Genomic Organization and Expression of a Human Gene for Myc-associated Zinc Finger Protein (MAZ)*

We have cloned and characterized the genomic structure of the human gene for Myc-associated zinc finger protein (MAZ), which is located on chromosome 16p11.2. This gene is transcribed as an mRNA of 2.7 kilobases (kb) that encodes a 60-kDa MAZ protein. A 40-kb cosmid clone was isolated that includes the promoter, five exons, four introns, and one 3′-untranslated region. All exon-intron junction sequences conform to the GT/AG rule. The promoter region has features typical of a housekeeping gene: a high G + C content (88.4%); a high frequency of CpG dinucleotides, in particular within the region 0.5 kb upstream of the site of initiation of translation; and the absence of canonical TATA and CAAT boxes. An S1 nuclease protection assay demonstrated the presence of multiple sites for initiation of transcription around a site 174 nucleotides (nt) upstream of the ATG codon and such expression was reflected by the promoter activity of a MAZ promoter/CAT (chloramphenicol acetyltransferase) reporter gene.Cis-acting positive and negative elements controlling basal transcription of the human MAZ gene were found from nucleotides (nt) −383 to −248 and nt −2500 to −948. Moreover, positive and negative autoregulatory elements were also identified in the regions from nt −248 to −189 and from nt −383 to −248 after co-transfection of HeLa cells with plasmids that carried theMAZ promoter/CAT construct and the MAZ-expression vector. Our results indicate that the 5′-end flanking sequences are responsible for the promoter activities of the MAZ gene.

The c-myc protooncogene is a member of a family of genes that encode DNA sequence-specific transcription factors with basic, helix-loop-helix, and leucine zipper domains. The Myc protein binds to DNA as heterodimers with a related polypeptide, Max (1)(2)(3)(4). Appropriate regulation of expression of the human c-myc gene is necessary for the proliferation and differentiation of cells and for progression of the cell cycle, and deregulation of the expression of c-myc is associated with tumorigenesis and apoptosis (1,4). Regulation of the expression of the human c-myc gene occurs at multiple levels, which include the initiation, the termination, and the attenuation of transcription (2,4). In proliferating cells, the initiation of transcription of the c-myc gene is controlled by two major promoters, P1 and P2, and the RNA initiated from the P2 promoter accounts for 80 -90% of the total RNA initiated from the P0, P1, and P2 promoters (4,5). Initiation of transcription from the P2 promoter requires at least three cis-elements: ME1a2, E2F, and ME1a1 (6,7). Several transcription factors, including Sp1 (5,8), the Myc-associated zinc finger protein (MAZ) 1 (9,10), Pur-1 (11), and E2F (12) bind to these elements in vitro and in vivo.
The MAZ protein was identified as a transcription factor that binds to a GA box (GGGAGGG) at the ME1a1 site, to the attenuator region of P2 within the first exon of the c-myc gene, and to a related sequence that is involved in the termination of transcription of the gene for complement 2 (C2) (5,9). Kennedy and Rutter (11) identified the Pur-1 protein as a GAGA box binding factor that binds to rat genes for insulin I and II and to the human gene for islet amyloid polypeptide (11). We recently reported the isolation of a cDNA clone for a member of the family of MAZ proteins in human islet cells (13). MAZ protein plays a role in the control of the initiation of transcription of genes for the adenovirus major late protein (14), CD4 (15), the serotonin receptor (16), and hematopoietic transcription factor (17), as well as in the termination of transcription between the closely spaced human genes for complement (8) and in the termination of transcription of the introns of the mouse gene for IgM-D (8). Therefore, MAZ appears to be a transcription factor with a dual role in the initiation and termination of transcription. We showed previously that MAZ is essential for the ME1a1-mediated expression of the c-myc gene during the neuroectodermal differentiation of P19 cells (18) and for the nuclease-hypersensitive element-mediated transcription of the c-myc gene in islet ␤-cells (13).
To gain a better understanding of the regulation of expression of MAZ, of the splicing mechanism, of the differential polyadenylation and of the potential interactions of MAZ with other factors, we isolated a human genomic gene for MAZ from cosmid and YAC libraries. We characterized the genomic structure of the gene for MAZ protein and identified regulatory elements in 5Ј-end flanking sequences that are involved in basal transcription and in the autoregulation of the gene for MAZ by the MAZ protein itself.
Screening of a Library of Human Genomic DNA-A cosmid library (20) was constructed from the genomic DNA of the HLA-homologous B-lymphoblastoid cell line AKIBA (A24, Bw52, Dw12, DQw1, and Cp63), which had been partially digested with Sau3AI, with subsequent ligation of fragments to the cosmid vector pWE15 (Stratagene, La Jolla, CA). The library was screened by colony hybridization with 1.8-and 0.7-kb EcoRI fragments of MAZ cDNA from pCMVMAZ (13) as probes (21). Filters were prehybridized at 68°C for 30 min in a solution that contained 6 ϫ SSC (1 ϫ SSC: 150 mM sodium chloride, 15 mM sodium citrate, pH 7.0), 5 ϫ Denhardt's solution (0.1% Ficoll (type 400; Pharmacia LKB, Uppsala, Sweden), 0.1% polyvinylpyrrolidone, 0.1% bovine serum albumin (fraction V; Sigma)), 0.1% SDS, and 0.1 mg/ml denatured calf-thymus DNA and then hybridized in the same buffer supplemented with 1 ϫ 10 6 cpm/ml of radioactive probe at 68°C for 20 h. The hybridization probe for MAZ was prepared by the random primed labeling method (22) using a Random Primed DNA Labeling Kit (Boehringer Mannheim, Mannheim, Germany). Filters were washed three times at room temperature for 20 min in 2 ϫ SSC, 0.1% SDS and once at 65°C for 30 min in 0.2 ϫ SSC, 0.1% SDS and then exposed to XAR-5 film (Eastman Kodak Co., Rochester, NY) with an intensifying screen. A 40-kb DNA insert of a cosmid clone was digested with EcoRI and then subcloned into the pBluescriptII SK ϩ vector (Stratagene) for further studies. The human MAZ-yeast artificial chromosome (YAC) recombinant clones, y645D4 and y976H4 were isolated from a library of CGM1 DNA (CHEF, Paris, France) by PCR-mediated methods according to the protocol provided by CHEF. The primers for YAC screening were described elsewhere (13,18).

FIG. 1-continued
Nucleotide Sequencing-DNA sequencing was carried out by the dideoxy chain termination method (23) with an automated DNA sequencer (ABI 373A; Applied Biosystems Inc., Foster City, CA) using a DNA Sequencing Kit (Ref. 24; Dye Terminator Cycle Sequencing Ready Reaction; Applied Biosystems Inc.). In the case of DNA fragments with an unusually high G ϩ C content, cycle sequencing was performed according to the protocols provided by the manufacturer (Applied Biosystems Inc.). Nucleotide and deduced amino acid sequences were analyzed with the GCG program (25).
Southern and Northern Blotting Analysis-High-molecular weight DNA from human cells was extracted, digested, fractionated on a 1% gel, and transferred onto a nylon membrane as described previously (21). The DNA on the membrane was allowed to hybridize with the probe, and the membrane was washed and exposed to Kodak XAR-5 film with an intensifying screen as described elsewhere (21). Multiple tissue Northern blots were obtained from CLONTECH (human MTN II, 7759 -1, and human MTN III, 7767-1; CLONTECH, Palo Alto, CA). The blots were hybridized and washed before autoradiography as described elsewhere (21). A 1.8-kb EcoRI fragment of cDNA for MAZ and a 0.6-kb EcoRI/BamHI fragment of the human DNA (Ϫ1.0 kb to Ϫ0.4 kb relative to the major site of initiation of transcription) were radiolabeled for use as DNA probes for further hybridization. S1 Nuclease Assay-Total RNA from HeLa cells or human peripheral blood lymphocytes (PBL) was prepared by the guanidine thiocyanate method, as described elsewhere (21). S1 nuclease protection was conducted essentially as described previously (26) using a 208-nt probe (nt Ϫ98 to 106) generated by the digestion of the promoter region of MAZ DNA with SmaI. The sizes of protected fragments were determined from comparisons with nucleotide sequencing ladders prepared from the protected DNA fragments with a Sequence 7-deaza-D-GTP kit (version 2.0) from Amersham.
Chromosome Mapping-Fluorescent in situ hybridization was performed as described by Ozawa et al. (27). Metaphase preparations were obtained from phytohemagglutinin-stimulated normal human male lymphocytes after synchronization with 5-bromodeoxyuridine (Sigma). The cosmid DNA was labeled with biotin-14-dATP (Life Technologies, Inc., Gaithersburg, MD) by nick-translation as described elsewhere (28). After hybridization, slides were washed, blocked, and incubated with rabbit antibodies against biotin (Enzo Diagnostic, Inc., New York) Slides were then incubated with second and third antibodies (goat antibodies against rabbit IgG and rabbit antibodies against goat IgG) conjugated with fluorescein isothiocyanate. After washing and drying, slides were counterstained with propidium iodide in anti-bleach mounting medium. Slides were examined under an Optiphot microscope (UFX-IIA; Nikon, Tokyo, Japan), and photographs were taken on Ektachrome 400 film (Kodak). Then, chromosomes were subjected to Qband staining. The localization of fluorescent signals on chromosomes was determined under a fluoresence microscope and photographed under the bright field of a light microscope with Minicopy HR II film (Fuji-Film Co., Kanagawa, Japan).
Culture and Transfection of Cells and Assay of CAT Activity-HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Life Technologies, Inc.) and 60 g/ml kanamycin (Sigma). Transfections for assays of shortterm expression of chloramphenicol acetyltransferase (CAT) were performed as described elsewhere (29,30) with one single change. After sonication, each crude extract was incubated for 10 min at 60°C to denature proteinases (30) Each thin-layer chromatography plate was exposed to RX film (Fuji, Tokyo, Japan). The extent of conversion of chloramphenicol to its acetylated form was determined with a Bio-Image analyzer (model BAS 2000; Fuji). ␤-Galactosidase activity was assayed as described by Katoh et al. (31). The ratio of CAT activity to that of ␤-galactosidase was used for normalization of results (31).
Polymerase Chain Reaction-Each polymerase chain reaction (PCR) was carrried out in a total volume of 20 l that contained 10 ng of each primer, 10 ng of DNA, 200 M each dNTP, ⌬Tth polymerase buffer (Toyobo, Kyoto, Japan), and 2.5 units of ⌬Tth polymerase (Toyobo). Samples were heated at 96°C for 2 min to denature the template DNA and then subjected to 30 cycles of 95°C for 40 s, 60°C for 30 s, and 72°C for 30 s in a DNA thermal cycler (9600 model; Perkin-Elmer, Foster City, CA). The products of PCR were separated by electrophoresis on a 8% acrylamide gel.
Synchronization of the Cell Cycle and Western Blotting Analysis-Human normal diploid cells (WI-38) were grown in DMEM supplemented with 10% FBS and 60 g/ml kanamycin. To induce the arrest of WI-38 cells in the G 0 phase, cells were allowed to grow exponentially in monolayer and maintained for 30 h in DMEM supplemented with 0.1% FBS. Arrested cells were switched to incubation in DMEM supplemented with 10% FBS and then harvested at 6-h intervals as indicated. G 0 -arrested cells were also maintained in DMEM supplemented with 10% FBS in the presence of 1 mM hydroxyurea (Sigma) for 24 h for resynchronization in early S phase. The cells were then incubated in DMEM supplemented with 10% FBS and 7.5 g/ml Hoechst 33342 (Sigma) for another 12 h for accumulation of cells in the G 2 phase. Cells were lysed and boiled for 3 min at 90°C and then lysates were sonicated for 10 s. The sonicates were fractionated by SDS-polyacrylamidg gel electophoresis (8% polyacrylamide) and then proteins were electroblotted on a polyvinylidene difluoride-nylon membrane (Daiichi Pure Chemicals Co., Tokyo, Japan). The membrane was blocked with 5% nonfat milk in PBS-T (0.2% Tween 20 in PBS) for 1 h at 25°C and then washed three times with PBS-T. The membrane was incubated with 1000-fold diluted polyclonal antibodies against MAZ (13,18) or against human ␤ 2 -microglobulin (BM-63; Sigma) in PBS-T for 2 h at 25°C and then washed three times with PBS-T. Antibodies that had bound to the membrane were detected with horseradish peroxidase-conjugated antibodies against mouse immunoglobin G (Zymed Laboratories Inc., South San Francisco, CA) and ECL detection reagents (Amersham Japan).

Cloning and Determination of the Genomic Structure of the
Human Gene for MAZ-Screening of a cosmid library of the genomic DNA from human B-lymphoblastoid AKIBA cells with the 1.8-and 0.7-kb EcoRI fragment of the coding region and the untranslated region of pMAZi cDNA, respectively (13), as probe yielded seven clones. DNA from the seven cosmid clones was digested with EcoRI and then allowed to hybridize with the same 1.8-kb DNA probe. A single 9.0-kb EcoRI fragment was detected in all seven cosmid clones and subcloned into Blue-scriptII SK ϩ . The complete restriction maps of the seven subclones were identical. From one of the seven subclones, pJSMAZ9.0E, we further subcloned the 1.3-kb PstI-PstI fragment and the 3.5-kb PstI-EcoRI fragment into pBluescriptII SK ϩ to generate pJSMAZ1.3P/P and pJSMAZ3.5P/E, respectively. A 1.1-kb EcoRI fragment adjacent to the inserted DNA of pJSMAZ9.0E was also isolated and further subcloned into pBluescriptII SK ϩ to generate pJSMAZ1.1E. These plasmids, pJSMAZ1.3P/P, pJSMAZ3.5P/E, and pJSMAZ1.1E were fully sequenced and characterized (Fig. 1).
Intron-exon boundaries were determined from the divergence of the genomic sequence from the sequence of the cDNA for the human protein and the presence of consensus splice donor-acceptor sequences was confirmed (32,33). The locations of five exons and four introns within the 6.0 kb of human genomic DNA cloned in this study are shown in Fig. 1. Both the 5Ј-and the 3Ј-splice junctions and the sizes of exons and introns are shown in Table I. The exons were small, with the exception of exon 5, which extended over 1028 nucleotides. Similarly, the introns were relatively small, ranging from 84 to 277 bp, with the exception of intron 4, which extended over 1.3 kb of DNA. The region that encoded the zinc finger motif of the MAZ protein extended from exon 2 to exon 4 and was interrupted by two introns. The 3Ј-untranslated region of the MAZ gene that we isolated was about 1.1 kb in length (Figs. 1 and 2 and Table  I). We also isolated two human MAZ-YAC recombinant clones, y645D4 and y976H4, from a YAC library of EB virus-transformed human peripheral lymphocytes (CGM1). The two cloned YACs spanned 560 kb of DNA that included the MAZ gene (Fig. 2). The NotI DNA fragment covered the entire genomic region of the MAZ gene and was subcloned into the pWE15 vector for generation of a physical map of the inserted DNA. The nucleotide sequence of the MAZ gene in the YAC clones was identical to that of the cloned cosmid DNA (data not shown).
Sequence of the 5Ј-Flanking Region of the MAZ Gene-We compared the genomic MAZ sequence with corresponding sequences of cDNAs for members of the MAZ family that had been reported previously (9 -11). We found relatively limited homology in the 5Ј-flanking region even though the sequences of the coding regions of these genomic and cDNA clones were identical (Fig. 3).
To examine whether the human genome contains several MAZ genes, we designed 5Ј-end primers on the basis of the 5Ј-end upstream sequences of human genomic MAZ DNA and two representative cDNAs (see Fig. 3b; 13, 15). The 3Ј-end primers corresponded to the conserved sequences in the genomic MAZ gene. Then human total genomic DNA and mix-tures of cDNA were used as templates for PCR. The primer combinations corresponding to the genomic MAZ sequences resulted in successful amplification, while with MAZi and clone 33813 cDNAs, we failed to amplify any DNA fragments (Fig.  3c). Thus, the differences in the 5Ј-flanking regions of both MAZi and clone 33819 cDNAs might have been due to artifacts in cloning due to the high G ϩ C content of this region. Alternatively, differential splicing is also a possible explanation in the case of MAZi. Detailed characterization of the physical map and the nucleotide sequences of YAC-MAZ clones supported this conclusion (data not shown).
A particular striking feature of the MAZ gene was its high G ϩ C content. The average G ϩ C content of the 6.0-kb genomic MAZ DNA was 68.9%; that of the upstream 2.5-kb part of the 6.0-kb genomic MAZ DNA was 76.2% and that of the downstream 2.5-kb part of the 6.0-kb genomic MAZ DNA was 61.2%. The 500-bp region upstream of the ATG initiation codon had a G ϩ C content of 88.4%. It is noteworthy that two regions of 71 and 77 bp (nt Ϫ103 to Ϫ33 and nt Ϫ306 to Ϫ230 relative to the ATG codon, respectively), had extremely high G ϩ C contents of  98.6 and 97.4%, respectively (Fig. 4). The GC-rich region contained restriction sites for some rare "CG cutters," such as NarI, BssHII, SrfI, and EagI (Fig. 2b), which are characteristic of "CpG islands" (34 -36). The 1.3-kb region at the 5Ј-end of the MAZ gene contained 171 copies of CpG and 196 copies of GpC dimers, whereas the 2-kb region at the 3Ј-end of the MAZ gene contained 165 GpC and only 70 CpG dinucleotides. A total of 45 HpaII sites was found in the MAZ gene (Fig. 2c). Thirty-four of them were found within the 3-kb region that encompassed the promoter and the first two exons of the MAZ gene, whereas only eight HpaII sites were found within a 3-kb region of the 3Ј-end (Fig. 2c). An abundance of HpaII sites, also referred to as HTF islands, in addition to the CpG islands, is characteristic of housekeeping genes (34 -36). The presence of a CpG island immediately upstream of the ATG codon strongly suggests that this region contains the promoter sequence of the gene for MAZ.
Several consensus motifs were found in the promoter region of the MAZ gene. Numerous CT-tract motifs, namely, one (TC-CCCC) 2 , one (TCC) 3 , one (TCC) 2 , one (TCCC) 3 , four (TCCC) 2 , and other CT tracts, were found in the upstream 77-bp G ϩ C stretch and a (CCG) 8 repeat was found in the 71-bp G ϩ C stretch (Fig. 4). Several (CCG) 3 , (GC) 4 , and (GC) 5 repeats were found in the region upstream of the ATG codon. Numerous putative binding sites for transcription factors were also identified in this region. Within this region, we found 16 potential Sp1-binding sites (5,8,37) and 26 potential AP-2-binding sites (38). No apparent TATA box or CAAT box was found in the 5Ј-flanking region. These data indicate that the gene for MAZ has features of the family of "housekeeping" genes (39). Moreover, the expression of the MAZ gene appeared to be ubiquitous and independent of cell type (see Fig. 8).
Characterization of the Site of Initiation of Transcription-The 5Ј-boundary of exon 1 was determined by S1 nuclease protection analysis of the SmaI/SmaI fragment (208 nt) of the promoter region of the MAZ genomic clone (Fig. 5). A strong signal corresponding to 106 bp was detected by the S1 protection analysis in the presence of the total RNA from HeLa cells. In addition, five weak signals corresponding to 100, 98, 97, 93, and 92 bp were also detected (Fig. 5). These results indicated that transcription of the MAZ gene started at multiple sites, namely, at positions ϩ1, ϩ7, ϩ9, ϩ10, ϩ14, and ϩ15. All of these sites are located within a CT repeated sequence (see Fig.  4).
MAZ is a Single-copy Gene Located on Chromosome 16p11.2-Southern hybridization analysis using DNA from HeLa cells and human PBL showed that a single 9.0-kb EcoRI fragment hybridized with the MAZ-specific cDNA probe (Fig.  6). When DNA was digested with BamHI, we detected three major MAZ fragments, of 4.4, 2.1, and 1.6 kb, respectively, using human PBL and HeLa cells DNAs. These results suggested that MAZ might be encoded by a single, unique gene. Digestion with BamHI revealed common 4.4-, 2.1-, and 1.6-kb fragments in WI-38, MKN7, MKN28, HGC27, GCY1, MKN45, PBL, and HeLa cells. The weak band of the EcoRI-generated 7.2-kb fragment might represent cross-hybridization with the DP-1 gene, which exhibits weak sequence homology to the gene for MAZ (Ref. 40; data not shown). The reciprocal hybridization of human PBL DNAs with the genomic BamHI fragments of MAZ DNA as probes yielded distinct fragments of 4.4, 2.1, and 1.6 kb (Fig. 6c). These results suggest that MAZ is encoded by a single, unique gene. The physical mapping of human-YAC recombinant clones confirmed the presence of a single, unique gene for MAZ (data not shown).
We next attempted to determine the chromosomal location of the human MAZ gene by in situ hybridization with the MAZ cosmid clone on spreads of replicated prometaphase chromosomes that had been prepared from phytohemagglutinin-stimulated normal human male lymphocytes. As shown in Fig. 7, fluorescent spots were observed on chromosome 16 and this result was reproducible. To map the MAZ gene with greater accuracy, we compared the immunofluorescence micrographs and the Q-banding patterns of the same cells. This analysis clearly indicated that the human MAZ gene mapped to band 16p11.2. To confirm this result, we performed a similar experiment using the immunogold detection method, with biotinylated DNA derived from the MAZ gene as the probe (41). The silver grains were concentrated on the same region of chromosome 16 (data not shown).
Characterization of the Three Major Transcripts-We examined the tissue distribution of MAZ transcripts using human multitissue Northern blots (Fig. 8). Transcripts were identified in all the tissues examined, albeit at different levels. We detected three mRNAs, of 1.6, 2.7, and 4.6 kb, respectively, with the latter two transcripts being major species. The transcripts of 2.7 and 4.6 kb were present in all tissues examined but in the liver, in particular, the level of the 2.7-kb transcript was very low. The levels of the 2.7-and 4.6-kb transcripts in the heart, placenta, pancreas, thymus, prostate, testis, colon, peripheral blood leukocytes, thyroid, and adrenal gland were higher than those in other tissues. The 1.6-kb transcript was detected mainly in the heart, placenta, pancreas, spleen, prostate, colon, thyroid, spinal cord, trachea, and adrenal gland.
Expression of MAZ during the Cell Cycle-As shown in Fig.  9a, the level of expression of MAZ protein was modulated by the cell cycle. The 60-kDa MAZ protein was produced during the G 0 and G 2 phases, but it was not detected during early S phase (Fig. 9a) When serum-starved normal diploid WI-38 cells were stimulated by addition to the medium of a high concentration of serum, the level of expression of MAZ appeared to be reduced at 12 h; the protein disappeared at 30 h and then it reappeared at 36 to 48 h (Fig. 9a). The level of expression of human HLA-associated ␤ 2 -microglobulin was not changed significantly during the cell cycle (Fig. 9a). These results indicated that the level of MAZ protein was modulated in a cell cycle-dependent manner. At early S phase, in particular, we were unable to detect the expression of MAZ protein. We next examined the level of MAZ mRNA by RT-PCR during cell cycle and found the similar changes as that of MAZ protein. By contrast, the level of expression of glyceraldehyde-3-phosphate dehydrogenase was unaltered (Fig. 9b). The CAT activities of WI-38 cells transformed with the pMAZCAT0 reporter construct reflected the variations in the level of the MAZ protein during the cell cycle (see Fig. 9c). Furthermore, the introduction of a MAZ expression vector into HeLa cells significantly and dose dependently enhanced the promoter activity of the pMAZCAT0 reporter plasmid (Fig. 9d). Thus, expression of the MAZ gene appears to be controlled during the cell cycle and to be regulated by the MAZ protein itself.
Characterization of the Promoter-To examine whether the 5Ј-flanking region of the human MAZ gene contained a functional promoter, we generated a series of chimeric MAZ promoter/CAT gene reporter constructs, as depicted schematically in Fig. 10. These constructs were used to transfect HeLa cells to characterize the regulatory elements of the human MAZ pro- moter. Promoter activity was retained after deletion to position Ϫ383 relative to the major site of initiation of transcription (pMAZCAT3; Fig. 10a). Further deletion to positions Ϫ248, Ϫ189, and Ϫ40 (pMAZCAT4, pMAZCAT5, and pMAZCAT6) resulted in significant decreases in promoter activity. Moreover, the internal deletion mutant pMAZCAT2-d (with a deletion from nt Ϫ383 to Ϫ248) had diminished promoter activity. Therefore, DNA sequences between nt Ϫ383 and Ϫ248 appeared to be required for high basal promoter activity in HeLa cells. Negative elements might be present between nt Ϫ948 and Ϫ2500 because the activity due to pMAZCAT0 was significantly repressed as compared with that due to pMAZCAT1. These results clearly demonstrated that the 5Ј-end flanking region contained the functional promoter of the human gene for MAZ.
As shown in Fig. 9, a and b, the levels of MAZ protein and MAZ mRNA and expression of the MAZ gene varied in a cell cycle-dependent manner. Moreover, increased concentrations of the MAZ expression vector resulted in significant induction of the CAT reporter activity of the MAZ-CAT gene (Fig. 9d). Therefore, we examined whether the MAZ protein might be able to regulate the expression of its own gene. We co-transfected cells with a MAZ-CAT reporter construct and a MAZ expression plasmid (pCMV-MAZ) and monitored expression of the MAZ gene. As shown in Figs. 10c, the CAT activity due to the pMAZCAT4 construct was increased significantly by cotransfection with pCMV-MAZ, while that of the deletion construct pMAZCAT3-d was not, suggesting that an element between nt 248 and 189 might be involved in positive enhancement of the regulation of expression of the MAZ gene. The CAT activity due to pMAZCAT2 was lower than that due to the deletion mutant pMAZCAT2-d in the presence of pCMV-MAZ, suggesting that the deleted region (nt Ϫ383 to Ϫ248) might play a role in down-regulation of the MAZ gene by the MAZ protein. The regulatory sequences seemed to overlap one another.
The CAT activity due to pMAZCAT0 was enhanced significantly in the presence of MAZ protein, whereas that of pMAZCAT1 was not. Thus, the regions from nt Ϫ248 to Ϫ189 and from nt Ϫ2500 to Ϫ948 appear to contain putative positive regulatory elements, in contrast to the region from nt Ϫ383 to Ϫ248 that seems to contain negative regulatory elements. DISCUSSION The present study revealed the exon-intron structure of the human gene for MAZ that spans approximately 6.0 kb and consists of promoters, five exons, four introns, and a 3Ј-untranslated region. In addition, physical mapping studies of MAZ-YAC clones demonstrated that the MAZ gene is a single and unique gene. All exon-intron boundaries begin with GT at the 5Ј-end and terminate with AG at the 3Ј end, conforming to the GT-AG rule (42). When compared with three reported cDNAs for human MAZ, the insert in our clone was most similar to clone 33819 from HeLa cells (the coding region was the same as that of clone 33819). The major differences were found in the 5Ј-end promoter region (see Fig. 3). The corresponding region of the gene for MAZi from human islets is missing about 35 nucleotides (G ϩ C; from nt ϩ69 to ϩ128). From the results of Southern blotting and RT-PCR with cosmids and YAC recombinant clones, it appeared that MAZ might be encoded by a single gene (Figs. 3 and 6). The heterogeneity of the 5Ј-end promoter regions of cDNA sequences for MAZ might be due to artifacts that arose because of the high G ϩ C content. Alternatively, differential splicing might have occurred in the case of the MAZi gene from human islets.
Multiple sites for initiation of transcription were found within 174 bp upstream of the site for initiation of translation by the S1 nuclease protection assay (Fig. 5). These sites are located in a putative initiator sequence near the major site for initiation of transcription that matches the 5Ј-YYC(A/ T)YYYYY-3Ј (Y, pyrimidine) consensus sequence (43). Furthermore, the results of a transcription experiment in vitro showed that the sequence in the vicinity of the defined site of initiation of transcription was sufficient to promote faithful transcription (data not shown). We tried to confirm the site of initiation of transcription in a primer extension assay. However, we were unsuccessful since the reverse transcriptase did not read through the 5Ј-end regions with a high G ϩ C content. Furthermore, the MAZ promoter-CAT construct also demonstrated the significant activity of the promoter (Figs. 9 and 10). Taken together, our results demonstrate that transcription of the MAZ gene originated from a TATA-less promoter in vivo and in vitro and that the 5Ј-end region really contained the promoter region of the MAZ gene required to generate a 2.7-kb mRNA. The exact nature of the transcription factors and the specificity of expression of the MAZ gene from this promoter region remain to be determined.
Sequence analysis of the first exon and the 5Ј-upstream region suggested that this region had an unusually high G ϩ C content. In particular, the average G ϩ C content of the 500-bp region upstream of the ATG codon was 88.4%. It is noteworthy that two regions, extending over 71 and 77 bp (nt Ϫ103 to Ϫ33 and nt Ϫ306 to Ϫ230 relative to the ATG codon, respectively), had extremely high G ϩ C contents of 98.6 and 97.4%, respectively (Fig. 4). The 1.3-kb fragment of the 5Ј-end boundary of the MAZ gene contained 171 copies of CpG and 196 copies of GpC islands. Furthermore, G ϩ C-rich sequences were located before and after the cap site (ϩ1) of the MAZ gene, within 280 bp. These two long G ϩ C-rich sequences might contribute to changes in DNA conformation and might be modified, for example, by methylation (34 -36). Such changes might explain the strength of the promoter of the MAZ gene under certain conditions and in different types of cells. FIG. 5. S1 mapping of sites of initiation of transcription of the MAZ gene. a, a [␥-32 P]ATP-labeled 208-bp DNA probe was allowed to hybridize to 25 g of total RNA from HeLa cells. The sample was size-fractionated on a 6% polyacrylamide gel together with a sequencing ladder for sizing of the protected fragments. Arrows show the multiple sites of initiation of transcription of the human gene for MAZ. b, the S1 probe of 208 bp was isolated from the pJSMAZ1.3P/P subclone and radiolabeled as described under "Experimental Procedures." CT tracts were present in the region upstream of the small 77-bp region with high G ϩ C content. Such CT tracts are strongly reminiscent of structures described in the gene for the receptor of epidermal growth factor (EGFR) (44) and in other promoters, such as those of the ets-2 (45), c-Ki-ras (46), and c-myc (4, 13) genes. Such repeats are thought to form triplex or H-DNA structures that are sensitive to a variety of nucleases (47). In an examination of CT tracts in the gene for epidermal growth factor receptor, Johnson et al. (44) showed that a specific factor (CTF) binds these sequences and that deletion of the CT tracts significantly down-regulates transcription. We reported similarly that a MAZ protein binds the CT tracts of the c-myc promoter region (13).
(CCG) 8 repeats were present in the 71-bp GC-rich stretch and several (CCG) 3 , (GC) 4 , (GC) 5 repeats were also present in the 500-bp long GC-rich region. GC-rich promoters and the absence of a TATA sequence are characteristic of housekeeping genes (34 -36, 43). In fact, Northern blotting analysis revealed the expression of MAZ transcripts in a large variety of human tissues. Examination of the DNA sequence upstream of exon 1 failed to reveal any TATA box or CAAT box. Similar findings have been obtained for a variety of oncogenes, such as human ets-1 (48), human ets-2 (45), human fgr (49), human c-src (50) and murine c-Ki-ras (46); for genes for growth factors and their receptors, such as human epidermal growth factor receptor (44) and insulin-like growth factor receptor (51); and for housekeeping genes, such as the gene for adenosine deaminase (52). The promoters of such genes have a number of common characteristics, such as the presence of multiple sites for initiation of Commercially purchased multiple-tissue Northern blot filters, with approximately 2.0 g/lane of poly(A) ϩ RNA from each indicated tissue, were allowed to hybridize with a 1.8-kb 32 P-radiolabeled DNA fragment of a cDNA clone for human MAZ (13).
transcription, presumably because of the absence of a TATA box and a CAAT box, and they often have an unusually high G ϩ C content. We found 16 consensus binding sites for Sp1 and 26 binding sites for AP-2 that were located close together or as partially overlapping sites in the 5Ј-end promoter region. There were also several potential binding sites for the epidermal growth factor receptor-specific transcription factor (ETF) in the promoter region. This protein is thought to replace TATA- Numbering is relative to the major site of initiation of transcription. Plasmid DNAs (10 g) were used to transfect HeLa cells and CAT activity was measured as described under "Experimental Procedures." b, CAT fusions with the human MAZ gene and transient expression of CAT driven by the MAZ promoter (5 g) after co-transfection with pCMV-MAZ (10 g). Plasmid DNA was use to transfect HeLa cells and CAT activity was measured as described under "Experimental Procedures." Percent conversion of acetyltransferase is indicated. c, promoter activities of MAZ-CAT fusion genes are expressed relative to the activity of pSV00CAT (5 g) in the absence of pCMV-MAZ, which was taken arbitrarily as 1.0. All values are the averages of results from at least three experiments and the standard deviation for each value is indicated. binding proteins in the control of expression of genes that lack TATA boxes (53).
We identified mRNAs of 1.6, 2.7, and 4.6 kb, respectively. It is of interest that only the 4.6-kb mRNA, was detected when the EcoRI/BamHI DNA fragment that corresponded to the promoter region of the MAZ gene (Ϫ1.0 to Ϫ0.4 kb relative to the site of initiation of transcription) was used as DNA probe (data not shown). Thus, it is likely that the transcription of the 4.6-kb mRNA is initiated from the far upstream promoter. Studies of transcription and promoter activities in vitro and nucleotide sequencing indicated that each transcript, namely, the 4.6, 2.7, and 1.6 kb transcripts, might have been generated by corresponding independent promoters, which the possibility of differential splicing or differentiated polyadenylation seems less likely. However, we failed to isolate a full-length cDNA clone that corresponded to the 4.6-kb mRNA by molecular cloning, probably because of the unusual high G ϩ C content of this region, as described above.
The 60-kDa MAZ protein was detected during the G 0 and G 2 phases but not during the early S phase. When the cells were stimulated to reinitiate the cell cycle from the G 0 phase by addition of a high concentration of serum, the expression of MAZ was down-regulated at 12 h, was lowest at 30 h and increased from 36 to 48 h (Fig. 9a). The similar changes in the levels of MAZ mRNA were obtained during the cell cycle (Fig.  9b). These data indicated that expression of the MAZ gene was controlled by the cell cycle, as was the activity of the MAZ promoter (Fig. 9c).
In order to identify the elements that regulate the basal transcription of the MAZ gene, we constructed a series of deletion mutants of a MAZ-CAT fusion gene and transfected HeLa cells with them (Fig. 10). We identified a positive control element from nt Ϫ383 to Ϫ248. This region was required for maximal promoter activity, containing multiple consensus binding sites for Sp1 and AP-2, as well as two (TCCC) 2 elements. It remains to be determined whether these or other factors are really involved in the regulation of expression of the human MAZ gene. Negative regulatory elements might be localized in the far upstream region between nt Ϫ2500 and Ϫ948.
We also showed that the expression of the MAZ gene is controlled by its own product, the MAZ protein. Positive elements for autoregulation by the MAZ protein were putatively identified in the proximal region from nt Ϫ248 to Ϫ189 and in the distal region from nt Ϫ2500 to Ϫ948. The former proximal region contains two consensus (TCCC) elements and Sp1-and AP-2-binding sites. Negative autoregulatory elements were found in the region between nt Ϫ383 and Ϫ248 that is adjacent to the positive element. The regulatory elements for autoregulation by the MAZ protein had a reciprocal relationship to the regulatory elements for basal transcription of the MAZ gene: the enhancer region for basal transcription was "shut off" and the region for negative control of basal transcription was "turned on" by the product of the MAZ gene. This scenario is supported by the observation that forced expression of the MAZ gene resulted in significant and dose-dependent enhancement of the promoter-CAT activities of the pMAZCAT0 construct (Fig. 9d). We do not yet know the functional significance of these elements in autoregulation. We are currently trying to identify the transcriptional factors that are involved in autoregulation of the MAZ gene. Numbering is relative to the major site of initiation of transcription. The DNA fragments of MAZ promoter (Ϫ948 to ϩ259 and Ϫ383 to ϩ259) were digested with FseI, blunt ended, added the pHindIII linker, and ligated into the HindIII sites of pSV00CAT to generate pMAZCAT1a and pMAZCAT3a, respectively. Plasmid (10 g) was used to transfect HeLa cells and CAT activities were measured as described under "Experimental Procedures."