The Human Copper-Zinc Superoxide Dismutase Gene (SOD1) Proximal Promoter Is Regulated by Sp1, Egr-1, and WT1 via Non-canonical Binding Sites*

Human copper-zinc superoxide dismutase (Cu,Zn-SOD) participates in the control of reactive oxygen intermediate intracellular concentration. In this study, we show that phorbol 12-myristate 13-acetate (PMA) increases Cu,Zn-SOD mRNA expression within 30 min. The sequence between nucleotides −71 and −29 is essential for both basal and PMA-induced gene expression. This region includes an Sp1-binding site that is also recognized by a possible Sp1-like protein and by Egr-1 in a PMA-inducible manner. Egr-1 and two splicing variants of the Egr-related protein WT1 were able to transactivate the SOD1 promoter in co-transfection experiments. Sp1 and the possible Sp1-like proteins bind to two overlapping, but distinct sequences. However, Egr-1 and Sp1 seem to interact with two sites that are either identical or very close to each other. None of these sites fit the consensus sequences previously reported for these proteins. Analysis of various mutants of theSOD1 proximal promoter revealed that the region that binds Sp1 and Egr-1 is required for both basal and Egr-1-driven expression. Interplay between different members of the Sp1 family, Egr-1, and different splicing variants of WT1 in the SOD1 proximal promoter may provide clues about the physiological function of Cu,Zn-SOD.

Superoxide dismutases (SODs) 1 are found in all aerobic organisms and are thought to participate in the detoxification of reactive oxygen intermediates by catalyzing superoxide anion (O 2 . ) dismutation, yielding hydrogen peroxide (H 2 O 2 ) and molecular oxygen (O 2 ) (1). Three distinct SODs have been described in mammalian cells: a mitochondrial manganese-associated SOD, whose function is likely to detoxify superoxide anion generated by partial O 2 reduction during oxidative phos-phorylation; an extracellular SOD, which is found in extracellular fluids and bound to the extracellular matrix; and a cytosolic Cu,Zn-SOD, whose endogenous role remains poorly understood (2,3). The fundamental importance of SOD activity is illustrated in various organisms by the phenotype of mutants that do not express a form of SOD. In bacteria and yeast, SOD deficiency leads to severe growth deficits under aerobic conditions (4,5). The life span of Drosophila that lack SOD is shortened. Conversely, Drosophila that overexpress SOD and catalase can live up to 150% longer (6,7). Manganese-associated SOD-deficient mice die shortly after birth (8,9). Human Cu,Zn-SOD is encoded by a gene (SOD1) located in the 21q22 region (10). This region is involved in the most common genetic disease, known as trisomy 21 or Down's syndrome. However, although a 1.5-fold increase in SOD1 gene expression may be associated with the disease, several Down's syndrome patients carrying a partial trisomy 21 have a normal Cu,Zn-SOD activity (11). It has been reported that some cases of familial amyotrophic lateral sclerosis, a serious neurodegenerative disease, are associated with dominant mutations in the SOD1 gene (12,13). However, a gain of function is likely to cause familial amyotrophic lateral sclerosis, rather than a defect of Cu,Zn-SOD activity (14). Accordingly, in mice, Cu,Zn-SOD homozygous null mutants are viable and do not exhibit the neuropathological symptoms characteristic of familial amyotrophic lateral sclerosis patients. However, these Cu,Zn-SOD Ϫ/Ϫ mice exhibit an enhanced neuronal cell death following axotomy or ischemia, suggesting a requirement for Cu,Zn-SOD under stress (15,16).
Although Cu,Zn-SOD activity has been found modified in some specific situations including hematopoietic cell differentiation, aging, and some tumor cells (17)(18)(19), the SOD1 gene is often considered as a "housekeeping gene," and its activity is often used as an internal control when variations of manganese-associated SOD activity are measured. We report here the analysis of the proximal region of the SOD1 gene promoter. Phorbol 12-myristate 13-acetate (PMA) was used as a model agonist to identify trans-and cis-acting factors for SOD1 gene expression. We found that SOD1 gene expression is rapidly enhanced following PMA treatment. A region between nucleotides Ϫ71 and Ϫ29 was essential for both basal and PMAinduced expression. This region could be further refined to a single Sp1/Egr-1/WT1-binding site located between nucleotides Ϫ59 and Ϫ48. These findings suggest that competition between Sp1, Egr-1, and variants of WT1 (and perhaps other Sp1-or Egr-related proteins) for the same binding site may play a role in the regulation of SOD1 gene expression in response to a variety of biological signals.

EXPERIMENTAL PROCEDURES
Cell Culture-The HeLa cell line was obtained from the American Tissue Culture Collection. Cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heatinactivated fetal calf serum, 2 mM glutamine, and antibiotics at 37°C and 5% CO 2 in a humidified incubator. Cells were periodically checked for absence of mycoplasma contamination.
Chemicals-PMA (Sigma) was resuspended in dimethyl sulfoxide to a stock concentration of 1.5 mM, aliquoted, and stored at Ϫ80°C. For Northern blotting, electrophoretic mobility shift assay (EMSA), and transfection experiments for which PMA has been used, a control including dimethyl sulfoxide alone was added to ensure the absence of solvent-generated artifact. No effect of dimethyl sulfoxide was found in any of the assays.
Oligonucleotides-Oligonucleotides were obtained from Oligo-Express (Paris, France) and used without any further purification, except for site-directed mutagenesis, for which purification on polyacrylamide gel electrophoresis was carried out.
Northern Blotting-Total cellular RNA were purified using the conventional guanidium thiocyanate-CsCl method (20). Thirty micrograms of total RNA were loaded on each gel lane. Electrophoresis, transfer onto nitrocellulose membrane (Hybond C, Amersham Pharmacia Biotech), and hybridization were carried out according to standard protocols (21). The SOD1 probe was obtained by polymerase chain reaction on total HeLa cDNA using primers 5Ј-ACCTAAGCTTATGGCGAC-GAAGGCCGTGTG-3Ј and 5Ј-ACTCAAGCTTCCTCAGACTACATC-CAAGGG-3Ј and corresponds to a 490-base pair fragment containing the full-length SOD1 coding sequence. HindIII sites (underlined) were introduced to facilitate subcloning. An internal control probe was derived from a cDNA clone encoding human 18 S rRNA and was a gift from Dr. A. Harel-Bellan (Institut Fédératif sur le Cancer, Villejuif, France). Both probes were labeled by the "random-priming" technique using a Multiprime kit (Amersham Pharmacia Biotech) following the manufacturer's instructions. Membranes hybridized with the SOD1 probe were autoradiographed for 24 h. When the 18 S probe was used, membranes were autoradiographed for 10 min.
Western Blotting-Protein extraction, polyacrylamide gel electrophoresis, and blotting were carried out according to standard procedures (22). A polyvinylidene difluoride membrane (Polyscreen, NEN Life Science Products) was incubated with anti-human Cu,Zn-SOD primary antibody from sheep (Calbiochem). Rabbit anti-sheep secondary antibody was from Jackson ImmunoResearch Laboratories, Inc. 125 I-Labeled protein A (Amersham Pharmacia Biotech) was used to detect the rabbit secondary antibody (protein A does not bind sheep antibodies). 125 I was used as the detection method because of the poor linearity of signals generated with the peroxidase/light-based systems. The membrane was autoradiographed for 12 h.
Isolation of a SOD1 5Ј-Flanking Region Clone-A EMBL4 human genomic library (a gift from Dr. Malek Djabali, CNRS, Marseille-Luminy, France) was screened according to standard procedures (21). The probe used to screen this library was generated by polymerase chain reaction using oligonucleotides 5Ј-ACGCGGATCCGCCATTTTCGCG-TACTGCAAC-3Ј and 5Ј-TGCAGGATCCTCGCAACACAAGCCTCCC-3Ј. These primers amplify a 458-base pair fragment corresponding to positions Ϫ271 to ϩ187 of the SOD1 sequence described by Levanon et al. (23). After four successive screening rounds, one positive clone was obtained and purified. Identification of this clone was further assessed by restriction mapping analysis and comparison with the predicted restriction map derived from the sequence described by Kim et al. (24).
pGLSϪx Constructs-Polymerase chain reaction fragments corresponding to various lengths of the SOD1 5Ј-flanking region were generated using Pfu DNA polymerase (Stratagene) to ensure high fidelity amplifications. The proximal primer (reverse primer) spans positions Ϫ3 to ϩ17 and was the same in all reactions: 5Ј-CCGAAAGCTT-GAGACTACGACGCAAACCAG-3Ј. A HindIII site (underlined) was added to facilitate subsequent cloning. Forward primers were designed according to the size of the desired fragment: Ϫ1499, 5Ј-CCGACTC-GAGCCCTTGGCAAGTTTACAATG-3Ј, Ϫ952, 5Ј-CCGACTCGAGGTG-GTCCCAGGTACTTGGGA-3Ј, Ϫ750, 5Ј-CCGACTCGAGTATTTCCCT-TGAAAGGTAAG-3Ј, Ϫ552, 5Ј-CCGACTCGAGACCGAATTCTGCCAA-CCAAA-3Ј, Ϫ355, 5Ј-CCGACTCGAGTGGCCAAACTCAGTCATAAC-3Ј, Ϫ157, 5Ј-CCGACTCGAGACGCGCCCCTTGCCCCGCCC-3Ј, Ϫ71, 5Ј-C-CGACTCGAGATTGGTTTGGGGCCAGAGTG-3Ј, and Ϫ29, 5Ј-CCGAC-TCGAGTATAAAGTAGTCGCGGAGAC-3Ј. The XhoI site (underlined) was introduced at the 5Ј-extremity of all forward primers to allow directional cloning into the reporter pGL2-Basic vector (Promega). Amplifications were carried out using 100 ng of bacteriophage DNA as a template and 15 cycles in a thermal cycler (Perkin-Elmer). The different polymerase chain reaction fragments were introduced into the promoterless pGL2-Basic Photinus pyralis luciferase reporter vector (Promega) between XhoI and HindIII sites. One recombinant clone for each construct was chosen, and plasmid DNA was extracted and purified by alkaline lysis and ion-exchange chromatography (QIAG-EN, Inc.) following the manufacturer's instructions. The nucleotide sequences of construct inserts were determined by dideoxy chain termination fluorescent automated sequencing (Euro Sequences Genes Services, Montigny-Le-Bretonneux, France). These sequences were compared with the corresponding regions of the sequence previously described by Kim et al. (24) to check for polymerase-related misincorporation of nucleotides. Site-directed mutagenesis was performed on pGLSϪ71-derived single-stranded DNA according to the protocol described by Kunkel (25).
Preparation of Nuclear Extracts-HeLa nuclear extracts were prepared using a simplified version of the method of Dignam et al. (26). Briefly, 1-2 ϫ 10 7 exponentially growing cells were collected by scraping; rinsed once in phosphate-buffered saline; and resuspended on ice in 3 pellet volumes of 20 mM Hepes (pH 7.0), 0.15 mM EDTA, 0.015 mM EGTA, 10 mM KCl, and protease inhibitors containing 1% Nonidet P-40. Membrane disruption and nucleus integrity were checked by visual inspection under a microscope. Nuclei were collected by centrifugation and resuspended in 5 pellet volumes of 10 mM Hepes (pH 8.0), 25% glycerol, 0.1 M NaCl, 0.1 mM EDTA, and proteases inhibitors. After centrifugation, nuclei were resuspended in 2 pellet volumes of hypertonic 10 mM Hepes (pH 8.0), 25% glycerol, 0.4 M NaCl, 0.1 mM EDTA, and proteases inhibitors and incubated for 30 min on a rotating wheel. Extracts were centrifuged at high speed to remove nuclear debris; supernatants were aliquoted, quickly frozen in liquid nitrogen, and stored at Ϫ80°C until use. Protein concentration was measured by Bradford assay microtitration using the Coomassie Plus kit (Pierce) according to the manufacturer's instructions. Typical protein concentration obtained under these conditions was in the 3-6 g/l range.
EMSA-Complementary single-stranded oligonucleotides were annealed by incubation at a concentration of 2.5 g/l in STE buffer (100 mM NaCl, 10 mM Tris (pH 8.0), and 1 mM EDTA) at 80°C for 2 min. The mixture was then slowly cooled down to 4°C at a rate of 1°C/min in a thermal incubator (Perkin-Elmer). Annealed oligonucleotides were diluted to 25 ng/l in STE buffer and stored at Ϫ20°C until further use. 5Ј-End labeling of double-stranded oligonucleotides was performed by polynucleotide kinase reaction (New England Biolabs Inc.) using 25 ng of oligonucleotide and 30 Ci (1.1 MBq) of [␥-32 P]ATP (3000 Ci/mmol; Amersham Pharmacia Biotech). Labeled probes were purified by spin column exclusion chromatography (G-50, Amersham Pharmacia Biotech). For a typical EMSA experiment, 5 g of nuclear extract were diluted in buffer containing (final concentrations) 40 mM Hepes (pH 8.0), 50 mM KCl, 0.05% Nonidet P-40, 1% dithiothreitol, 10 g/ml poly(dI⅐dC), and 100 g/ml sheared salmon sperm DNA in a total volume of 20 l. Antibody (1 g) or unlabeled competitor DNA (25 ng) was added (when required), and the mixture was incubated for 20 min at room temperature. Labeled probe (0.25 ng) was added to the mixture, and the reaction was submitted to an additional 20-min incubation at room temperature. Samples were run on a 5% nondenaturing polyacrylamide gel in 0.5ϫ Tris borate/EDTA. Following electrophoresis, the gel was dried and autoradiographed. Anti-Sp1 (sc-059X) and anti-Egr-1 (sc-189X) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Transient Transfections and Luciferase Assays-HeLa cells were divided 48 h prior to transfection to generate 40 -60% confluence in 150-mm plates at the time of transfection. Culture medium was replaced 6 h prior to transfection, and cells were transfected by electroporation. Fifteen micrograms of test plasmid DNA (pGLSϪ29 to pGLSϪ1499) were combined with 1 g of pRLTK (Renilla reniformis luciferase under the control of a minimal herpes simplex virus thymidine kinase promoter; Promega) and introduced into 10 7 cells by electroporation (200 V, 900 microfarads; Easy CellJect, Eurogentec). Transfected cells were then plated in three wells of a six-well plate. After 16 h, the culture medium was replaced with fresh medium, and cells were collected and assayed for luciferase activities 42-44 h after transfection. Luciferase activities (P. pyralis and R. reniformis) were quantitated with the Dual-luciferase kit (Promega) in a luminometer (Berthold) following the manufacturers' instructions. The mean of Photinus luciferase activities measured in the three wells of the triplicates was plotted after adjusting for Renilla luciferase activities. Two to five independent transfections were performed for each experiment with similar results. For each assay, a representative experiment is displayed in the figures.

RESULTS
PMA Treatment Increases the SOD1 mRNA Level-To determine whether SOD1 gene expression can be induced by PMA treatment, Northern blot analysis was carried out. As shown in Fig. 1A, a major mRNA species of 0.7 kilobases was detected with the SOD1 cDNA probe and was constitutively expressed in HeLa cells. PMA treatment of HeLa cells induced a 2-fold increase in the SOD1 mRNA level, which could be detected as soon as 30 min following PMA addition and remained constant up to 24 h (Fig. 1B). As shown in Fig. 1C, this mRNA induction was followed by a 1.7-fold increase in the Cu,Zn-SOD protein level after 4 h of PMA treatment. Interestingly, in contrast with the observations obtained for the steady-state mRNA levels, the Cu,Zn-SOD protein level decreased to initial values after 8 h of PMA treatment, suggesting the existence of post-transcriptional and/or post-translational regulation mechanisms.
A PMA-responsive Element Is Located between Nucleotides Ϫ71 and Ϫ29 -To identify the most proximal element responsible for basal expression and PMA induction in the SOD1 5Ј-flanking region, we performed a luciferase assay using a series of constructs transfected in HeLa cells. These constructs carry different 5Ј-deletions of the SOD1 promoter linked to the P. pyralis luciferase gene. As shown in Fig. 2, constructs pGLSϪ1449 to pGLSϪ71 displayed basal promoter activity as well as PMA inducibility. In contrast, luciferase expression from transfected construct pGLSϪ29 (in which the regulatory sequences immediately upstream of the TATAA box have been deleted) was similar to that observed with the promoterless pGL2-Basic control vector and was barely detectable. It is noteworthy that both basal and PMA-induced promoter activities gradually decreased following the progressive removal of 5Ј-regulatory flanking sequences, indicating the existence of several cis-acting positive regulatory sequences spread along 1.5 kilobases of the SOD1 promoter proximal region. This observation led us to define the Ϫ71/ϩ1 region as the minimal sequence required to provide SOD1 promoter activity. Kinetics experiments showed that PMA induction could be detected as soon as 30 min following treatment for constructs pGLSϪ71 and pGLSϪ1499 (data not shown).
Sp1, Egr-1, and a Possible Sp1-like Protein Bind to the Ϫ60/ Ϫ38 Region-The Ϫ71/Ϫ29 region seems to be critical for both basal and PMA-induced expression of the SOD1 promoter. To further define the sequences involved in this regulation and to identify trans-acting factors bound to this region, we performed a series of band shift experiments using overlapping oligonucleotides covering this region. We used three different probes spanning nucleotides Ϫ71 to Ϫ51, Ϫ60 to Ϫ38, and Ϫ50 to Ϫ29, respectively. Incubation of the Ϫ71/Ϫ51 and Ϫ50/Ϫ29 probes with HeLa nuclear extracts (PMA-treated or not) did not reveal any specific retarded complex (data not shown). In contrast, incubation of the oligonucleotide corresponding to the Ϫ60/Ϫ38 sequence with nuclear extracts from untreated HeLa cells revealed two major shifted complexes (complexes I and III) (Fig.  3B). Incubation of the same probe with extracts from PMAtreated cells allowed the detection of a third complex (complex II) (Fig. 3B). Minor complexes were also inconsistently observed (complexes IV and V). Transcription factors containing zinc finger DNA-binding domains have been found to be involved in the regulation of several GC-rich proximal promoters (27). Because the Ϫ60/Ϫ38 probe contains 77% G or C nucleotides, we investigated the possible involvement of Sp1 and Egr family members in the complexes identified above. Thus, we carried out competition experiments using consensus binding sequences for Sp1 and Egr family members. As shown in Fig. 3, complexes I and III became barely detectable following incubation with an excess of Sp1 unlabeled competitor oligonucleotide, whereas complex II was not affected. On the other hand, the PMA-induced complex II was efficiently displaced by an Egr consensus oligonucleotide, whereas complexes I and III were not. Control experiments including consensus NF-B-and AP-1-binding sites did not reveal any change in electrophoretic profiles (data not shown). Altogether, these results suggest that complexes I and III contain two Sp1-related proteins, whereas complex II contains a member of the Egr family. Indeed, Fig. 3 shows that complex I was completely supershifted following incubation with an anti-Sp1 monoclonal antibody, whereas complex II and III were not. Conversely, complex II disappeared when the binding reaction was performed in the presence of an anti-Egr-1 antibody. The absence of supershift for this complex may be explained if the epitope recognized by the anti-Egr-1 antibody is part of (or located near) the Egr-1 DNA-binding site. Antibody binding would compete DNA binding, resulting in the disappearance of complex II rather than in a supershifted complex. Alternatively, antibody binding might result in a conformational change in Egr-1 structure, which would become unable to bind DNA. Altogether, these results allow the identification of Sp1 and Egr-1 as components of complexes I and II, respectively. Complex III is likely to contain an Sp1-related protein since this complex is efficiently competed by an excess of unlabeled Sp1 oligonucleotide competitor, although it is not supershifted by an anti-Sp1 monoclonal antibody. Thus, complex III may contain an Sp1 proteolytic fragment lacking the epitope recognized by the anti-Sp1 antibody or contains another possible Sp1-like protein. The minor complexes IV and V are likely to contain degradation products of Sp1-related proteins since they are efficiently competed by a consensus Sp1-binding site.
Mapping of Sp1-, Egr-1-, and Sp1-like Protein-binding Sites-Sequence analysis of the Ϫ60/Ϫ38 region did not allow the identification of any obvious Egr-1-binding site. To identify Sp1-and Egr-1-binding sites within the Ϫ60/Ϫ38 region, we used mutant oligonucleotides derived from the Ϫ60/Ϫ38 sequence in band shift assays. Nucleotides were substituted by groups of three, allowing complete coverage of the Ϫ60/Ϫ38 sequence using seven mutant oligonucleotides (Fig. 4A). The EMSA results displayed in Fig. 4B show that mutations between positions Ϫ59 and Ϫ48 disrupted all the complexes seen with the wild-type oligonucleotide. Oligonucleotide m(Ϫ47/ Ϫ45), carrying mutations on nucleotides Ϫ47, Ϫ46, and Ϫ45, bound Sp1 and Egr-1, but poorly bound the possible Sp1-like protein. Mutations in the 3Ј-region of the Ϫ60/Ϫ38 sequence had no effect on the recognition of the corresponding oligonucleotides. The identities of the different factors bound to mutant oligonucleotides have been assessed by competition and supershift assays with antibodies (data not shown). Thus, in this region, binding sites for Egr-1 and Sp1 are indistinguishable and cover positions Ϫ59 to Ϫ48, whereas a possible Sp1like factor binds to an overlapping sequence located between nucleotides Ϫ59 and Ϫ45. These results were confirmed using the Ϫ60/Ϫ38 oligonucleotide as a probe and various mutant oligonucleotides as competitors. As expected, oligonucleotides carrying mutations between positions Ϫ59 and Ϫ48 were unable to compete for Sp1 and Egr-1 binding (Fig. 4C). Conversely, competitors mutated at positions Ϫ59 to Ϫ45 did not compete for the binding of the possible Sp1-like factor, confirming assessment of binding site positions. Surprisingly, oligonucleotide m(Ϫ44/Ϫ42) did not completely compete the possible Sp1-like complex, suggesting that another unidentified protein may be present within this complex. FIG. 3. Gel shift analysis of nuclear factors bound to the ؊60/ ؊38 region of the SOD1 promoter. A, shown is the sequence of the Ϫ71/Ϫ30 region of the SOD1 promoter. The Ϫ60/Ϫ38 region used as a probe for gel shift assays is underlined. The TATAA box is indicated in italics. Sequences of Sp1 and Egr-1 oligonucleotides used as competitors. Binding sites are underlined. B, EMSA was performed using nuclear extracts from HeLa cells (PMA-treated or not) and a 32 P-labeled double-stranded Ϫ60/Ϫ38 oligonucleotide as a probe. Competition and supershift assays were carried out using Sp1 binding sequence, Egr binding sequence, anti-Sp1 or anti-Egr-1 antibodies.
The Sp1/Egr-1 Site Is Essential for Basal and Egr-1-induced SOD1 Proximal Promoter Expression-To test the functionality of the Egr-1 site, the pGLSϪ71 construct was transfected in HeLa cells together with an expression vector for Egr-1. The results shown in Fig. 5A indicate that the SOD1 proximal promoter-driven luciferase expression increased up to 4-fold when increasing amounts of Egr-1 expression vector were added. We then analyzed the response to Egr-1 of various mutant promoters. Mutant construct pGLSϪ71 m(Ϫ41/Ϫ39) is derived from oligonucleotide m(Ϫ41/Ϫ39), which binds Sp1, Egr-1, and the possible Sp1-like protein. Mutant construct pGLSϪ71 m(Ϫ47/Ϫ45) is derived from oligonucleotide m(Ϫ47/ Ϫ45), which efficiently binds Sp1 and Egr-1, but poorly binds the possible Sp1-like factor. Mutant construct pGLSϪ71 m(Ϫ50/Ϫ48) is derived from oligonucleotide m(Ϫ50/Ϫ48), which does not bind any of the three proteins. The results displayed in Fig. 5B show that all constructs but one, pGLSϪ71 m(Ϫ50/Ϫ48), displayed promoter activity and Egr-1 inducibility. Luciferase activity driven by an SOD1 proximal promoter mutated at positions Ϫ50 to Ϫ48 was almost abolished, both for constitutive and Egr-1-induced expression, confirming the crucial role of the Sp1/Egr-1 site. Interestingly, luciferase expression (both basal and Egr-1-induced) driven by construct pGLSϪ71 m(Ϫ47/Ϫ45) was comparable to that of the wild-type pGLSϪ71 control, suggesting that the possible Sp1-like factor(s) does not play a major role in SOD1 transcription in HeLa cells. We used the same mutant constructs to map the PMAinduced element within the SOD1 proximal promoter. Fig. 5C shows that, in a manner similar to the data described above, mutation of nucleotides Ϫ50 to Ϫ48 strongly reduced PMAinduced luciferase expression, whereas other mutants tested remained inducible by PMA. These data suggest that the PMAinduced element and the Sp1/Egr-1 site are identical. Thus, PMA-induced activation of an Egr family member would result in SOD1 proximal promoter induction via the Sp1/Egr-1-binding site.
The Sp1/Egr-1 Site Is a Target for WT1-Since numerous Egr-1 sites have been described to bind other Egr-related transcription factors such as the Wilms' tumor protein (27), we tested the ability of two splicing variants of WT1 to modulate transcription controlled by the SOD1 proximal promoter. HeLa cells were transiently transfected with pGLSϪ71 together with increasing amounts of expression vectors for either the WT1 splicing variant containing the tripeptide KTS or the WT1 variant lacking KTS. The results depicted in Fig. 6 show that, in contrast to what is observed in many systems, both splicing variants of WT1 transactivated the SOD1 proximal promoter. Excessive amounts of WT1 expression vectors were inhibitory, a "squelching" phenomenon often observed in other systems and thought to be an artifact linked to episomal expression in transient transfection assays (28). In accordance with the results obtained for Egr-1, mutations in the Ϫ50/Ϫ48 region completely abolished basal and WT1-induced expression, whereas other mutations (m(Ϫ47/Ϫ45) and m(Ϫ41/Ϫ39)) had no significant effect on WT1-induced expression (data not shown). These results suggest that the two WT1 splicing variants bind to the same sequence as Egr-1 in the SOD1 proximal promoter. However, since WT1 often induces transcriptional repression rather than activation through sites that respond to Egr-1 in a positive manner (27,29), we cannot exclude the possibility that the observed enhancer effect of WT1 overexpression may be indirect. DISCUSSION We report the analysis of the human Cu,Zn-SOD gene proximal promoter. Deletion analysis of the SOD1 gene promoter suggests the existence of several regulatory sequences spread along the 5Ј-flanking region of the SOD1 gene. Indeed, previous reports had allowed identification of several cis-acting sequences. Two AP-2 sites located between positions Ϫ134 and Ϫ127 and positions Ϫ118 and Ϫ111 are involved in the promoter response to Panax ginseng extracts (30). An Sp1 site (positions Ϫ104 to Ϫ89) and a CCAAT/enhancer-binding protein (C/EBP) site (positions Ϫ64 to Ϫ55) have also been described (31). Our data demonstrate that, although several regulatory sequences are present within 1.5 kilobases of 5Ј- upstream sequences, the SOD1 promoter region located between nucleotides Ϫ71 and ϩ1 is sufficient to provide both basal and PMA-induced promoter activities. Within this region, we identified a site located between nucleotides Ϫ59 and Ϫ48 that is able to bind Sp1 constitutively and Egr-1 following PMA treatment. This sequence can also mediate transactivation by two splicing variants of the Wilms' tumor protein WT1. Interestingly, the C/EBP-binding site (positions Ϫ64 to Ϫ55) described by Seo et al. (31) partially overlaps the Sp1/Egr-1 site (positions -59 to Ϫ48) described in this study, suggesting the existence of a complex interplay between different trans-acting factors for the transcriptional regulation of the SOD1 gene.  7. A, putative positions of Sp1-, Sp1-like protein-, and Egr-1binding sites on the SOD1 proximal promoter according to this study and to consensus sequences previously described (32). B, consensus binding sequences for Sp1 and Egr-1 according to the Transfac data base (32). Differences from the proposed sites on the SOD1 proximal promoter are indicated by boxes. Comparison of the Sp1/Egr-1 site with consensus Sp1-, Egr-1-, or WT1-binding sites (Fig. 7) revealed that this site does not fit consensus sequences described so far (32). In addition, in most other sequences recognized by both Sp1 and Egr-1, the binding sites of the two proteins overlap on six nucleotides, but are not identical (27). In our study, Sp1 and Egr-1 sites seem to be identical or at least do not seem to differ by more than two nucleotides since we could not find a mutation that affects binding of only one of the two factors. This non-canonical Sp1/Egr-1/WT1 binding sequence might represent the prototype of a novel binding site involved in transcriptional regulation by members of the zinc finger transcription factor family. Another sequence is recognized by a possible Sp1-like protein. Surprisingly, the Sp1-like site fits perfectly the Sp1 consensus sequence, but does not seem to be involved in SOD1 transcription regulation in HeLa cells since a mutation at this site does not lead to significant modification of transcriptional activity.
The apparent lack of phenotype in SOD1 null mice (15) indicates that Cu,Zn-SOD is probably not necessary for normal development and life, but rather is involved in the response to stress. Indeed, the only phenotype described for these mice is a partial defect in motor neuron regeneration following section. Interestingly, Egr-1 has been described as a key factor in response to mechanical stress in endothelial cells (33). Thus, Egr-1 regulation of the SOD1 gene provides a link between this key factor and Cu,Zn-SOD in response to mechanical stress in endothelial cells. Accordingly, a recent report has described an induction of Cu,Zn-SOD activity in human aortic endothelial cells in response to shear stress (34). This increase in activity was, at least in part, mediated by increased transcription of the SOD1 gene and was not observed in aortic smooth muscle cells. Although it was not demonstrated that this increase was mediated by Egr-1, we propose that Egr-1 regulation of SOD1 participates, via the antioxidant properties of Cu,Zn-SOD, in an atheroprotective pathway.
The function of the tumor suppressor protein WT1 in SOD1 gene regulation remains unknown. A connection between WT1 and apoptotic processes has been described in several reports. WT1 stabilizes p53 and inhibits its ability to induce apoptosis in response to various stimuli (35). Down-regulation of endogenous WT1 by antisense oligonucleotides induces apoptosis in myeloid leukemia cell lines (36). On the other hand, a proapoptotic role for WT1 has been reported: overexpression of WT1 induces apoptosis in an osteosarcoma cell line (37), and p53-independent programmed cell death was induced in U2OS and Saos-2 cells (38), HepG2 and Hep3B cells (39), and the myeloblastic leukemia M1 cell line (40) following WT1 induction. In most cases, the pro-apoptotic effect of WT1 seems to be mediated by the splicing variant that lacks the KTS tripeptide (38,39). In addition, WT1 has been described as able to downregulate transcription of the anti-apoptotic bcl-2 gene. Although reactive oxygen intermediates are not likely to be involved in all apoptotic processes, sympathetic neurons and neuronally differentiated PC12 cells undergo a reactive oxygen intermediate-dependent apoptosis when deprived of nerve growth factor. This phenomenon is prevented by injection of Cu,Zn-SOD (41). Taken together, these results suggest a function for the WT1 pathway of SOD1 regulation in the prevention of stress-induced apoptosis in neuronal cells.