Independent Repression of a GC-rich Housekeeping Gene by Sp1 and MAZ Involves the Same cis-Elements*

The transcription factors Sp1 and MAZ (Myc-associated zinc finger protein) contain several zinc finger motifs, and each functions as both a positive and a negative regulator of gene expression. In this study, we characterized the extremely GC-rich promoter of the human gene for MAZ, which is known as a housekeeping gene. Unique symmetrical motifs in the promoter region (nucleotides −383 to −334) were essential for the expression of the gene for MAZ, whereas an upstream silencer element (nucleotides −784 to −612) was found to act in a position-dependent but orientation-independent manner. Sp1 and MAZ bound to the same cis-elements in the GC-rich promoter, apparently sharing DNA-binding sites. The relative extent of binding of Sp1 and MAZ to these cis-elements corresponded to the extent of negative regulation of the expression of the gene for MAZ in various lines of cells. Furthermore, novel repressive domains in both Sp1 (amino acids 622–788) and MAZ (amino acids 127–292) were identified. Suppression by Sp1 and suppression by MAZ were independent phenomena; histone deacetylases were involved in the autorepression by MAZ itself, whereas DNA methyltransferase 1 was associated with suppression by Sp1. Our results indicate that both deacetylation and methylation might be involved in the regulation of expression of a single gene via the actions of different zinc finger proteins that bind to the same cis-elements.

Regulation of the expression of many genes is mediated by the binding of transcription factors to cis-elements in their promoter regions. The promoter regions of many eukaryotic genes contain GC-rich sequences (1) and some of the most widely distributed promoter elements are GC boxes and related motifs. The zinc finger proteins Sp1 and MAZ 1 (Myc-associated zinc finger protein) are transcription factors that bind to GCrich sequences, namely GGGCGG and GGGAGGG, respectively, to activate the expression of various target genes.
Sp1 was originally characterized as a ubiquitous transcription factor, consisting of 778 amino acids, that recognized GC-rich sequences in the early promoter of simian virus 40 (2,3). The DNA-binding domain of Sp1 consists of three contiguous C2H2-type zinc fingers (4). The amino-terminal region contains two serine-and threonine-rich domains and two glutamine-rich domains, which are essential for transcriptional activity (5). The carboxyl-terminal domain of Sp1 is involved in synergistic activation and interactions with other transcription factors. Sp1 is considered to be a constitutively expressed transcription factor and has been implicated in the regulation of a wide variety of housekeeping genes, tissue-specific genes, and genes involved in the regulation of growth (6). Sp1 is a phosphorylated (7) and highly glycosylated protein (8). It interacts with many factors, such as the TATA box-binding protein, which is a major component of the general transcription machinery, and the TATA box-binding protein-associated factors dTAFII110 (9), hTAFII130 (10), and hTAFII55 (11). Other proteins, such as transcription factor YY1 (12,13), E2F (14,15), and p300 (16,17), have also been reported to associate with Sp1. Sp1-null mice embryos exhibited severely retarded growth and died within 10 days (18), after displaying a wide range of abnormalities. Some of the embryos appeared as an undifferentiated mass of cells, whereas others had all the typical hallmarks of early embryogenesis, such as a developing heart, eyes, optic vesicles, somites, erythroid cells, and extra-embryonic tissues (18). Thus, it is likely that Sp1 is essential for the differentiation of embryonal stem cells after day 10 of development.
MAZ was first identified as a transcription factor that bound to a GA box (GGGAGGG) at the ME1a1 site of the c-myc promoter and to the CT element of the c-myc gene (19 -21). It is a zinc finger protein with six C2H2-type zinc fingers at the carboxyl terminus, a proline-rich region, and three alanine repeats. It is expressed ubiquitously, albeit at different levels in different human tissues (22). It can regulate the expression of numerous genes, such as c-myc (19,20,23,24), genes for insulin I and II (25), the gene for CD4 (26), the gene for the serotonin receptor (27), and the gene for nitric-oxide synthase (28). MAZ might be involved in the termination of transcription by interrupting elongation by RNA polymerase II (29).
The promoter region of the gene for MAZ has features typical of the promoter of a housekeeping gene, namely a high GϩC content, a high frequency of CpG (where p stands for "phosphoric residue") dinucleotides, the absence of canonical TATA boxes, and multiple sites for initiation of transcription (30,31). Moreover, the gene is ubiquitously expressed in human tissues (22).
We have attempted to clarify some aspects of the relationship between the factors that bind to GC-rich cis-elements and the promoters of housekeeping genes with a high GϩC content. A previous study showed that Sp1 binds to GC-rich DNA sequences in nucleosomes (32). Moreover, the large coactivator complex known as CRSP (cofactor required for activation of Sp1) stimulates Sp1-mediated transcription (33). Both Sp1 and MAZ can exert positive and negative control over the expression of target genes. Thus, regulation by individual DNA-binding factors seems to be coordinated via recruitment of other factors that participate in the regulated expression of target genes and via recognition of the modification of nucleotide sequences, for example, by methylation or demethylation and acetylation or deacetylation (34 -37). The binding affinities of transcription factors for individual target sequences are likely to be essential parameters in the regulation of gene expression, together with the recruitment of related factors.
We demonstrate here a possible mechanism for regulation of the expression of the human gene for MAZ. The mechanism involves the recruitment of different repressors by two different DNA-binding factors, Sp1 and MAZ, that interact with the same cis-elements. Our results indicate that deacetylation and methylation might be involved in the regulation of a single gene via the binding of different zinc finger proteins.

MATERIALS AND METHODS
Plasmids-A series of DNA fragments from the MAZ promoter was excised with appropriate restriction enzymes. Each fragment was filled in and inserted into the HindIII site of pSV00CAT (38), via a HindIII linker, to generate pMAZCAT1, pMAZCAT2, pMAZCAT3, pMAZCAT4, and pMAZCAT5, respectively. Internal deletion mutants of the MAZ promoter were created by amplification by the polymerase chain reaction, ligation of the appropriate DNA fragments, and insertion into the HindIII site of pSV00CAT to generate pMAZCAT2-d, pMAZCAT3-wt, pMAZCAT3-⌬I, pMAZCAT3-⌬II, and pMAZCAT3-⌬III, respectively. Mutant forms of pMAZCAT3 were further generated by mutation of dinucleotides (AA to GG; TT to GG (see Fig. 2)) to generate a series of mutants, pMAZCAT3-f1-pMAZCAT3-f11. Mutations in the putative Sp1-binding sites and putative MAZ-binding sites in pMAZCAT3-wt were generated by converting the GC-rich motif GGGCGG to GGTTGG and the GC-rich motif GGGAGGG to GGTATGG (39,40). Amplification by polymerase chain reaction and ligation into the HindIII site of pSV00CAT generated pMAZCAT3-m1-pMAZCAT3-m8. pCMV-MAZ was constructed as described previously (22). pCMV-Sp1 and pCMV-DNMT1 were provided by R. Chiu and R. Raenish, respectively.
Cell Culture, Transfection, and Assay of Chroramphenicol Acetyltransferase (CAT) Activity-HeLa cells, 293 cells, and NIH3T3 cells were grown in Dulbecco's modified Eagle's medium that had been supplemented with 10% fetal bovine serum (Life Technologies, Inc.). NCI-H460 cells were grown in RPMI 1640 medium that had been supplemented with 10% fetal bovine serum. Cells were treated with tricostatin A (TSA) at a final concentration of 100 ng/ml and with 5-azacytidine at a final concentration of 1 mM. Cells were transfected with plasmid DNA using the FuGENE TM 6 transfection reagent (Roche Molecular Biochemicals) according to the protocol provided by the manufacturer. All plasmids were purified by ultracentrifugation before transfection, as described previously (41). Assays of CAT activity were performed as described elsewhere (21).
Gel Shift Assay-DNA probes were radiolabeled at their 5Ј-ends with polynucleotide kinase (New England BioLabs, Inc., Beverly, MA) and [␥-32 P]ATP. The DNA probes designated M, S, and MS corresponded to DNA fragments between nt Ϫ313 and Ϫ284, nt Ϫ232 and Ϫ216, and nt Ϫ151 and Ϫ137. The binding reaction was performed in 30 l of a buffer that contained 20 mM Tris-HCl (pH 7.5), 2 mM MgCl, 0.5 mM EDTA, 10% glycerol, 0.5 mM dithiothreitol, 25 mM NaCl, 1 g of poly(dI-dC), and an extract of HeLa cells or purified glutathione S-transferase (GST) fusion proteins. Reactions were incubated at 4°C for 40 min after addition of the labeled DNA probe. The incubation was continued for 30 min at room temperature after the addition of appropriate antibodies. Products of reactions were loaded onto a 5% non-denaturing polyacrylamide gel in 0.5ϫ TBE buffer (1ϫ TBE: 45 mM Tris-borate, 1 mM EDTA). Electrophoresis was performed at 100 V for 4 -6 h at 4°C.
Immunoprecipitation and Assay of Histone Deacetylase (HDAC) Activity-HeLa cells were cultured with or without TSA (100 ng/ml) for 48 h, and then proteins in cell extracts were immunoprecipitated with antibodies specific for HDACs (Santa Cruz Biotechnology, Santa Cruz, CA) or DNA methyltransferase 1 (DNMT1) (New England BioLabs, Inc.). Cell extracts were subjected to assays of HDAC activity using a histone deacetylase assay kit (Upstate Biotechnology, Lake Placid, NY) in accordance with the instructions from the manufacturer.

Unique Symmetric Elements in the Minimal MAZ Promoter
Are Essential for Transcriptional Activity-The promoter region of the human gene for MAZ has an extremely high GϩC content, namely 88.4%. Various GC-rich elements are present in the promoter region, including consensus Sp1-binding sites (GGGCGG) and consensus MAZ-binding sites (GGGAGGG) (Fig. 1). Some of these sites overlap one another. Assays of CAT activity using constructs with various deletions in the MAZ promoter demonstrated that the minimal promoter activity was localized between nt Ϫ383 and ϩ259 ( Fig. 2A). Internal deletion of the region between nt Ϫ383 and Ϫ248 (pMAZCAT2-d) resulted in a decrease in promoter activity. This result suggested that the region from nt Ϫ383 to Ϫ248 might be critical for minimal promoter activity. We next attempted to identify the elements that were essential for minimal promoter activity. Promoter activity was reduced with the construct that lacked the region between nt Ϫ383 and Ϫ334, whereas constructs with internal deletion of the region between nt Ϫ334 and Ϫ279 or between nt Ϫ279 and Ϫ248 did not have reduced promoter activity (Fig. 2B). Thus, the region from nt Ϫ383 to Ϫ334 was essential for the promoter activity. Four symmetric elements, namely two CAAC and two CTTC elements, were present in this region (Fig. 2C). CAAC and CTTC elements have also been found in other promoters, such as the promoter of the gene for the ␣-myosin heavy chain (42), the gene for hydroxymethylbilane synthase (43), the gene for myosin light chain 2 (44), and the gene for lactoferrin (45). Some of these sites have been shown to contribute to the activation of promoter activity. We used a series of CAT constructs with mutations in these elements to investigate whether these putative elements might activate the promoter of the gene for MAZ (Fig. 2D). The results of CAT assays demonstrated that promoter activity was reduced when each of the four elements was mutated (pMAZCAT3-f1, pMAZCAT3-f2, pMAZCAT3-f3, and pMAZCAT3-f4). The promoter activity was reduced still further when two of the four elements were mutated simultaneously (pMAZCAT3-f5, pMAZCAT3-f6, pMAZCAT3-f7, pMAZCAT3-f8, pMAZCAT3-f9, and pMAZCAT3-f10). The CAT activity fell to about 20% of that of the wild type when all of the four elements were mutated (pMAZCAT3-f11). These results indicated that the four symmetric elements were essential for the activity of the MAZ promoter.
Sp1 and MAZ Recognize the Same cis-Elements in the MAZ Promoter-To determine whether Sp1 and/or MAZ could bind to the various putative binding sites for both Sp1 and MAZ, we performed gel shift assays using extracts of HeLa cells and DNA probes derived from the MAZ promoter. The SM probe, nt Ϫ313 to Ϫ284, contained one putative Sp1-binding site and one putative MAZ-binding site, and these two sites partially overlapped. The M probe, nt Ϫ232 to Ϫ216, contained one putative MAZ-binding site; and the S probe, nt Ϫ153 to Ϫ137, contained one putative Sp1-binding site (Fig. 3A). We detected two prominent DNA-protein complexes with the SM probe that contained the overlapping binding sites for Sp1 and MAZ. The rapidly migrating band was more intense than the slowly migrating band (Fig. 3B, left panel). The retarded bands corresponding to B1 and B2 were shifted even further upon addition of antibodies against Sp1 and MAZ (Fig. 3B, lanes 2 and 4). Control antibodies did not affect the mobility of the DNAprotein complexes (Fig. 3B, lane 6). These results indicated that both Sp1 and MAZ specifically recognized the overlapping sites in the same cis-element. To examine the DNA binding specificity of Sp1 and MAZ, we used the purified GST-Sp1 and GST-MAZ in the assays. A DNA-protein complex was detected using GST-Sp1, and a supershifted band was detected in the presence of antibodies specific for Sp1 but not in the presence of antibodies specific for MAZ. Similarly, a DNA-protein complex was detected using GST-MAZ, and a supershifted band was detected in the presence of antibodies to MAZ but not in the presence of antibodies to Sp1 (Fig. 3B, lanes 7-12). Supershifted bands were also detected in the presence of antibodies against MAZ or Sp1 when we used the M probe or the S probe (Fig. 3C, lanes 2, 4, 8, and 10). The results indicated that both Sp1 and MAZ interacted with the putative MAZ-binding sites and the putative Sp1-binding sites in all three probes that we used. We also examined other GC-rich cis-elements in the MAZ promoter for recognition by Sp1 and MAZ. We found that Sp1 and MAZ bound to the same GC-rich cis-elements in other regions of the MAZ promoter (data not shown). Thus, it was clear that Sp1 and MAZ bound to the same GC-rich cis-elements in the MAZ promoter.
Both Sp1 and MAZ Repress the Activity of the MAZ Promoter through the Various cis-Elements-We next focused on the effects of Sp1 and MAZ in transactivation of the gene for MAZ. The reporter constructs were used to transfect HeLa cells in the presence or absence of an Sp1 or a MAZ expression vector. The promoter activity was inhibited significantly in the presence of ectopically expressed Sp1 (Fig. 4A) and MAZ (Fig. 4B), whereas ectopic expression of Sp1 and of MAZ had no effect on the transcription of pRSVCAT (19), the control plasmid. These results indicated that both Sp1 and MAZ repressed transcription of the gene for MAZ.
The binding sites for Sp1 and MAZ in the minimal promoter region (-303 to ϩ3) were mutated in an attempt to identify the cis-elements that were involved in the negative regulation FIG. 2. A symmetric element in the minimal MAZ promoter is essential for the promoter activity of the gene for MAZ. A, the minimal basal promoter of the gene for MAZ was located in the region between nt Ϫ383 and ϩ259. A summary is shown of the human MAZ-CAT deletion constructs and corresponding CAT activities. Numbering is relative to the major site of initiation of transcription (ϩ1). CAT, gene for CAT. CAT fusion plasmids were used to transfect HeLa cells, and CAT activity was measured as described under "Materials and Methods." Promoter activities of MAZ-CAT fusion genes are expressed relative to the activity of pMAZCAT1, which was taken arbitrarily as 1.0. All values in this and other figures are the averages of results from at least three experiments, and the standard deviation for each value is indicated. B, the region between nt Ϫ383 and Ϫ334 is critical for the promoter activity of the MAZ gene. Promoter activities of MAZ-CAT fusion genes are expressed relative to the activity of pMAZCAT3, which was taken arbitrarily as 1.0. C, unique symmetric elements were present in the region between nt Ϫ383 and Ϫ334. Boxes indicate four dinucleotide repeats. D, the symmetric elements are essential for the promoter activity of the gene for MAZ. The wild type CAAC and CTTC elements and the mutated element CGGC were examined for their effects on the promoter activity of the gene for MAZ. Promoter activities of MAZ-CAT fusion genes are expressed relative to the activity of pMAZCAT3, which was taken arbitrarily as 1.0. (Fig. 5). In the presence of ectopically expressed Sp1 or MAZ, repression of the MAZ promoter was detected in the presence of mutations in the region between nt Ϫ383 and Ϫ248 (pMAZCAT3-m1 and pMAZCAT3-m4). We then mutated other sites (pMAZCAT3-m2, pMAZCAT3-m3, pMAZCAT3-m5, pMAZCAT3-m6, and pMAZCAT3-m7), and again we observed reduced transcriptional activity. All the mutated constructs mentioned above contained wild type binding sites for Sp1 and/or MAZ. Thus, those binding sites for Sp1 and/or MAZ might still have been active in the negative regulation of the MAZ promoter. Our results indicated that most, if not all, of the binding sites for Sp1 and MAZ were involved in negative regulation of the expression of the gene for MAZ. This possibility was confirmed by studies of promoter activity with pMAZCAT3-m8, in which all the binding sites for both Sp1 and MAZ had been mutated. No repression by Sp1 or by MAZ was observed with pMAZCAT3-m8. These results strongly sug-gested that repression by Sp1 and/or MAZ was mediated by the DNA-binding sites for Sp1 and MAZ and that most or all of these sites were involved in the repressive activity.
Independent Repression by Sp1 and by MAZ Is Mediated by Their Respective Repression Domains-Sp1 and MAZ repressed the expression of the gene for MAZ by binding to the same cis-elements. Therefore, we next asked whether repression by Sp1 and by MAZ might be related. The results of a yeast two-hybrid assay and immunoprecipitation-Western blotting analysis showed that Sp1 did not interact with MAZ (data not shown), indicating that repression by Sp1 and repression by MAZ were independent.
A series of Sp1 and MAZ expression plasmids was constructed to identify the domains responsible for repression. These plasmids were used to cotransfect HeLa cells in combination with the reporter construct (pCATMAZ1), and then CAT assays were performed (Fig. 6, A and B). Repression of CAT activity was observed with constructs that did not encode domains in the amino-terminal region of Sp1 (amino acid positions 1-503; Fig. 6A). These results suggested that the aminoterminal region of Sp1 was not involved in repression. The promoter activity was released from repression when Sp1 without the carboxyl-terminal region (⌬622-C) was expressed. Moreover, repression of the promoter activity was diminished when Sp1 was expressed without the zinc finger domain (⌬531-605) that is essential for binding to DNA. Taken together, the results indicated that the carboxyl-terminal region of Sp1 (amino acids 622-778) was responsible for the repression and that the repression was also dependent on the DNA binding ability of Sp1.
The promoter activity was partially repressed when MAZ without amino acids 54 -195 (⌬54 -195) was expressed, and the activity of the promoter was completely released from repression when MAZ without amino acids 127-292 (⌬127-292) was expressed. Thus, it appeared that the region between amino acids 127 and 292 was responsible for repression of the expression of the gene for MAZ. Moreover, repression of the promoter was reduced when a mutant form of MAZ was expressed without the five zinc fingers in the carboxyl-terminal region (⌬317-441), which was essential for DNA binding activity. Taken together, the results demonstrated that amino acids 127-292 of MAZ were responsible for autorepression and that autorepression was also dependent on the DNA binding activity of MAZ. We concluded that independent repression by Sp1 and by MAZ was mediated by the repression domains of each protein and that the DNA binding activities of these zinc finger proteins were also essential for repression.
Recruitment of Histone Deacetylases by MAZ-HDACs are known to act as repressors in the regulation of the expression of many genes. We attempted to determine whether histone deacetylases might be involved in repression of the gene for MAZ. HeLa cells were transfected with pCMV-MAZ or pCMV-HDAC1 in the presence and absence of TSA, a specific inhibitor of histone deacetylases. We then monitored the CAT activity due to a reporter plasmid, pMAZCAT1, with which the cells had been cotransfected. Ecpotic expression of HDAC1 repressed the activity of the MAZ promoter, and such repression was overcome in the presence of TSA (Fig. 7A), indicating that histone deacetylases might be involved in repression by MAZ. This possibility was confirmed by measurement of the HDAC activity of proteins that were recruited by MAZ. The HDAC activity of a MAZ-specific immunoprecipitate was more than five times higher than that of the complex that was immunoprecipitated by the control IgG, and the activity of the former complex was repressed in the presence of TSA (Fig. 7B). We performed immunoprecipitation and Western blotting analysis using nuclear extracts from HeLa cells to determine whether histone deacetylases were included in the complex of proteins recruited by MAZ. Western blotting analysis indicated the presence of MAZ in immunoprecipitates of extracts obtained with antibodies specific for HDAC1, HDAC2, and HDAC3 (Fig.  7C). The proteins in the same extracts were also immunoprecipitated by antibodies specific for MAZ, and all three kinds of histone deacetylase were detected (Fig. 7C). These results implied that MAZ recruited proteins that included HDAC1, HDAC2, and HDAC3.
Association of DNMT1 with Repression by Sp1-HeLa cells were transfected with pCMV-Sp1 and pMAZCAT1 in the presence or absence of pCMV-HDAC1 and TSA, respectively. The results of CAT assays revealed that repression by Sp1 was insensitive to TSA and that ectopic expression of HDAC1 had no effect on repression by Sp1 (Fig. 8A), suggesting that repression by Sp1 might be HDAC-independent. Methylation is known to be important in the regulation of gene expression. Thus we examined whether methylation might be involved in repression by Sp1. HeLa cells were transfected with pCMV-Sp1 (or just with the reporter) in the presence or absence of 5-azacytidine, a specific inhibitor of methylation. Repression by Sp1 was released in the presence of 5-azacytidine using the wild type reporter but not the mutant reporter (pMAZCAT3-m8) and the control reporter, pRSVCAT (Fig. 8B). Furthermore, the forced expression of DNMT1 enhanced the repression of transcription by Sp1, whereas treatment with 5-azacytidine reversed the repression due to Sp1 and DNMT1 (Fig. 8C). We performed immunoprecipitation and Western blotting analysis using nuclear extracts from HeLa cells, and the results showed that DNMT1 was included in the complex of Sp1 and vice versa (Fig. 8D). Taken together, these results suggest that DNMT1 might be involved in the repression mediated by Sp1.

DISCUSSION
The promoter regions of human housekeeping genes are usually GC-rich, and, by definition, these genes are expressed ubiquitously, as is, for example, the human gene for MAZ (22,30). Many GC-rich cis-elements can be found in the promoters of housekeeping genes, and they might be expected to regulate the transcription of various genes. In this study, we analyzed the GC-rich promoter of the human gene for MAZ in an attempt to identify the role of GC-rich cis-elements in the regulation of transcription of this gene.
The minimal MAZ promoter was located between nt Ϫ383 and ϩ259 (Fig. 2A). We showed that a 135-base pair sequence, from nt Ϫ383 to Ϫ248 in the minimal promoter region of the gene for MAZ was associated with the promoter activity. Further studies indicated that the region from nt Ϫ383 to Ϫ334 was critical for the promoter activity (Fig. 2B). The GϩC content of this region is relatively low, and there are two CAAC elements and two CTTC elements within this region (Fig. 2C). The region containing these four elements is 33 base pairs long, with an average GϩC content of only 49%, the lowest GϩC content in the extremely GC-rich promoter region of the gene for MAZ. It has been reported that, in some promoters, the proximal upstream region is extremely GC-rich, whereas the distal region is AT-rich (46 -49). It has also been reported that a stretch of GC-rich sequences is followed by AT-rich sequences in some promoters (49). A specific cis-element in the promoter region of the c-myc gene is localized in an AT-rich domain that is flanked by GC-rich sequences (49). The cited studies suggest that relatively AT-rich elements in extremely GC-rich sequences might be recognition sites for transcription factors that are associated with the initiation of transcription (49). To determine whether these motifs are critical for the activity of the promoter of the gene for MAZ, we examined a series of constructs with mutations in these motifs. As shown in Fig. 2D, two symmetric CAAC elements and two CTTC symmetric elements were required for basal transcriptional activity, and the contribution of each element to the total transcriptional activity was lower than that of all the elements together. Both CAAC and CTTC elements have been found in the promoters of other genes, such as the gene for the ␣-myosin heavy chain, the gene for hydroxymethylbilane synthase, the gene for myosin light chain 2, and the gene for lactoferrin, and some of these sites have been shown to be important for promoter activity (42)(43)(44)(45). The factors that bind to the CAAC and/or CTTC elements in the minimal MAZ promoter remain to be identified.
The consensus sequence of MAZ-binding sites is very similar A, repression by Sp1 was independent of HDAC1. Transfections and CAT assays were performed using HeLa cells in the presence and absence of pCMV-Sp1 and pCMV-HDAC1. The cells were harvested after a 48-h treatment with TSA. Promoter activities of MAZ-CAT fusion genes are expressed relative to the activity of pMAZCAT3-wt in the absence of pCMV-Sp1, which was taken arbitrarily as 1.0. B, repression by Sp1 was sensitive to 5-azacytidine (5-aza). HeLa cells were transfected with pMAZCAT3-wt, pMAZCAT3-m8, and the control, pRSVCAT, and then stable clones were treated with 5-azacytidine for 72 h before assays of CAT activity. C, transfected HeLa cells with pMAZCAT3-wt were transfected with pCMV-Sp1 or pCMV-DNMT1 and incubated with or without 5-azacytidine for 72 h before assays of CAT activity. D, immunoprecipitation-Western blotting analysis. Proteins in extracts of HeLa cells were immunoprecipitated with antibodies against Sp1 or DNMT1, and then Western blotting analysis was performed. Mouse IgG was used as the negative control.
to that of Sp1-binding sites. The GC-rich minimal promoter of the gene for MAZ contains multiple binding sites for Sp1 and MAZ. We found that Sp1 bound to consensus Sp1-binding sites as well as to consensus MAZ-binding sites. Similarly, MAZ bound to the consensus binding sites for both MAZ and Sp1 (Fig. 3). The results of our gel shift assays indicated that both Sp1 and MAZ recognized the same cis-elements in the MAZ promoter. It has been reported that a GC-rich motif in the c-myc promoter region is a high affinity binding site for both MAZ and Sp1 (50). It has also been reported that Sp1 binds to a series of GC-rich nucleotide sequences as well as to the consensus Sp1-binding site (51). The fact that MAZ and Sp1 shared binding sites indicates that the regulatory activity of some GC-rich cis-elements is consistent with cooperative interactions by multiple transcription factors, such as zinc finger proteins, with the same or overlapping cis-elements.
The binding of both Sp1 and MAZ to the same cis-elements in the promoter region of the gene for MAZ might regulate transcription of the gene. Both Sp1 and MAZ suppressed transcription from the MAZ promoter (Fig. 4). There are seven consensus binding sites for Sp1 and nine consensus binding sites for MAZ in the minimal promoter region of the gene for MAZ. We tried to identify the cis-elements that are involved in repression of the transcription of the gene for MAZ, and we found that the extent of repression by Sp1 and by MAZ was reduced only when all of the consensus binding sites for Sp1 and MAZ had been mutated (pMAZCAT3-m8; Fig. 5). However, we did not detect enhanced expression of the mutated construct (pMAZCAT3-m8), as compared with that of the wild type construct, even when the possible consensus GC-rich motifs in the promoter regions of the MAZ gene were mutated (Fig. 5). We do not know the exact reason why the overexpression of Sp1 or MAZ stimulated the expression of pMAZCAT3-m8. One possible explanation is that the weak binding sites of Sp1 or MAZ are still present in pMAZCAT3-m8, which might be the other GC-rich sequences in the promoter region, and are functional for the residual activity of repression. In fact, it has been reported that Sp1 or MAZ binds other GC-rich elements besides the consensus motifs (27,51). Further studies are required for the identification of other motifs for Sp1 and MAZ. Alternatively, we cannot rule out the possibility that the overexpression of Sp1 and MAZ might titrate the coactivators or general transcription factors and result in the repression of the MAZ promoter. Further studies are required to answer these questions.
The activity of the promoter was repressed when any of the wild type binding sites remained (pMAZCAT3-m1-pMAZCAT3-m7; Fig. 5), indicating that almost all of the ciselements were involved in repression. Autorepression of the gene for MAZ by MAZ itself also indicates that negative feedback might possibly be involved in the control of the expression of housekeeping genes. Both the multiple GC-rich cis-elements and the upstream silencer element were involved in the negative regulation of the gene for MAZ, indicating that suppression of transcription of this gene is the major basal regulatory mechanism that controls its expression.
Both Sp1 and MAZ repressed the activity of the gene for MAZ through binding to the same cis-elements. We tried to determine whether the repression by Sp1 and by MAZ might be linked, but the results failed to reveal any interaction between Sp1 and MAZ (data not shown). Repression by Sp1 and repression by MAZ were independent phenomena, even though both involved the same GC-rich cis-elements. We identified novel repressive domains in both Sp1 and MAZ. The carboxyl-terminal region of Sp1 (amino acids 622-778) and amino acids 127-292 of MAZ were responsible for the respective repressive activities (Fig. 6, A and B). Moreover, repression was also dependent on the zinc finger domains of both Sp1 and MAZ, which were essential for binding to DNA (Fig. 6, A and B). It is possible that Sp1 and MAZ might bind to cis-elements through their zinc finger motifs, recruiting other factors through their repression domains.
Histone deacetylases act negatively to regulate the expression of many genes (52)(53)(54). Therefore, we examined whether histone deacetylases might be involved in repression of the gene for MAZ. HeLa cells were transfected with pCMV-Sp1 or pCMV-MAZ in the presence and absence of TSA, a specific inhibitor of histone deacetylases. Only repression by MAZ was released in the presence of TSA, whereas the repression by Sp1 was insensitive to treatment with TSA (Figs. 7A and 8A). Thus, it appears that histone deacetylases are involved in repression by MAZ. We confirmed this possibility by measuring HDAC activity of immunoprecipitated complexes that contained MAZ. The HDAC activity of complexes was about five times higher than that of control immunoprecipitates, and the HDAC activity of the former complexes was repressed in the presence of TSA (Fig. 7B). Immunoprecipitation and Western blotting analysis using nuclear extracts from HeLa cells indicated that MAZ could recruit proteins that included HDAC1, HDAC2, and HDAC3 to form a multiple protein complex (Fig. 7C).
We found that the action of Sp1 was insensitive to TSA and that HDAC1 had no effect on repression by Sp1 (Fig. 8A). Thus, repression mediated by Sp1 appeared to be independent of HDACs. It has been reported that methylation plays an important role in the suppression of transcription, and the interaction of Sp1 with MeCP2 has also been reported (55). We examined whether methylation might be involved in repression of the gene for MAZ and found that repression by Sp1 was sensitive to 5-azacytidine, a specific inhibitor of methylation (Fig.  8B). The ectopic expression of DNMT1 enhanced repression by Sp1, whereas 5-azacytidine reversed the repression induced by Sp1 and DNMT1 (Fig. 8C). Furthermore, it is highly possible that DNMT1 is recruited by Sp1 (Fig. 8D). Therefore, DNMT1 appeared to play a role in the repression mediated by Sp1.
We have demonstrated here a possible mechanism for the down-regulation and autorepression of a human housekeeping gene, namely the gene for MAZ, through the recruitment of different repressors by two different DNA-binding proteins, Sp1 and MAZ, which interact with the same cis-elements (Fig.  9). Our data suggest that different levels of suppression of the transcription of this housekeeping gene might be responsible for the different levels of expression of the gene in different tissues. Moreover, deacetylation and methylation appear to play distinct roles in the regulation of a single gene, namely the human gene for MAZ, in a process that is mediated by different DNA-binding transcription factors.