Redox regulation of GA-binding protein-alpha DNA binding activity.

We have investigated the reduction/oxidation (redox) regulation of the heteromeric transcription factor GA-binding protein (GABP). GABP, also known as nuclear respiratory factor 2, regulates the expression of nuclear encoded mitochondrial proteins involved in oxidative phosphorylation, including cytochrome c oxidase subunits IV and Vb, as well as the expression of mitochondrial transcription factor 1. GABP is composed of two subunits, the Ets-related GABP-α, which mediates specific DNA binding, and GABP-β, which forms heterodimers and heterotetramers on DNA sequences containing the PEA3/Ets motif ((C/A)GGA(A/T)(G/A)). We demonstrate here that GABP DNA binding activity and GABP-dependent gene expression in 3T3 cells are inhibited by pro-oxidant conditions. DNA binding of recombinant GABP-α was activated by chemical reduction (dithiothreitol) and by thioredoxin; however, GSSG inhibited GABP DNA binding activity. Treatment of GABP-α, but not GABP-β1, with sulfhydryl-alkylating agents also inhibited GABP DNA binding activity. Our results suggest that GABP DNA binding activity is redox-regulated in vivo, possibly by thioredoxin-mediated reduction and by GSSG-mediated oxidation of the GABP-α subunit. The regulation of GABP (nuclear respiratory factor 2) DNA binding activity by cellular redox changes provides an important link between mitochondrial and nuclear gene expression and the redox state of the cell.

Transcriptional regulation in eukaryotes takes advantage of products of cellular metabolism to control gene expression in response to physiological changes within or outside of the cell (1,2). Receptor-bound hormones induce a myriad of signal transduction pathways that modulate the activities of transcription factors at the level of DNA binding and/or transcriptional activation. In recent years, reactive oxygen species have been recognized as modulators of cellular processes including transcription (3). Transcription factors such as NF-B and Ap1 are regulated by direct reduction/oxidation (redox) reactions at the level of DNA binding as well as by pro-oxidant-induced phosphorylation reactions (4 -13).
Mitochondrial proteins involved in oxidative phosphorylation are encoded in the mitochondrial and nuclear genomes, including factors involved in mitochondrial gene regulation and replication. As a consequence, nuclear and mitochondrial gene expression is coordinately regulated in response to changes in cellular conditions, including neoplastic transformation and hypoxia (14,15). The means by which the cell coordinates nuclear and mitochondrial gene expression is not clear. Recent reports suggest that expression of the nuclear encoded mitochondrial proteins cytochrome oxidase subunits IV and Vb and mitochondrial transcription factor 1 (mtTF-1) 1 is regulated by nuclear respiratory factor 2 (NRF-2) (16 -22). Human NRF-2 is identical to human E4TF-1 and is the likely human homolog of murine GABP (18,20,(23)(24)(25)(26)(27)(28)(29).
GABP belongs to the Ets family of transcription factors (8, 30 -42), which bind to the consensus sequence (C/A)GGA(A/ T)(A/G), also known as the PEA3 motif (AGGAAG), an important element in the polyoma virus enhancer and viral late gene initiator element (43)(44)(45). The GABP-binding site, first identified in the herpes simplex virus immediate early gene promoter, is composed of two adjacent PEA3/Ets-binding sites (CGGAAGCGGAAA) (25,28). Ets family members share a common highly conserved 85-amino acid DNA-binding domain (Ets domain) that belongs to the winged helix-turn-helix family of DNA-binding proteins (46 -49). GABP (NRF-2 and E4TF-1) is unique among Ets proteins in that it is composed of two subunits that form stable dimer and tetramer complexes (19,23,25,27,29). The ␣ subunit (GABP-␣) is an Ets protein and is absolutely required for DNA binding activity of GABP heteromer complexes. The ␤ subunit (GABP-␤ 1 ) contains four tandemly repeated Notch-ankyrin repeats that mediate dimerization with GABP-␣, but lacks a discernible DNA-binding domain and is incapable of binding to DNA alone, although it greatly stabilizes GABP-␣ DNA binding. Tetramer formation is mediated by a leucine zipper-like domain located in the carboxyl terminus of GABP-␤ 1 (24,25).
The cytochrome c oxidase subunit IV and Vb genes have TATA-less promoters, whose expression is driven from initiator elements (see Fig. 1) that include two PEA3/Ets motifs that have been shown to be bound by GABP (NRF-2) (5, 14, 17, 18, 20 -22). The cytochrome c oxidase subunit IV initiator contains Sp1 sites in addition to the two PEA3/Ets sites, and the cytochrome c oxidase subunit Vb promoter contains Sp1-binding sites and a putative YY1-binding site (3). Since mitochondrial function is dependent on large enzyme complexes that include both nuclear and mitochondrial encoded subunits, the expression of nuclear and mitochondrial genes is coordinated during mitochondrial biogenesis. The demonstration that mtTF-1 is regulated, in part, by GABP (NRF-2) provides a potential link between nuclear and mitochondrial gene regulation. While specific cellular modulators of mitochondrial protein expression have yet to be identified, there is evidence that suggests that hormonal and metabolic stimuli affect expression of both nuclear and mitochondrial encoded proteins, suggesting that one or more signal transduction pathways may modulate expression of these genes (14,15). Mitochondrial protein expression might also be regulated through cellular redox changes. Our studies demonstrate that both DNA binding and, as a result, transcriptional activation by GABP are redox-regulated through the oxidation/reduction of one or more cysteine residues in the DNA-binding and dimerization domains of GABP-␣. These results suggest that GABP is an important regulatory link between mitochondrial metabolic function, nuclear gene expression, and the redox state of the cell.
For transient transfection of NIH 3T3 cells, test DNAs (20 g) and internal control DNA (pSV-␤-galactosidase expression vector; 10 g; Promega) were coprecipitated with calcium phosphate and added to the plates (90 mm) as described previously (43,45). After 24 h, the cultures were treated with 0.5 mM DEM for 6 or 8 h, and the cells were harvested by scraping. Cell pellets were resuspended in 100 l of extraction buffer (100 mM potassium phosphate, pH 7.8, 1 mM dithiothreitol) and subjected to three cycles of freezing and thawing at Ϫ70 and 37°C, respectively. Cell debris was pelleted, and the supernatants were collected and immediately analyzed for luciferase and ␤-galactosidase activities. The means Ϯ S.D. were determined from three independent transfections of each construct, and the data were expressed as arbitrary luciferase units. All luciferase measurements were normalized for transfection efficiency to ␤-galactosidase expressed from the plasmid pSV-␤galactosidase and corrected for cell viability as determined by trypan blue exclusion (97 and 70% of untreated controls at 6 and 8 h of DEM treatment, respectively).
To determine luciferase activity in cell extracts, 10 l of the cell lysate were added to 90 l of assay buffer (0.25 mM ATP, 10 mM MgCl 2 , and 100 mM potassium phosphate, pH 7.8). Luminescence was measured in a Turner model TD-20e luminometer, and the raw data were expressed in arbitrary units according to the manufacturer. ␤-Galactosidase assays were performed by adding 30 l of the cell extract to 270 l of assay buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 1 mM MgCl 2 , 50 mM ␤-mercaptoethanol, and 0.665 mg/ml o-nitrophenyl ␤-D-galactoside), followed by incubation at 37°C for 45 min. The reaction was terminated by addition of 500 l of 1.0 M Na 2 CO 3 , and the absorbance was measured at 420 nm.
The individual rGABP proteins were expressed in E. coli BL21 LysS according to the manufacturer's instructions (Novagen). The recombinant proteins were recovered by nickel chelating chromatography (QIA-GEN Inc.). The purified recombinant proteins were renatured by dialysis against several changes of buffer A (10 mM Tris, pH 8.5, 300 mM NaCl, 6.0 M urea) containing decreasing amounts of urea, followed by dialysis against EMSA buffer. The presence of the rGABP proteins in bacterial extracts was confirmed by immunoblotting with rabbit anti-GABP-␣ and anti-GABP-␤ 1 sera and by EMSA. 2 Treatment of rGABP Proteins with Thioredoxin (Trx) and the Alkylating Agent N-Ethylmaleimide (NEM)-Bacterial Trx (1.0 mM; Sigma) was first reduced with 1.0 mM DTT for 1 h at 4°C. rGABP proteins were combined and treated with 0.05 mM oxidized or reduced Trx for 30 min, followed by subsequent treatment with GSH (1.0 mM) or GSSG (1.0 mM) as indicated, and DNA binding activity was measured by EMSA analysis. For NEM treatment, concentrated solutions (ϳ0.2 mg/ml) of rGABP proteins were first pre-reduced by treatment with 1.0 mM DTT for 30 min at 25°C and then diluted to a final DTT concentration of 0.01 mM. The reduced rGABP proteins (␣, ␤, or ␣c) were treated with 0.5 mM NEM for 20 min at 25°C in EMSA buffer, pH 7.1, and the excess NEM was removed by dialysis prior to combining the GABP subunits and subsequent EMSA analysis. The lower pH used in these experiments was used to minimize non-thiol-specific reactions between NEM and other amino acid side chains (54).

Effect of Oxidants/Antioxidants on GABP DNA Binding Activity in 3T3
Cells-GABP (NRF-2) has been associated with transcriptional regulation of several genes whose products are involved in oxidative phosphorylation, including cytochrome oxidase subunits IV and Vb and mtTF-1 (Fig. 1). These observations suggest that GABP, as an important link between nuclear and mitochondrial gene regulation, may be redox-reg- ulated. To determine if cellular redox changes affect GABP activity, we have analyzed the effect of oxidant and antioxidant treatment of mouse 3T3 fibroblasts on GABP DNA binding activity ( Fig. 2A). Mouse 3T3 cells were treated for 2 h with the GSH-depleting agent DEM in the presence or absence of NAC, an antioxidant and a precursor of GSH synthesis (55). Following treatment, the cells were harvested, and nuclear extracts were prepared in the absence of a reducing agent. GABP DNA binding activity was measured by EMSA analysis using oligonucleotide probes containing a single (PEA3m) or two tandem (dPEA3-0) PEA3/Ets-binding sites.
It has been shown previously that GABP binds to a monomeric PEA3/Ets-binding site as a heterodimer (␣␤ 1 ) and to a dimeric PEA3/Ets motif as a heterotetramer (␣ 2 (␤ 1 ) 2 ) (25,26,28). One major complex was detected using the PEA3m probe, and two complexes were detected using the dPEA3-0 probe, one of which comigrated with the major complex formed on the PEA3m probe, consistent with their identity as GABP dimer and tetramer complexes ( Fig. 2A, lanes 1 and 5). The presence of GABP-␣ and GABP-␤ 1 proteins in these DNA-protein complexes was confirmed by treatment with anti-GABP-␣-and anti-GABP-␤ 1 -specific sera (Fig. 2B). Preimmune serum had no effect on complexes formed with either the PEA3m or dPEA3-0 probe (Fig. 2B, lanes 1 and 4); however, serum specific for GABP-␣ (lanes 2 and 5) or GABP-␤ 1 (lanes 3 and 6) inhibited both complexes, indicating that the DNA-protein complexes formed with 3T3 nuclear extracts contain GABP-␣ and GABP-␤. Furthermore, the two complexes formed on the dPEA3-0 probe with 3T3 nuclear extracts comigrated with complexes formed with purified recombinant GABP-␣ and GABP-␤ 1 proteins and likely represent dimer and tetramer complexes, as indicated in Fig. 2. 2 Treatment of 3T3 cells with DEM resulted in a dramatic decrease in the formation of GABP dimer ( Fig. 2A, lane 2) and tetramer (lane 6) complexes. Conversely, treatment of 3T3 cells with NAC alone resulted in little or no effect on GABP dimer and tetramer complex formation ( Fig. 2A, lanes 3 and 7). C/EBP DNA binding activity in 3T3 cell nuclear extracts, which is not redox-regulated, was not affected by treatment with either DEM or NAC ( Fig. 2A, lanes 10 and 11) or by treatment with both DEM and NAC (lane 12). Inhibition of GABP DNA binding activity by DEM treatment was nearly completely prevented by the simultaneous addition of NAC, an antioxidant and precursor of GSH synthesis ( Fig. 2A, lanes 4 and 8). The reduction of GABP DNA binding activity observed in nuclear extracts from 3T3 cells treated with DEM was not due to loss of GABP protein since the amount of GABP-␣ (see Fig. 2A, Immuno-Blot, lanes 1-4) and GABP-␤ 1 2 in these extracts was unaffected by DEM or NAC treatment. Furthermore, treatment of the nuclear extracts prepared from DEM-treated 3T3 cells with DTT restored GABP binding activity present in these extracts, 2 indicating that GABP DNA binding activity is inhibited by oxidative stress, i.e. GSH depletion.
Previous reports have demonstrated that DEM treatment increases the relative GSSG levels of cells (55,60,61), and our results show that DEM treatment of 3T3 cells leads to inactivation of GABP DNA binding activity, possibly by direct oxidation by GSSG. To test the ability of GSSG to directly oxidize GABP and thereby inhibit its DNA binding activity, 3T3 cell nuclear extracts were treated with DTT, GSSG, or GSH, and the effect on GABP dimer and tetramer DNA binding activities was determined by EMSA analysis using the dPEA3-0 probe (Fig. 3). Addition of 5 mM DTT or GSH to 3T3 nuclear extracts had little effect on GABP dimer and tetramer DNA binding activities, although a modest increase in DNA binding activity was consistently observed (Fig. 3, compare lanes 2 and 4 with  lane 1). Treatment of 3T3 nuclear extracts with 5 mM GSSG, however, nearly abolished GABP DNA binding, demonstrating that GABP DNA binding activity is inactivated upon oxidation by GSSG. As reported previously (5), GSSG (but not GSH) treatment also inhibited Ap1 DNA binding activity in 3T3 nuclear extracts (Fig. 3, lanes 9 -12), but had little or no effect on C/EBP DNA binding activity (compare lanes 5 and 7). These results suggest that GSSG likely plays a role in redox regulation of GABP DNA binding activity.
Glutathione Depletion Inhibits GABP-dependent Gene Ex- pression-We have demonstrated that GABP is the predominant factor that binds to the PEA3m or dPEA3-0 motifs present in 3T3 cells (Fig. 2B). Depletion of GSH levels in 3T3 cells by treatment with DEM inhibited GABP DNA binding activity, without reducing the levels of GABP proteins ( Fig. 2A), suggesting that GABP-dependent gene activation is potentially regulated by cellular redox changes. To directly demonstrate such regulation, we measured the effect of DEM treatment on expression of transiently transfected luciferase reporter constructs containing a TATA box with either upstream GABP-or C/EBP-binding sites (Fig. 4). DEM treatment had no effect on luciferase expression from C/EBP-TA-Luc after 6 or 8 h of treatment (Fig. 4), consistent with previous demonstrations that C/EBP DNA binding is not subject to redox regulation. DEM treatment of cells transfected with dPEA3-0-TA-Luc, however, resulted in a 28% decrease in luciferase expression after 6 h of DEM treatment and a 62% decrease in luciferase expression after 8 h of DEM treatment (Fig. 4). These results confirm our observations that glutathione depletion inhibits GABP DNA binding activity and, as a consequence, the expression of a GABP-dependent reporter construct.
Thioredoxin Efficiently Activates rGABP DNA Binding Activity-We have demonstrated that GSH depletion in 3T3 cells inhibits GABP DNA binding activity and GABP-dependent gene expression. To more precisely analyze GABP redox regulation, we expressed GABP-␣ and GABP-␤ 1 as His 6 fusion proteins in E. coli and purified and renatured the recombinant proteins in the absence of a reducing agent. Reconstituted rGABP-␣ and rGABP-␤ 1 proteins were unable to bind to the PEA3m probe in the absence of a reducing agent (Fig. 5, lane 1). When treated with 1.0 mM DTT, however, DNA binding was activated (Fig. 5, lane 3), whereas treatment with 0.1 mM DTT, 1.0 mM GSH, or 1.0 mM GSSG failed to activate rGABP DNA binding activity (lanes 2, 6, and 7). These results demonstrate that GABP must be reduced to activate DNA binding activity and further suggest that GSH is an inefficient reducing agent for activation of GABP DNA binding activity.
GSH is not the only reducing agent present in mammalian cells; Trx is also an important reducing agent in both eukaryotic and prokaryotic cells (56,57). The thioredoxins are highly conserved proteins, which, together with thioredoxin reductase, are potent reducing agents in vivo. To determine whether Trx is capable of activating GABP DNA binding activity, we treated rGABP-␣ and rGABP-␤ 1 with bacterial Trx. The commercial source of Trx was primarily in the oxidized form, but was converted to the reduced form by treatment with an equimolar amount of DTT (58). Treatment of rGABP with oxidized Trx (0.05 mM) failed to activate rGABP DNA binding (Fig. 5, lane 4), whereas treatment with reduced Trx (0.05 mM) restored DNA binding activity of oxidized rGABP to levels nearly equivalent to those obtained by treatment with 1.0 mM DTT (lanes 3 and 5). DTT carry-over (0.05 mM) following reduction of Trx with DTT was not responsible for activation of rGABP DNA binding since 0.1 mM DTT alone failed to activate DNA binding (Fig. 5, compare lanes 2 and 5), indicating that reduced Trx is responsible for activation of rGABP DNA binding activity.
Treatment of 3T3 cell nuclear extracts with GSSG inhibited GABP DNA binding activity. To confirm these results, we pretreated rGABP with reduced or oxidized Trx and then subsequently added GSH (1.0 mM) or GSSG (1.0 mM). As in Fig. 5  (lane 4), rGABP pretreated with oxidized Trx failed to bind DNA, and GSH was unable to activate DNA binding (Fig. 5,  lanes 8 and 10). rGABP pretreated with reduced Trx, however, bound to DNA, but was not further activated by treatment with GSH (Fig. 5, lane 9). When rGABP pretreated with reduced Trx was subsequently treated with GSSG, however, nearly complete inhibition of rGABP DNA binding activity was observed (Fig. 5, compare lanes 3, 5, and 11), demonstrating that  2, 6, and 10), 5.0 mM GSSG (lanes 3, 7, and 11), or 5.0 mM GSH (lanes 4, 8, and 12). GABP DNA binding activity is subject to redox regulation through reduction by Trx (activation) and oxidation by GSSG (inhibition).

Demonstration That Cysteine(s) in GABP-␣, but Not in GABP-␤ 1 , Are Responsible for GABP Redox Regulation-
The previous experiments demonstrate that GABP DNA binding is redox-regulated; however, since GABP is composed of two subunits, we sought to identify which subunit was the target of this regulation. Alkylation of cysteines by NEM has been used to identify functionally important cysteine residues in proteins including Ap1, NF-B, and Ets-1 (4,5,8,10,54,59). As can be seen in Fig. 6, GABP-␣ contains nine cysteines, including two residues located within the DNA-binding domain (Cys-338 and Cys-388) and two in the carboxyl-terminal region associated with GABP-␣␤ dimerization (Cys-401 and Cys-421) (25,45).
To determine the effect of NEM treatment on GABP DNA binding activity, rGABP-␣ and rGABP-␤ were reduced separately with 5 mM DTT in order to ensure that each of the proteins was completely reduced. The reduced proteins were then dialyzed to remove excess DTT (Ͻ0.01 mM) and were immediately treated with NEM (0.5 mM), followed by dialysis of the alkylated proteins to remove excess NEM. After normalizing the concentration of the alkylated and non-alkylated GABP-␣ and GABP-␤ 1 proteins by SDS-polyacrylamide gel electrophoresis analysis, 2 the proteins were combined, and their DNA binding activities were analyzed by EMSA (Fig. 7). Since the NEM-Cys conjugate is not cleaved by reducing agents, the non-alkylated GABP proteins were maintained in the reduced state by inclusion of 1.0 mM DTT during EMSA analysis. Treatment of GABP-␣ with NEM abolished its capacity to bind DNA alone (Fig. 7, lanes 1 and 2) or when combined with untreated GABP-␤ 1 (lanes 5 and 6). In contrast, treatment of GABP-␤ 1 with NEM had no effect on the DNA binding activity of the GABP-␣␤ dimer complex (Fig. 7, lane 7), indicating that alkylation of cysteines in GABP-␣, but not in GABP-␤ 1 , inhibits GABP DNA binding activity. Furthermore, identical results were obtained using the sulfhydryl-specific reagent 5Ј,5Ј-dithiobis(2-nitrobenzoic acid) (Ellman's reagent), except that inhibition of GABP DNA binding was reversed by DTT, as was expected, since the reaction of 5Ј,5Ј-dithiobis(2nitrobenzoic acid) with sulfhydryl groups results in the formation of a disulfide bond that is cleavable by DTT reduction. 2 From these results, we conclude that one or more cysteines present in the GABP-␣ subunit are targets for redox regulation of GABP DNA binding activity.
The carboxyl-terminal 138 amino acids (positions 316 -454) of GABP-␣ have been previously shown to be sufficient for DNA binding and dimerization with GABP-␤ (25,28). We therefore constructed a recombinant DNA in pET15b encoding this Cterminal region of GABP-␣ (GABP-␣c) (Fig. 6) and measured its sensitivity to NEM treatment. The GABP-␣c monomer, in the absence of GABP-␤ 1 , binds poorly to the PEA3m probe used in these studies, and the monomeric GABP-␣c DNA-protein complex is not apparent in this gel (Fig. 7, lanes 3 and 4). When complexed with GABP-␤ 1 , however, the GABP-␣c␤ dimer complex is readily observed (Fig. 7, lanes 9 -12). Like full-length GABP-␣, treatment of GABP-␣c with NEM, but not GABP-␤ 1 , abolished DNA binding of the GABP-␣c␤ dimer complex (Fig. 7, lanes 9 -12), demonstrating that one or more of the four cysteines present in the DNA-binding and dimerization domains present in the GABP-␣c protein (Cys-338, Cys-388, Cys-401, or Cys-421) are responsible for NEM inactivation of GABP DNA binding and are likely targets of redox regulation of GABP DNA binding activity in vivo.

DISCUSSION
Redox regulation of transcription factors is now recognized as an important mechanism for regulating gene expression, as has been demonstrated for the transcription factors Ap1 and NF-B. Reduction of specific cysteine residues in c-Jun (Cys-272), c-Fos (Cys-154), and the p50 subunit of NF-B (Cys-62) is required for DNA binding of these transcription factors (3-6, 8 -12, 59). We have presented several lines of evidence that support the conclusion that the Ets protein GABP is redoxregulated through one or more cysteine residues located in the DNA-binding and dimerization domains of the GABP-␣ subunit. Since GABP (NRF-2) is involved in the regulation of several promoters of genes encoding proteins involved in cellular respiration, our observations suggest that GABP may function as an important link between the redox state of the cell and nuclear and mitochondrial gene regulation.
GABP-␣, like Ap1 and NF-B, requires reduction of cysteines in the DNA-binding Ets domain; however, the regulation of  7. Effect of NEM treatment on rGABP DNA binding activity. rGABP-␣, rGABP-␤ 1 , and the truncated rGABP-␣c proteins were treated with 0.5 mM NEM as described under "Experimental Procedures." DNA binding activity was measured by EMSA analysis using the PEA3m probe. The presence of individual rGABP proteins in the EMSA assay is indicated above the gel (ϩ, NEM-treated; Ϫ, untreated). A blank indicates the absence of that protein in the EMSA assay. The rGABP-␣ monomer and GABP-␣␤ dimer complexes are indicated by arrows to the left. The GABP-␣c␤ complex is indicated by the arrow to the right. The GABP-␣c monomer complex is not apparent in this gel because the truncated protein binds to the monomeric PEA3m probe very poorly under the conditions used in this experiment. GABP DNA binding activity seems to be very different from that of Ap1 and NF-B. Pro-oxidant conditions (DEM treatment) resulted in nearly complete inhibition of GABP DNA binding activity in nuclear extracts from DEM-treated 3T3 cells, and the antioxidant NAC substantially protected GABP DNA binding activity from DEM-mediated inhibition. DEM is irreversibly conjugated with GSH through the action of glutathione transferase and results in reduced intracellular levels of GSH. NAC may protect cells from the action of DEM by direct chemical conjugation with DEM, by antioxidant action, or by replenishing GSH levels through biosynthesis of GSH from NAC, its immediate biosynthetic precursor (55). Chemical (nonenzymatic) reaction of NAC with DEM is relatively slow (Ͻ20% after 1 h), whereas enzymatically catalyzed DEM depletion of GSH in cultured cells is nearly 80% complete in Ͻ1 h (60,61). Under the conditions of our experiments, only a small fraction of the DEM would likely be lost to direct chemical conjugation with NAC, suggesting that the protective effect of NAC on DEM inhibition of GABP DNA binding is the restoration of GSH levels. These observations are consistent with the conclusion that GABP DNA binding activity is negatively affected by oxidative changes in vivo.
In contrast to the results presented here, pro-oxidant conditions have been reported to result in activation of NF-B and, to a lesser extent, Ap1 activities. The activation of these factors by redox changes is, in part, an indirect consequence of activation of signal transduction pathways that lead to induction of Ap1 synthesis and phosphorylation of the inhibitor of B, IB (MAD-3), which results in the targeted degradation of IB and the subsequent activation and nuclear transport of NF-B (4,8,9,11,12). Although NF-B and Ap1 may be induced by prooxidant conditions, the DNA binding activities of both factors depends on reduction of specific cysteines in their DNA-binding domains, possibly by Trx, GSH, or factors such as Ref-1 (4). Induction of Ap1 and NF-B activities by pro-oxidant conditions has been proposed to constitute an early event in the cellular antioxidant response to oxidative damage (9,11). Thus, oxidative conditions result in activation of Ap1 and NF-B, which, in turn, induce the expression of factors involved in the antioxidant response.
The fact that GABP DNA binding activity is inhibited by pro-oxidant conditions and activated by antioxidants suggests that GABP does not play a positive role in an antioxidant response, as has been proposed for NF-B and Ap1 (9,11). Rather, GABP responds negatively to pro-oxidative conditions, leading to reduced expression of genes regulated by GABP (NRF-2). Our demonstration that GSH depletion in 3T3 cells inhibited expression from a GABP-dependent reporter gene supports the conclusion that redox regulation of GABP DNA binding activity may have a significant effect on GABP-dependent expression of enzymes involved in oxidative phosphorylation, such as cytochrome c oxidase subunits IV and Vb. In addition, pro-oxidant conditions would likely diminish expression of mtTF-1, leading to reduced mitochondrial gene expression and biogenesis. It is therefore possible that the redox state of GABP helps modulate oxidative metabolism of the cell through feedback inhibition, in which increased production of reactive oxygen species during pro-oxidant conditions would lead to decreased GSH levels and increased GSSG levels, resulting in inhibition of GABP DNA binding activity and, consequently, decreased expression of mtTF-1, cytochrome c oxidase subunits IV and Vb, and possibly other oxidative phosphorylation proteins.
We have also demonstrated that Trx is a very efficient reducing agent for GABP and is capable of activating GABP DNA binding activity in vitro. Since Trx reductase is dependent upon NADPH for enzymatic reduction of oxidized Trx and since NADPH is the central source of reducing equivalents for reductive biosynthesis as well as reduction of GSSG by glutathione reductase, the availability of NADPH is a sensitive measure of the oxidative state of the cell (56). Thus, NADPH levels could determine the levels of available reduced Trx and thereby regulate the expression of oxidative phosphorylation genes through redox regulation of transcription factors (including GABP) that depend upon reduced Trx for maintenance of DNA binding activity. Preliminary experiments show that cotransfection of a Trx expression construct (pCMV-ADF) (11) enhances transcription of a luciferase reporter construct whose expression is dependent on an artificial GABP-binding site, 2 consistent with the notion that Trx mediates GABP reduction and subsequent DNA binding and transcriptional trans-activation in vivo.
Of the two GABP subunits, only GABP-␣ is subject to redox regulation in vitro. We have demonstrated that alkylation of the four cysteine residues in the truncated GABP-␣c protein by NEM treatment inhibits DNA binding, suggesting that one or more of these cysteines are the targets of GABP redox regulation. Comparison of other Ets domain proteins with GABP-␣ revealed no member of the family with a cysteine residue in a position analogous to Cys-388 of GABP-␣ (Fig. 8). Many Ets domain proteins, however, contain cysteines at positions analogous to or in close proximity to Cys-338 and Cys-401 of GABP-␣. These observations suggest a potentially unique role for Cys-388 in the redox regulation of GABP-␣ DNA binding activity. Mutational analysis is currently underway that should identify the specific cysteine residues involved in redox regulation of GABP-␣ and may lend insights into the mechanism(s) of GABP transcriptional regulation.  (20, 25, 28, 30 -40, 42, 49). Structural motifs predicted by NMR analysis of the c-Ets-1 and Fli-1 Ets domains are indicated at the bottom of the figure (46,47). Predicted ␣ helix and ␤ strands are indicated by filled and open boxes, respectively. m, murine; h, human; d, Drosophila.