repair systems such as base excision repair and nucleotide excision repair are involved in repair of these DNA lesions (21,23). If damaged DNA is not repaired, it can lead to mutations, chromosomal aberrations, aging, or cancer (24).
In an effort to identify novel proteins that interact with the DNA-bound ER␣, we have isolated and identified the DNA repair protein 3-methyladenine DNA glycosylase (MPG). ER␣ stimulates binding of MPG to modified DNA and increases the catalytic removal of modified bases by MPG. In turn, MPG enhances the binding of ER␣ to the ERE and modulates estrogen-mediated transcription. Our studies provide evidence of a physical and functional link between these proteins involved in transcription and DNA repair.

EXPERIMENTAL PROCEDURES
Pull-down Assay-Pull-down assays were carried out by adsorbing 200 pmol of annealed biotinylated oligos containing the A2 ERE or a nonspecific DNA sequence to streptavidin-agarose beads as described (25). The immobilized DNA was incubated with 400 pmol of baculovirus-expressed, purified ER␣ in the presence of 1 M 17␤-estradiol (E 2 ) for 10 min followed by addition of 500 g of HeLa nuclear extract prepared as described (25). Immobilized DNA was also incubated with HeLa nuclear extracts in the absence of ER␣ as a control for nonspecific binding. After 2 h at 4°C, the beads were washed as described (25) and the bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE, and silver stained. A 35-kDa band, which was observed in the presence, but not in the absence of ER␣, was excised and analyzed by micro-liquid chromatography-electron spray ionization tandem mass spectrometry (MS) as previously reported (26). MS/MS analysis identified three peptides, IVETEAYLGPEDEAAHSR, GPLEPSEPAVVA-AAR, and AFLGQVLVR with amino acid sequences identical to that found in 3-methyladenine DNA glycosylase. These three peptides comprised 14.3% of the total MPG protein.
Cell Cultures and Transfections-U2-osteosarcoma (U2-OS) cells were transfected as described (27) (10, 50, or 100 ng of pCDNA3 MPG with ConsERE (ϩ10)-CAT, 10 or 40 ng of pCDNA3 MPG with pS2ERE-CAT, and 1, 10, or 100 ng of pCDNA3 MPG with pS2-CAT). ␤-Galactosidase activity was measured (30) and used to normalize for differences in transfection efficiency. CAT assays were performed as described in Ref. 5, quantitated using a Amersham Biosciences PhosphorImager, and analyzed with ImageQuant 5.0 software. At least three independent experiments were carried out in duplicate.
Immunoprecipitation Assays-ER␣-associated proteins were immunoprecipitated from MCF-7 nuclear extracts, which had been prepared from cells that were exposed to ethanol or 10 nM E 2 for 15 min. The ER␣-specific antibody sc-8005 (Santa Cruz Biologicals, Santa Cruz, CA) was cross-linked to beads using the Seize antibody immobilization kit as described by the manufacturer (Pierce) and then incubated with MCF-7 nuclear extracts in IP buffer (10 mM Tris, 0.5 mM EDTA, 1 mM dithiothreitol, 1 M E 2 ) with 100 mM NaCl and protease inhibitors (50 g/ml leupeptin, 5 g/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 5 g/ml aprotinin). After incubation at 4°C for 2 h, the beads were washed and bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE, and transferred to a nitrocellulose membrane. Western analysis was carried out using the ER␣-specific antibody sc-543 (Santa Cruz Biotechnologies) or an MPG-specific antibody (kindly provided by Timothy O'Connor, Beckman Research Institute, City of Hope, CA). The ER␣-MPG interaction was also examined by immunoprecipitation of MPG-associated proteins using the MPGspecific antibody. The MPG antibody was adsorbed to Protein A beads and incubated with MCF-7 nuclear extracts prepared from cells that had been exposed to ethanol or 10 nM E 2 for 15 min. Immunoprecipitation and Western analysis of proteins was carried out as described above except that the MPG-specific and ER␣-specific antibody sc-8002 (Santa Cruz Biotechnologies) was used for detection.
Interaction of ER␣ with Immobilized MPG-Bacterially expressed GST-MPG was immobilized on glutathione-Sepharose (Amersham Biosciences) and washed three times with TEG buffer (20 mM Tris, 0.5 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 0.1% Nonidet P-40) containing 500 mM NaCl and protease inhibitors. The beads were then washed and resuspended in TEG buffer containing 200 mM NaCl. Full-length and truncated receptors (ER␣, AB, CD, E, EF, and 1-530), which had been transcribed and translated in vitro in the presence of [ 35 S]methionine using the TNT T7 Quick-coupled transcription translation system (Promega, Madison, WI) as described (26) were then combined with immobilized GST-MPG. The 35 S-labeled proteins were incubated with immobilized MPG for 1 h at 4°C in the absence or presence of 1 M E 2 and the unbound proteins were removed by four washes with TEG buffer containing 200 mM NaCl. The bound 35 S-labeled proteins were eluted with SDS sample buffer, separated by SDS-PAGE, and subjected to autoradiography. For interaction studies with nuclear receptors, progesterone receptor B (PRB), thyroid receptor ␤, retinoid X receptor ␣ (RXR␣), and peroxisome proliferator-activated receptor ␥ (PPAR␥) were expressed as 35 S-labeled proteins (26) and studies were carried out using GST-MPG as described above.
Pull-down assays were also used to determine whether the receptor interacted directly with MPG. Immobilized GST-MPG was combined with purified ER␣ in TEG buffer containing 200 mM NaCl and 1 M E 2 and samples were processed as described above.
To examine the effect of DNA on ER␣-MPG interaction, GST-MPG was combined with the endogenous ER␣ present in MCF-7 nuclear extracts, which was prepared from cells that had been exposed to 10 nM E 2 for 15 min. The immobilized GST-MPG was prepared as described above and incubated with 35 g of MCF-7 nuclear proteins in TEG buffer containing 100 mM NaCl in the absence or presence of oligos containing a nonspecific DNA sequence or the A2 ERE sequence for 1 h at 4°C. After four washes with TEG buffer containing 100 mM NaCl, the bound proteins were eluted with SDS sample buffer, separated by SDS-PAGE, and transferred to a nitrocellulose membrane. Western analysis was carried out using the ER␣-specific antibody sc-543 (Santa Cruz Biotechnologies) to determine the levels of bound ER␣.
Gel Mobility Shift Assays-Gel mobility shift assays were carried out as previously reported (25) with the following modifications. To examine the effect of MPG on ER␣-DNA interaction, 25 or 75 fmol of purified ER␣ was combined with 32 P-labeled, A2 ERE-containing oligos and incubated for 5 min at 25°C in Binding buffer A (20 mM Tris, 0.1 mM EDTA, 5 mM magnesium acetate, 10% glycerol, 1 mM dithiothreitol, 1 M E 2 , 0.1 mM ZnCl 2 , 20 mM KCl) containing 1 g of poly(dI-dC). Increasing amounts (0.1, 1, or 10 pmol) of purified MPG were added as indicated. The protein concentration was adjusted to 10 g with bovine serum albumin in a final volume of 10 l. Competition experiments included 1 pmol of unlabeled oligos containing a nonspecific DNA sequence or the A2 ERE. Supershift experiments included the ER␣specific antibody sc-8002 or the MPG-specific antibody. After incubation for 10 min at 25°C, the complexes were separated on a nondenaturing acrylamide gel, dried, and subjected to autoradiography. The film was scanned and bands were quantitated using ImageQuant 5.0 (Molecular Dynamics, Sunnyvale, CA).
The effect of ER␣ on the interaction of MPG with DNA was examined with hypoxanthine-containing oligos (HxDNA, 5Ј-CCTAGATTACTTC-TCATGTT/Hx/GACATACTCAGATCTACC-3Ј and 5Ј-GGTAGATCTGA-GTATGTCTAACATGAGAAGTAATCTAGG-3Ј), which were purified on a denaturing 20% acrylamide gel, isolated, annealed, and end-labeled with [␥-32 P]dATP. The labeled oligos were combined with 250 or 500 fmol of purified MPG in Binding buffer B (20 mM Hepes, pH 7.5, 0.1 mM EDTA, 5 mM MgOAc, 100 mM potassium L-glutamate, 10% glycerol, 1 mM dithiothreitol, 1 M E 2 ) containing 10 g of bovine serum albumin and 0.1 g of poly(dI-dC) per reaction in a 10-l final volume. Increasing amounts of purified ER␣ (0.5, 1, or 2 pmol) were added to the reaction mixture as indicated. Protein and salt concentrations were held constant by addition of ovalbumin and/or KCl. After 10 min at 4°C, the complexes were separated on a nondenaturing acrylamide gel. The free probe and the bound complex were quantitated using an Amersham Biosciences PhosphorImager and ImageQuant 5.0. Student's t tests were conducted to determine whether statistical differences existed in the binding of MPG to the HxDNA in the absence or presence of ER␣.
MPG Activity Assays-To determine the effect of ER␣ on MPGcatalyzed removal of hypoxanthine, 32 P-labeled HxDNA was combined with 2 pmol of ER␣, 500 fmol of MPG, or 500 fmol of MPG and increasing amounts of ER␣ (0.5, 1, or 2 pmol). Protein and salt concentrations were held constant by addition of ovalbumin or KCl. The binding reaction was incubated at 4°C for 10 min to allow binding of MPG to HxDNA, followed by incubation for 2 min at 37°C to allow MPG-catalyzed removal of hypoxanthine to occur (31). The DNA was denatured under alkaline conditions (2 mM EDTA, 0.2 N NaOH), precipitated in the presence of 1 g of tRNA, resuspended in formamide buffer, and separated on a denaturing 20% acrylamide gel. The wet gel was analyzed using Amersham Biosciences PhosphorImager and Im-ageQuant 5.0. Student's t tests were used to determine whether statistical differences existed between MPG assays carried out in the absence and presence of ER␣.
Acetylation of MPG and ER␣ by p300 -Acetylation assays were carried out with purified p300 and 3 H-labeled acetyl-CoA as described (26). Purified MPG (20 pmol), ER␣ (1 pmol), or histones were included as indicated in the presence of 1 M E 2 . Assays were conducted and quantitated using ImageQuant 5.0. Student's t tests were used to de-termine whether statistical differences existed in protein acetylation when both ER␣ and MPG were present.

RESULTS
Interaction of Nuclear Proteins with the DNA-bound ER␣-To identify novel ER␣-associated proteins, we examined the ability of the ERE-bound receptor to recruit HeLa nuclear proteins in DNA pull-down experiments. Immobilized EREcontaining oligos were incubated with purified ER␣ and HeLa nuclear extracts. After extensive washing, the proteins associated with the DNA-bound receptor were fractionated on denaturing gels and silver-stained. A unique 35-kDa band was seen in the presence of ER␣ and the A2 ERE (Fig. 1, lane 4), but was absent when ER␣ was not included in the binding reaction (lane 3). Likewise, this band was also absent when a nonspecific DNA sequence was included in the binding reaction (lanes 1 and 2). The 35-kDa silver-stained band was excised and subjected to mass spectrometry analysis. Three peptides, which contained amino acid sequences identical to that found in the MPG, demonstrated that MPG was associated with the DNAbound ER␣. MPG is a 35-kDa DNA repair protein involved in base excision repair. MPG recognizes and removes damaged bases such as 3-methylpurines leaving behind an apurinic site that is further processed by other DNA repair proteins (31)(32)(33).
Influence of A2 ERE-containing DNA on the ER␣-MPG Interaction-To determine whether MPG could interact directly with ER␣, immobilized GST-MPG was combined with purified ER␣. ER␣ bound to the GST-MPG (Fig. 2, panel A, lane 3) demonstrating that the interaction between ER␣ and MPG was direct and did not require additional proteins. In contrast, ER␣ was not retained on the resin when GST-MPG was omitted (lanes 2).
Our initial pull-down experiments suggested that ERE-containing DNA might influence the ER␣-MPG interaction. To determine whether this was the case, immobilized GST-MPG was incubated with MCF-7 nuclear extracts, which contained endogenously expressed ER␣. ER␣ was pulled down by the GST-MPG in the absence of DNA (Fig. 2, panel B, lane 3).
Addition of oligos containing a nonspecific DNA sequence did not affect the amount of ER␣ pulled down (lane 4). However, inclusion of oligos containing the A2 ERE enhanced the amount of ER␣ associated with GST-MPG (lane 5) confirming that the A2 ERE enhanced the ER␣-MPG interaction.
Regions of ER␣ Required for Interaction with MPG-Because we had established that ER␣ could interact directly with MPG, we next sought to identify the region of the receptor responsible for this interaction. In vitro synthesized, 35 S-labeled full-length or truncated ER␣ was incubated with immobilized GST-MPG in the absence or presence of E 2 . Specifically bound proteins were eluted, fractionated by SDS-PAGE, and detected by autoradiography. The full-length receptor (Fig. 3, ER␣) and the carboxyl-terminal deletion mutant (1-530) interacted with the immobilized GST-MPG in a ligand-independent manner (lanes 3 and 4). However, the amino terminus of ER␣ (AB) was incapable of interacting with GST-MPG. The CD, EF, and E domains of the receptor interacted with GST-MPG in the absence and presence of E 2 . In contrast, neither full-length nor truncated ER␣ was retained on the beads when GST-MPG was omitted (lane 2). Unprogrammed lysate, which contained no ER␣ DNA template, did not interact with GST-MPG. These data demonstrate that the DNA-binding domain, hinge region, and the LBD are involved in the ER␣-MPG interaction.
Interaction of Endogenous ER␣ and MPG-We next assessed the levels of MPG protein present in nuclear extracts from human osteosarcoma (U2-OS), cervical (HeLa), and breast cancer (MCF-7) cells using Western analysis. MPG was detected in HeLa and MCF-7 nuclear extracts, but not in U2-OS nuclear extracts (Fig. 4, panel A). Thus, MCF-7 cells, which express both MPG and ER␣ and are estrogen-responsive, provided an appropriate cell line to study the ER␣-MPG interaction in a biologically relevant environment. When MCF-7 nuclear extracts were incubated with an ER␣-specific antibody, similar levels of ER␣ were immunoprecipitated from ethanol and E 2treated cells (Fig. 4, panel B, lanes 4 and 5, upper bands). Probing of the same blot with an MPG-specific antibody revealed that MPG was immunoprecipitated with ER␣ in the absence and presence of E 2 (Fig. 4, panel B, lanes 4 and 5, lower  bands). Likewise when MCF-7 nuclear proteins were immunoprecipitated with an MPG-specific antibody, both ER␣ and MPG were present regardless of whether the cells had not or had been treated with E 2 (panel C, lanes 4 and 5, upper and lower bands). In contrast, when no antibody was included in the reaction, neither ER␣ nor MPG was detected (lane 3). Comparable levels of ER␣ and MPG were detected in nuclear extracts from ethanol-and E 2 -treated cells (lanes 1 and 2). Thus, endogenous MPG and ER␣ present in MCF-7 cells interact in the absence and presence of E 2 .
Effect of ER␣ on Interaction of MPG with Hypoxanthinecontaining DNA-MPG recognizes and binds to damaged bases such as 3-methyladenine and hypoxanthine and brings about catalytic removal of these damaged bases (31)(32)(33)(34)(35). When 500 fmol of purified MPG was incubated with a radiolabeled oligo containing a single hypoxanthine residue (HxDNA), a protein-DNA complex was formed (Fig. 5, 4-6). This increase in MPG-DNA complex formation was not because of an increase in protein concentration because total protein levels were held constant by the addition of ovalbumin. Although the presence of MPG in this complex was confirmed by supershifting the protein-DNA complex with an MPG-specific antibody, there was no change in the migration of the protein-DNA complex when an ER␣-specific antibody was used (data not shown). The specificity of the protein-DNA complex formation was confirmed by the ability of unlabeled hypoxanthine-containing oligos, but not an unmodified DNA sequence, to compete for MPG binding (data not shown). Control experiments also demonstrated that unlike MPG, ER␣ was unable to bind to the hypoxanthine-containing oligos (data not shown). Statistical analysis of three independent experiments revealed more than a 2.5-fold increase (p Ͻ 0.005) in MPG-HxDNA complex formation at the highest concentration of ER␣ used (Fig. 5, panel B). Thus, ER␣ enhanced the stability of the MPG-HxDNA complex, but did not bind to the oligo or form a ternary complex with MPG and HxDNA.
Effect of ER␣ on MPG-catalyzed Removal of Hypoxanthine from the DNA-MPG binds to hypoxanthine-containing DNA  4) were separated by SDS-PAGE and transferred to a nitrocellulose membrane, which was probed with an MPG-specific antibody. Purified MPG was included as a control (lane 1). Panels B and C, nuclear extracts from MCF-7 cells that had been treated with ethanol (ϪE 2 ) or 17␤-estradiol (ϩE 2 ) were incubated with an ER␣-(B, lanes 4 and 5) or MPG-(C, lanes 4 and 5) specific antibody. Lane 3 contained MCF-7 nuclear extract but lacked the ER␣-or MPG-specific antibodies. 20% of the MCF-7 nuclear extract input was included in lanes 1 and 2. The coimmunoprecipitated proteins were identified by Western analysis using an ER␣-(upper bands) or MPG-specific antibody (lower bands). 32 P-labeled hypoxanthine-containing oligos (HxDNA) were run alone (lane 1) or were combined with 500 (lane 2) or 250 (lanes 3-6) fmol of purified MPG. Increasing amounts of purified ER␣ were included as indicated (0.5, 1, and 2 pmol, lanes 4 -6). The complexes were separated on a nondenaturing acrylamide gel and subjected to autoradiography. Panel B, data from four independent experiments were combined, statistically analyzed using Student's t test, and expressed as the mean Ϯ S.E. An asterisk indicates that the amount of complex formed was statistically greater in the presence than in the absence of ER␣ (p Ͻ 0.04). and catalyzes the hydrolytic cleavage of the N-glycosylic bond between the hypoxanthine and sugar-phosphate backbone resulting in the generation of an apurinic site (31)(32)(33). Because the phosphodiester bonds at these apurinic sites can be cleaved in vitro under denaturing alkaline conditions, it is possible to monitor MPG activity by quantitating the amount of intact and cleaved DNA. Using this methodology, we determined whether ER␣-enhanced stabilization of the MPG-HxDNA complex had a functional consequence. When 32 P-labeled oligos containing an internal hypoxanthine on one strand were fractionated on a denaturing acrylamide gel, the full-length 39-nucleotide oligo was observed (Fig. 6, panel A, lane 1). However, when MPG was included in the reaction, the full-length oligo and a 20nucleotide cleaved product were observed (lane 2). As increasing concentrations of ER␣ were included in the reaction, there was a dose-dependent increase in the 20-nucleotide product (lanes [3][4][5]. This cleavage product was absent when ER␣, but not MPG, was incubated with the radiolabeled hypoxanthinecontaining oligos (lane 6). Analysis of data from three independent experiments demonstrated that hypoxanthine removal was significantly increased in the presence of ER␣ (p Ͻ 0.0005, panel B). Thus, ER␣ not only stabilized the MPG-DNA interaction, but also increased MPG-catalyzed removal of the modified DNA base.

FIG. 5. ER␣ stabilizes the MPG-HxDNA interaction. Panel A,
Effect of MPG on Interaction of ER␣ with the A2 ERE-containing DNA-We had thus far shown that ER␣ enhanced binding of MPG to Hx-containing DNA and increased MPGmediated repair of damaged DNA. To determine whether MPG could in turn affect the interaction of ER␣ with the A2 ERE, gel mobility shift assays were carried out. Purified ER␣ was incubated with 32 P-labeled oligos containing the A2 ERE and the complexes were fractionated on a nondenaturing acrylamide gel. As expected, 75 fmol of ER␣ formed a complex with the ERE (Fig. 7, lane 2. When only 25 fmol of ER␣ was included with the probe, there was a distinct decrease in ER␣-ERE complex formation (lane 3). On addition of increasing concentrations of MPG, there was a dramatic increase in ER␣-ERE complex formation (lanes 4 -6). The presence of ER␣ in this complex was confirmed by supershifting the receptor-DNA complex with an ER␣-specific antibody (lane 9). However, no supershifted band was seen when an MPG-specific antibody was used (lane 10). The specificity of ER␣-ERE interaction was demonstrated by the ability of unlabeled oligos containing the consensus A2 ERE to compete for ER␣ binding (lane 11). In contrast, oligos containing a nonspecific DNA sequence (lane 12) were unable to compete for ER␣ binding. Thus, MPG greatly enhanced ER␣-ERE complex formation even though MPG was not present in this complex. Thus far, our pull-down assays with HeLa and MCF-7 nuclear extracts and with purified ER␣ and MPG provided evidence that MPG was associated with the DNA-bound receptor. Our inability to detect MPG associated with the DNA-bound receptor in these gel shift assays may have resulted from the stringency of the gel shift conditions, which did not allow a stable complex to form.
MPG Modulates ER␣-mediated Transcription-Our gel shift experiments had shown that MPG dramatically enhanced the interaction of the receptor with the consensus A2 ERE. It seemed possible that MPG might also influence ER␣-mediated transcription. To test this possibility, U2-OS cells, which express no detectable level of MPG (Fig. 2, panel A), were transiently transfected with a CAT reporter vector containing a single A2 ERE and an ER␣ expression vector in the absence or presence of an MPG expression vector. When no MPG expression vector was included, a 15-fold increase in transcription was observed in the presence of 10 nM E 2 (Fig. 8, panel A). Surprisingly, when increasing amounts of an MPG expression vector were included, there was a dose-dependent decrease in E 2 -mediated transcription. Similar transfection experiments Increasing amounts of purified ER␣ (0.5, 1, and 2 pmol) were added as indicated (lanes [3][4][5]. A control lane, which included the 32 P-labeled hypoxanthine-containing oligos and ER␣ (2 pmol) was also included (lane 6). The DNA was denatured, run on a denaturing acrylamide gel, and subjected to autoradiography. Panel B, data from four independent experiments were combined, analyzed using Student's t test, and expressed as the mean Ϯ S.E. An asterisk indicates that the MPGmediated removal of hypoxanthine was significantly greater in the presence than in the absence of ER␣ (p Ͻ 0.0005). were carried out with a reporter plasmid containing an imperfect pS2 ERE. As seen with the A2 ERE-containing reporter plasmid, inclusion of increasing concentrations of the MPG expression vector led to a dose-dependent decrease in E 2 -mediated transcription (panel B). However, the level of transcription observed with the pS2 ERE-containing reporter plasmid was significantly less than that observed with the A2 EREcontaining reporter plasmid. In contrast to our findings with the ERE-containing reporter plasmids, a control ␤-galactosidase reporter plasmid was unaffected by addition of increasing amounts of the MPG reporter plasmid. These findings demonstrated that MPG decreased transcription of reporter plasmids containing simple promoters with a single ERE and a TATA box. We also determined whether MPG influenced transcription of a complex promoter. Transient transfections were carried out with a reporter plasmid containing 1.1 kb of the pS2 5Ј-flanking region (7). Again, a dose-dependent decrease in E 2 -mediated transcription was observed (Fig. 8, panel C) indicating that MPG decreases transcription of both simple and complex estrogen-responsive promoters.
ER␣-MPG Interaction Modulates p300-mediated Acetylation-Hyperacetylation of histones has been linked to increased transactivation (36,37). Likewise, acetylation of ER␣ has been associated with increased hormone sensitivity and ER␣-mediated transcription (38). The coregulatory protein p300 acetylates a number of proteins including ER␣ and modulates their activity (26, 38 -40). To determine whether MPG might affect p300-mediated acetylation of ER␣, acetylation assays were carried out. When only p300 and [ 3 H]acetyl-CoA were included in the reaction, autoacetylation of p300 was observed (Fig. 9, panel A, lane 1) as we have previously reported (26). As expected, p300 acetylated purified histones and ER␣ (panel A, lanes 2 and 3). Interestingly, MPG was also acetylated by p300 (panel A, lane 4). When ER␣ and MPG were included in the same reaction, MPG decreased ER␣ acetylation and ER␣ increased MPG acetylation (panel B). Statistical analysis of three independent experiments indicated that MPG acetylation was significantly increased in the presence of ER␣ and that ER␣ acetylation was significantly decreased in the presence of MPG (Fig. 9, panel C).
MPG Exhibits Selectivity in Its Interactions with Nuclear Receptors-Because ER␣ is a member of a large family of highly conserved nuclear receptors, we determined whether MPG might interact with other members of the nuclear receptor family. Interaction studies were carried out with immobilized GST-MPG and in vitro synthesized 35 S-labeled PRB, thyroid receptor ␤, RXR␣, or with PPAR␥. MPG interacted efficiently with thyroid receptor ␤ and PPAR␥ but bound weakly with RXR␣ in a ligand-independent manner (Fig. 10,  lanes 3 and 4). Interestingly, no interaction was observed between MPG and PRB. Thus, MPG exhibits selectivity in its interactions with these nuclear receptors. DISCUSSION We have identified a novel physical and functional interaction between a nuclear transcription factor, ER␣, and a DNA repair protein, MPG. ER␣ stabilizes the interaction of MPG with hypoxanthine-containing DNA, increases the catalytic removal of the modified base, and enhances p300-mediated acetylation of MPG. In turn, MPG increases ER␣-ERE complex formation, decreases ER␣ acetylation, and reduces ER␣-mediated transcription. Thus, our studies have provided a functional link between gene expression and DNA repair and illustrate the complementary activities of these two proteins.
Role of MPG in Cells-One of the largest classes of mutagens are alkylating agents, which include cancer therapeutics, environmental agents, and products of cell metabolism (21,22). Base alkylation produces a wide range of products including 3-methyladenine (23) that are recognized and removed by several classes of DNA gylcosylases (41). If it is not repaired, 3-methyladenine can be mutagenic or cytotoxic. 3-Methyladenine in mammalian cells is repaired primarily by MPG through the base excision repair pathway (32,33). In addition to 3-methyladenine, MPG is able to repair 7-methylguanine, hypoxanthine, which is a deamination product of adenine, and 1-N 6 -ethanolamine, which is a by product of lipid peroxidation (34,42,43).
MPG activity has been detected in nearly every organism from bacteria to humans suggesting that it plays a critical role in detecting and repairing DNA lesions (44). The fact that mouse cells, which fail to express MPG, exhibit sister chromatid exchange, chromosomal aberrations, S phase arrest, and apoptosis highlights the importance of MPG in maintaining genome integrity (44). Surprisingly, however, overexpression of MPG increases sensitivity of the cell to alkylating agents by creating an imbalance in DNA repair. Increased expression of MPG leads to increased excision of N-methylpurines leaving behind unprocessed apurinic sites, which can then serve as intermediates in the formation of chromosomal aberrations and play a role in carcinogenesis (45). Interestingly, T47D and MCF-7 breast cancer cells express higher levels of MPG protein than normal mammary cells (46). High levels of MPG have also been observed in the colon cancer cell line HT29 (47).
Interaction of MPG with ER␣-We isolated MPG in a screen MPG expression vector was included as indicated. CAT activity was normalized to ␤-galactosidase (␤gal) activity to correct for differences in transfection efficiency. Data from three independent experiments carried out in duplicate were combined and expressed as the mean Ϯ S.E. that utilized DNA-bound ER␣. These experiments combined with our GST-MPG pull-down studies indicate that ERE-containing DNA increases the interaction of ER␣ with MPG. We have previously demonstrated that DNA acts as an allosteric modulator of ER␣ conformation and thereby influences the interaction of the receptor with coregulatory proteins (5,25,27). Our current studies suggest that the interaction of the receptor with the ERE enhances the recruitment of MPG to estrogen-responsive genes.
Our gel shift studies indicated that the interaction of MPG with hypoxanthine-containing DNA was enhanced more than 2.5-fold in the presence of ER␣. This ER␣-enhanced binding of MPG to its substrate resulted in increased DNA damage recognition and repair. Studies carried out by Miao et al. (48) have demonstrated that the interaction of MPG with the DNA damage repair proteins hHR23A and -B increases the affinity of MPG for hypoxanthine-containing DNA. The ability of selected proteins to increase MPG recognition of damaged DNA may help to provide flexibility and enhanced surveillance of genomic DNA at times when additional surveillance is required. For example, E 2 induces expression of estrogen-responsive genes and thereby increases exposure of this transcriptionally active DNA to mutagens. However, the recruitment of MPG by ER␣ to target genes may help to enhance DNA surveillance and repair and reduce genome instability.
Transcriptional Repression by DNA Repair Proteins-MPG has a profound effect on ER␣-mediated transactivation. Previous studies have demonstrated that DNA repair proteins can repress transcription by sequestering transcription factors (49,50). For example, transcription is repressed during DNA repair involving transcription factor II H, which is required for transcription. It seems possible that the MPG-mediated recruitment of ER␣ to damaged DNA could limit the participation of the receptor in establishing active transcription complexes and thereby decrease ER␣-mediated gene expression.
We have demonstrated that MPG decreased p300-mediated acetylation of ER␣. Because acetylation of ER␣ increases ligand sensitivity and ER␣-mediated transcription (38), an MPG-mediated decrease in ER␣ acetylation could help to limit ER␣-mediated transcription. In addition, the ability of MPG to facilitate binding of these less acetylated receptors to estrogenresponsive promoters could further limit ER␣-mediated transcription. It is possible that the acetylation state of MPG could also influence the association of MPG with proteins involved in regulating gene expression and DNA repair.
Our studies suggest that DNA-bound ER␣ recruits MPG and other chromatin modifiers, which in turn facilitate chromatin remodeling, transcription, and DNA repair. We find that the interaction of ER␣ with MPG has functional consequences for both of these proteins. MPG modulates ER␣ binding and estrogen-mediated transcription and in turn ER␣ enhances the association of MPG with DNA and promotes repair of damaged DNA. Thus, in addition to regulating the expression of estrogen-responsive genes, ER␣, through its interaction with MPG, is involved in maintaining the integrity of the genome. In vitro transcribed and translated, 35 S-labeled full-length PRB, thyroid receptor ␤ (TR␤), RXR␣, and PPAR␥ were incubated with glutathione beads without (lanes 2) or with immobilized GST-MPG (lanes 3 and 4). The absence (ϪH) or presence (ϩH) of the receptor-specific hormones R5020 (PRB), Triac (thyroid receptor ␤), RXR␣, or BRL49653 (PPAR␥) is as indicated. The GST-MPG-associated proteins were eluted, fractionated by SDS-PAGE, and subjected to autoradiography. Lane 1 contains 10% of the input 35 S-labeled protein.