Regulation of the Human MAT2A Gene Encoding the Catalytic a 2 Subunit of Methionine Adenosyltransferase, MAT II GENE ORGANIZATION, PROMOTER CHARACTERIZATION, AND IDENTIFICATION OF A SITE IN THE PROXIMAL PROMOTER THAT IS ESSENTIAL FOR ITS ACTIVITY*

Mammalian methionine adenosyltransferase II (MAT II) consists of a catalytic a 2/ a 2 * and a regulatory b subunit. Up-regulation of a 2 subunit expression is associ-ated with increased intracellular levels of S -adenosyl-methionine, the major methyl group donor and a key compound in cell metabolism and polyamine synthesis. Previous studies have shown that expression of the a 2 subunit is differentially regulated in normal and malignant cells. To delineate the molecular basis for the differential regulation of a 2 subunit expression, we cloned and characterized the human MAT2A gene and its promoter and defined regions that contain enhancer and repressor elements. Detailed functional characterization of the proximal promoter of the MAT2A gene revealed the formation of three major protein-DNA complexes with probes containing three Sp1 sites (Sp1-1 at 2 14, Sp1-2 at 2 47, and Sp1-3 at 2 69). Competition with a probe copying sequence between 2 76 and 2 54, which contains the Sp1-3 site only, or mutation of this site, abolished complex formation. Furthermore, mutation of the Sp1-3 site, but not the Sp1-1 or Sp1-2 sites, inhibited the in vivo promoter activity by ; 85%. serum-containing medium to inhibit trypsin activity. The cell mixture was incubated for 2 h inplastic tissue culture plates to separate the adherent Cos-1 cells from the nonadherent Jurkat cells. Cos-1 cells adhered readily, and the nonadherent Jurkat were recovered by gentle pipetting. Jurkat cells isolated in this fashion were subjected to two additional cycles of plastic adherence to remove any residual Cos-1 cells. Jurkat 93.0% as analyzed by flow cytometry using a PerCP-Cy5.5-labeled monoclonal anti-human CD3 antibody (Becton and Dickinson, clone SK7). adherent Cos-1 cells were also recovered, the presence of Jurkat T cells was tested for by flow cytometry using the anti-CD3 mAb, population shown to be 100% pure. Luciferase activity assayed in extracts of recovered Jurkat and Cos-1 as above.

Methionine adenosyltransferase (MAT) 1 is a key enzyme in metabolism, because it catalyzes the synthesis of S-adenosylmethionine (AdoMet), the major methyl group donor (1,2). Methylation is essential for the synthesis and modification of proteins, lipids, RNA, and DNA. It is well established that DNA methylation can control the expression of certain genes (reviewed in Ref. 3). Moreover, AdoMet is essential for polyamine biosynthesis and regulates essential metabolic pathways (4,5). Inasmuch as the levels of AdoMet vary in different tissues and at different stages of differentiation, it is believed that the regulation of AdoMet synthesis and utilization is important for normal cell development and function (6 -11). Delineating the mechanism by which the MAT enzymes are regulated will, therefore, be a key to understanding the control of AdoMet level and its contribution to cell growth and differentiation.
The catalytic subunit of MAT, designated ␣, is highly conserved among species. In mammals, the two major MAT isozymes are designated MAT I/III and MAT II, and their catalytic subunits are designated ␣1 and ␣2, respectively (12). The MAT1A gene encodes the ␣1 protein and the MAT2A gene encodes the ␣2 protein (12). Despite the high degree of homology (84% identity at the amino acid level), the mammalian MAT isozymes differ considerably in their physical and kinetic properties as well as in their tissue distribution (12)(13)(14)(15)(16)(17). Expression of ␣1 is confined to liver, whereas ␣2 is expressed in all tissues, including liver. In addition to the normal differences in the tissue distribution of the ␣1 and ␣2 subunits, expression of these proteins is altered in liver cancer. In hepatic tumors, ␣1 is markedly down-regulated while ␣2 expression is increased (18 -21). Molecular mechanisms responsible for the differential expression of MAT ␣1 and ␣2 subunits in normal and malignant tissues are not well defined.
In 1985, a human 185-kDa MAT II enzyme was purified to homogeneity from leukemic T cells (22). It was demonstrated to consist of two related 53-and 51-kDa ␣2 and ␣2Ј subunits, complexed with a structurally unrelated 38-kDa protein, which was designated the ␤ subunit (22). Ensuing studies showed that the regulation of the MAT II ␣2 and ␤ subunits expression differs among normal resting, activated, and malignant T cells (23)(24)(25). Physiological activation of normal human T cells induces the up-regulation of ␣2/␣2Ј and the down-regulation of the ␤ subunit; by contrast, expression of both ␣2/␣2Ј and ␤ is constitutively high in leukemic B and T cells (23)(24)(25).
The present study reports functional analysis of 2.4 kbp of the MAT2A promoter, which identified regions with positive and negative regulatory activities that were not reported in previous studies investigating the rat MAT2A genes and a segment of the human MAT2A promoter (26,27). Detailed functional analysis of the proximal promoter of the human MAT2A gene, reported here, reveal a region that acts as a nucleation site for formation of protein-DNA complexes and that appears to be essential for promoter activity in vivo. We believe that this information may lead to a better understanding of the differential regulation of MAT II expression and AdoMet levels in different cell types as well as in normal and leukemic human T lymphocytes.

MATERIALS AND METHODS
Isolation and Genomic Organization of the MAT2A Gene-Forward and reverse primers based on the sequence of the previously reported MAT II ␣2 subunit cDNA (28) spanning the entire open reading frame were designed and used to amplify genomic DNA isolated from normal human lymphocytes. The presence of introns was indicated by PCR products larger than predicted from coding sequence, and these products were cloned and sequenced. Intron-exon boundaries of the MAT2A gene were determined by aligning the cDNA sequence of MAT2A cDNA with the genomic sequence. Probes of 80 -150 bp spanning exon/intron junctions were used to screen a human P1 genomic library (Genome Systems Inc., St. Louis, MO), and clone 2007 was selected for further characterization. Sequence and PCR analyses of the human MAT2A locus identified the location and sizes of the exons and introns.
Mapping the MAT2A Gene Transcription Start Site-Poly(A) ϩ RNA was isolated from 500 ml of human blood by the Poly(A)Ttract mRNA isolation system (Promega) and used for primer extension analysis. Two primers, PE-1 and PE-2, complementary to nucleotides 5Ј(ϩ10)-GTC-CGTTCATGTTGGTG-3Ј and nucleotides 5Ј(Ϫ1)-GTTGGTGTCGGTGT-GCGGAAGT-3Ј of the human MAT II ␣2 cDNA were synthesized and end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase. Free ATP was removed by the nucleotide removal kit (Qiagen), and labeled primers (500,000 cpm) were hybridized to 2-4 g of poly(A) ϩ RNA. Extension was performed in the presence of 15 units/g avian myeloblastosis virus reverse transcriptase (Promega). Excess labeled primer was removed using a QIAquick PCR purification system (Qiagen). The eluted cDNA was concentrated and dissolved in 3 l of loading buffer containing 80% formamide, 1ϫ TBE buffer (90 mM Tris borate with 2 mM EDTA), xylene cyanol, and bromphenol blue, heated at 95°C for 3 min, and analyzed on 6% polyacrylamide 7 M urea sequencing gels. An M13mp18 sequencing ladder was used as a size marker.
Cloning of the Human MAT2A Gene Promoter-We cloned and sequenced ϳ2.4 kbp of the 5Ј-flanking DNA of the human MAT2A gene. PCR primers were designed to amplify the region ϩ102/Ϫ2374 (see Table I) in the presence of Pfu DNA polymerase (Promega), and the product was cloned into the pGEM-TEasy vector (Promega). The primers introduced a NheI site at the 5Ј-end and a XhoI site at the 3Ј-end of the cloned promoter. MAT2A promoter sequence was confirmed by sequencing from both directions with overlapping segments by the fmol sequencing system (Promega).
Generation of Luciferase Reporter Constructs of the MAT2A Promoter-A 2.4-kbp NheI/XhoI-digested fragment of the MAT2A gene containing the 5Ј-flanking region starting at position Ϫ19 from the translation start site was cloned upstream of a firefly luciferase (LUC) reporter gene in the pGL3 Enhancer vector (Promega). Insertion orientation was verified by restriction enzyme digestion and sequencing from both ends. The pGL3-MAT2A plasmid was treated with T4 DNA ligase (Promega) to repair any nicks. The 5Ј-deletions were generated by exonuclease III treatment using KpnI/NheI-digested pGL3-MAT2A and the Erase-a-base kit (Promega). Briefly, the vector was digested with NheI to provide 5Ј-overhangs for digestion with exonuclease III and with KpnI to provide a protected 3Ј-overhang. The exonuclease III reaction was incubated at 37°C, and at specific time points aliquots were removed and added to tubes containing S1 nuclease to digest single-stranded DNA. Resultant deletion constructs were blunt-ended with Klenow and ligated to the pGL3 Enhancer vector with T4 DNA ligase (Promega). Further deletion constructs were generated by taking advantage of the SacII, ApaI, PstI, and StuI restriction sites within the promoter sequence, or were generated by amplification using specific primers (Table I). All constructs were confirmed by sequence analysis.
In Vivo Analysis of the MAT2A Promoter Activity-Functional expression of deletion constructs of the MAT2A promoter was analyzed in Cos-1 and Jurkat T cells. Transient transfection of Cos-1 cells was performed by the TransFast method (Promega) according to manufacturer instructions. Cos-1 cells (1 ϫ 10 5 ) were cultured in DMEM ϩ 10% fetal calf serum until roughly 70% confluency was reached and transfected with the LUC reporter pGL3-MAT2A promoter deletion constructs along with cytomegalovirus Renilla luciferase vector (Promega) pRL-CMV DNA (40:1 ratio) as an internal control. Briefly, 400 l of transfection media (DMEM without fetal bovine serum) containing 600 ng of LUC reporter plasmid were added to the semi-confluent Cos-1 cells along with 15 ng of the pRL-CMV DNA and 4 l of TransFast. The plates were incubated for 2 h, then 2 ml of DMEM ϩ 10% fetal calf serum was added and incubated for another 48 h at which time the cells were harvested, and luciferase activity was assayed using the dual luciferase reporter assay system (Promega). After normalization to the Renilla luciferase control, the activity of each MAT2A promoter construct was calculated and expressed as relative light units (RLU). All samples were analyzed in triplicate, and experiments were performed at least four times. The data are presented as mean Ϯ S.E. of RLU of at least four different experiments as well as percent RLU relative to the activity of the largest construct used, pGL3-MAT2A(Ϫ2374)-LUC, which was set at 100%.
Initial attempts to transfect human Jurkat T cells by electroporation yielded low levels of transfection efficiency. However, an almost 300fold increase in efficiency of transfection of Jurkat cells with the pGL3-MAT2A plasmid was achieved by the method of Keller et al. (29), with a few modifications. Briefly, 2.5 ϫ 10 5 Jurkat cells were seeded into 12-well tissue culture plates that had been coated with an adherent monolayer of Cos-1 cells at ϳ70% confluency. The cells were co-cultured overnight to allow the Jurkat cells to adhere to the Cos-1 cell monolayer. The next day, the wells were rinsed twice with Hanks' balanced salt solution, and then the pGL3-MAT2A promoter deletion constructs (1.25 g) were added to the adherent cells along with the control Renilla luciferase vector pRL-CMV DNA (31.25 ng; i.e. a 40:1 ratio) in a final volume of 400 l of serum-free DMEM:RPMI medium (1:1) containing 7.7 l of TransFast (Promega). After incubation at 37°C for 4 -5 h, the transfection mixture was removed, and the cells were further incubated for 36 h in DMEM:RPMI medium (1:1) plus 10% fetal calf serum. The adherent cells were washed with 1ϫ Hanks' balanced salt solution and then recovered by trypsinization for 5 min followed by the addition of serum-containing medium to inhibit trypsin activity. The cell mixture was incubated for 2 h in plastic tissue culture plates to separate the adherent Cos-1 cells from the nonadherent Jurkat cells. The Cos-1 cells adhered readily, and the nonadherent Jurkat cells were recovered by gentle pipetting. Jurkat cells isolated in this fashion were subjected to two additional cycles of plastic adherence to remove any residual Cos-1 cells. The purity of the recovered Jurkat cells was Ͼ93.0% as analyzed by flow cytometry using a PerCP-Cy5.5-labeled monoclonal anti-human CD3 antibody (Becton and Dickinson, clone SK7). The adherent Cos-1 cells were also recovered, the presence of Jurkat T cells was tested for by flow cytometry using the anti-CD3 mAb, and the population was shown to be 100% pure. Luciferase activity was assayed in extracts of the recovered Jurkat and Cos-1 cells as described above.
a Primers denoted WP and CD were used for promoter constructs. Primers denoted WGS and GS were used to make probes for use in EMSA.
In Vitro Analysis of the MAT2A Promoter Activity by Electrophoretic Mobility Shift (EMSA) and Supershift Assays-Double-stranded oligonucleotide probes were generated by PCR amplification with 32 P-endlabeled primers. The amplified DNA representing specific regions of the proximal promoter of the MAT2A gene were generated using the primers listed in Table I. Cos-1 and Jurkat cell nuclear extracts were prepared by washing cell pellets in 1ϫ Hanks' phosphate-buffered saline and resuspending them in a hypotonic buffer (10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM PMSF, and 0.5 mM DTT). After centrifugation for 5 min at 3000 rpm, the pellet was resuspended in the same buffer and allowed to swell on ice for 10 min, then homogenized. Nuclei were harvested by spinning for 15 min at 4000 rpm, resuspended in hypotonic buffer, and homogenized. An equal volume of high salt buffer (20 mM HEPES, pH 7.9, 25% glycerol, 1.5 mM MgCl 2 , 1.2 M KCl, 0.2 mM EDTA, 0.2 mM PMSF, and 0.5 mM DTT) was added, and the extracted nuclei were pelleted by centrifugation for 30 min at 14,500 rpm at 4°C. Supernatants were dialyzed for 1 h against 2 liters of dialysis buffer (20 mM HEPES, pH 7.9, 20% glycerol, 100 mM KCl, 0.2 mM EDTA, 0.2 mM PMSF, and 0.5 mM DTT). Aliquots were stored at Ϫ80°C.
Binding reactions were performed in a 20-l final volume by incubating 60 fmol (Ϸ50,000 cpm) of 32 P-labeled probe with 5 g of crude nuclear extracts from Cos-1 or Jurkat cells in the presence of 2 g of poly(dI-dC), 1 g of salmon sperm DNA in 20 mM Hepes, pH 7.9, 50 mM KCl, 1 mM MgCl 2 , 0.1 mM EDTA, 0.5 mM DTT, 10% glycerol, and 6 g of bovine serum albumin for 40 min on ice. The reactions were conducted in the absence or presence of specific or nonspecific competitors, both added at a 100-fold molar excess. Specific competitors were cold probes; nonspecific competitors were unlabeled PCR amplification products of an unrelated DNA sequence, free of sites of interest, and similar in size to specific competitor. An Sp1-specific competitor was purchased from Santa Cruz Biotechnology Inc., Santa Cruz, CA (catalogue no. SC2502). The DNA-protein complexes formed were analyzed by electrophoresis on nondenaturing 4% polyacrylamide gels. The gels were prerun for 1 h at 100 V, and electrophoresis was done at 30 mA of constant current.
Supershift assays were performed using antibodies (Abs) to specific factors that have corresponding recognition elements within the MAT2A promoter region analyzed. Binding reactions were carried out as described above in the absence or in the presence of 2 g of Ab specific to one of the transcription factors of interest. The Ab was premixed with the nuclear extract, incubated for 30 min at room temperature prior to addition of radiolabeled probe, or added after the binding reaction was completed and incubated for 30 min on ice. The following Abs (Santa Cruz Biotechnology Inc.) were used in supershift analyses: mouse anti-Sp1 mAb (catalogue no. sc420X), anti-Sp2 (catalogue no. sc-643X), anti-Sp3 (catalogue no. sc-644X), anti-Sp4 (catalogue no. sc-645X), anti-PEA3 mAb (catalogue no. sc113X), rabbit poly-clonal anti-AP-2 (catalogue no. sc184X), anti-c-Myb (catalogue no. sc7874X) Abs, and goat polyclonal anti-GATA1 (catalogue no. sc1233X).
Western Blot Analysis of Proteins in Protein-DNA Complexes-32 P-Labeled probe G representing sequences between Ϫ76 and Ϫ54 was incubated with nuclear extracts from Jurkat T cells and separated on native 4% acrylamide gels as described above. The region containing the radiolabeled protein-DNA complexes formed was excised from the gel and boiled in SDS sample buffer for 5 min and loaded onto 10% SDS-polyacrylamide gel electrophoresis gels, followed by immunoblotting as described previously (30) using antibodies to members of the Sp1 family of transcription factors that were used in the supershift analysis.
In Vivo and in Vitro Analysis of the Effect of Mutations in the Sp1-1, Sp1-2, and Sp1-3 Sites-Sp1 site mutations of clone pGL3-MAT2A (110)-LUC were generated by site-directed mutagenesis (23) using the following mutagenic oligos: 5ЈmSp1-1F, CCT CCCGCT GCT TCG TTC; 3ЈmSp1-1R, CAG CGA ACG AAG CAG CGG GAG GTG CG; 5ЈmSp1-2F, TCG CGG GCG CTG CTC TA; 3ЈmSp1-2R, TTA TAG AGC AGC GCC CGC GAG C; 5ЈmSP1-3F, AGG TGC GTG GCC CGA GC; and 3ЈmSp1-3R, GCT CGG GCC ACG CAC CT. Primers 5ЈCD3 and 3ЈWP, which flank the 3 Sp1 sites, were used to amplify the whole region (see Table  I). PCR products were cloned into pGEM-T-Easy (Promega). The resultant mutagenized fragments were confirmed by sequence analysis, excised using KpnI and XhoI, and individually cloned into the pGL3-Enhancer vector (Promega). The in vitro and in vivo activity of each of the three Sp1 mutants was determined as described above. EMSA probes were generated using the mutated pGEM-T-Easy clones as templates and outside primers 5ЈCD3 and 3ЈWGS (Table I).
Statistical Analysis-Paired t tests were used to test for significant differences between the in vivo activity of the MAT2A promoter constructs. All constructs were analyzed in triplicate, and experiments were performed at least four times.

RESULTS
Genomic Organization of the MAT2A Gene-Human P1 genomic clone 2007, encompassing the entire MAT2A gene, was used to characterize the gene and its promoter. The human MAT2A gene consisted of nine exons separated by eight introns, contained within ϳ6 kbp of genomic DNA (31) ( Table II). The exons ranged from 78 (exon 2) to 1639 bp (exon 9) in size (Fig. 1). Exon 1 contained 91 bp of coding sequence and 120 bp of 5Ј-untranslated region, and exon 9 contained 100 bp of coding sequence and 1539 bp of 3Ј-untranslated region (Fig. 1). The overall organization of the human MAT2A gene was very similar to the rat MAT2A (26) and the human MAT1A genes (32), however, important differences in the promoter were found as detailed below. Mapping the MAT2A Gene Transcription Start Site-Identification of the transcription start site was determined by primer extension as detailed under "Materials and Methods" using poly(A) ϩ RNA from normal human lymphocytes and primers PE-1 and PE-2 complementary to nucleotides starting at positions ϩ10 and Ϫ1, respectively, relative to the translation start site. As shown in Fig. 2, transcription begins at position Ϫ120 relative to the translation start site. These data are in conflict with data by Mao et al. (27) who reported that the transcription of the human MAT2A gene start site is located at Ϫ132 upstream of the translation start site, but are in close agreement with Hiroki et al. (26) who demonstrated that the transcription initiation site for the rat MAT2A gene is located 123 bp upstream of the translation start codon.
The sequence of 2.4 kbp of the 5Ј-flanking region of the MAT2A promoter is shown in Fig. 3. Sequence analysis identified the presence of recognition sequences for a large number of transcription factors. They include the canonical TATA box at Ϫ32 and an inverted CAT box at Ϫ88, and three Sp1 sites, which we designated Sp1-1 (Ϫ14/Ϫ9), Sp1-2 (Ϫ47/Ϫ39), and Sp1-3 (Ϫ69/Ϫ63). The Sp1-3 site is also the GC box. Putative binding sites for several transcription factors were found, and they included an AP-2 site at Ϫ104, two PEA3 sites at Ϫ119 and Ϫ273, a GATA1 site at Ϫ212, and an NF-1 site at Ϫ446 relative to the transcription start site. A series of deletion constructs were used to test the functionality of the various sites, as discussed below.
In Vivo Analysis of the MAT2A Promoter-A series of 5Ј-MAT2A promoter deletion constructs coupled to a firefly luciferase reporter gene in a pGL3 Enhancer vector were generated as described under "Materials and Methods" and used to analyze functional expression in Cos-1 and Jurkat human leukemic T cells. The pattern of the MAT2A promoter activity was very similar in both cell types (Fig. 4). Promoter sequences contained between Ϫ87 and Ϫ58 from the transcription start site harbor most of the necessary elements required to drive transcription of the MAT2A gene (Fig. 4); however, sequences between Ϫ325 and Ϫ128 significantly contributed to enhanced promoter activity. Sequential deletions gradually reduced promoter activity with a significant (p Ͻ 0.05) drop in activity when sequences between Ϫ128 and Ϫ110 were deleted (Fig. 4), suggesting that this region contains elements that enhance promoter activity. The inclusion of the region from Ϫ1281 to Ϫ423 significantly reduced promoter activity (p Ͻ 0.001), suggesting the presence of repressor elements particularly in the region from Ϫ423 to Ϫ325. This region contains recognition sites for a c-Myb motif at Ϫ359 (Ϫ), an AP-2 site at Ϫ396 (Ϫ), and an Sp1 site at Ϫ406. Interestingly, a marked and highly significant enhancement of promoter activity was observed when sequences between Ϫ1881 and Ϫ1281 were included (p Ͻ 0.001). This enhancement, which was seen in five different experiments, was significantly suppressed by the inclusion of 5Ј-flanking sequences from Ϫ2374 to Ϫ1881 (p Ͻ 0.001). The latter region contains the following putative transcription factors recognition sites: an AP-2 site at Ϫ1886, an NF-Y site at Ϫ2025, a c-Myb site at Ϫ2309 and two GR sites at Ϫ2281 and Ϫ2347, respectively.
In this study, we focused on the proximal promoter of the MAT2A gene (up to bp 285 relative to transcription start site) and hereby describe detailed characterization of this region.
In Vitro Analysis of the Proximal MAT2A Promoter Activity by Electrophoretic Mobility Shift (EMSA) and Supershift Assays-Analysis of the proximal MAT2A promoter sequence identified several recognition sites for known transcription factors. These regions were subjected to further functional analysis using EMSA and supershift assays. A series of doublestranded 32 P-labeled oligonucleotide probes were generated by PCR amplification (Table I and Fig. 5). The amplified DNA representing the region from Ϫ285 to ϩ102 sequences was designated probe A. When Jurkat or Cos-1 cells nuclear extracts were incubated with probe A and analyzed by EMSA, three complexes I, II, and III were formed (Fig. 5). An additional complex IV was seen with nuclear extracts from Cos-1 cells. The shift pattern observed with probes C and D was very similar to that obtained with probe A. No complexes were formed with probe B (Ϫ16 to ϩ102), whereas probe F (Ϫ110 to Ϫ45) and probe G (Ϫ76 to Ϫ54) formed complexes that were very similar to probes A-D, except that complexes I and II seemed to comigrate (Fig. 5).
In competition experiments, probes D-G, which together cover the region from Ϫ110 to ϩ5, abolished all complex formation (Fig. 6). Oligonucleotides representing sequences from Ϫ59 to ϩ3, which contain the Sp1-1 site plus the Sp1-2 site, or representing sequences Ϫ59 to Ϫ29, which contain the Sp1-2 site alone, had no effect on the formation of complexes I, II, or III. In addition, a consensus Sp1 oligonucleotide, used as a competitor, reduced but did not completely block the formation of complexes. By contrast, when probe G (Ϫ76 to Ϫ54 bp), which contains the Sp1-3 site only, was used as a competitor, all complex formation was abolished (Fig. 6b). No competition was seen with sequences representing the region from Ϫ93 to Ϫ74 bp, which is upstream of the Sp1-3 site. Therefore, the data suggest that the major complexes are formed on the region from Ϫ76 to Ϫ54 bp, which contains the Sp1-3 site. Based on consensus sites for known transcription factors, initial supershift analyses were performed with antibodies to these putative factors, and they included anti-Sp1, anti-PEA3, anti-AP2, anti-GATA1, anti-CBF-C, and anti-c-Myb; however, only the anti-Sp1 antibody produced a supershift with probes A, C, and D, which contain all three Sp1 sites but not with probe E (Ϫ110 to Ϫ20), which contains sites Sp1-2 and Sp1-3),

FIG. 4. Promoter activity of the human MAT2A gene in Cos-1 cells and Jurkat human leukemic T cells.
A series of MAT2A promoter deletion constructs fused to the firefly luciferase reporter gene are depicted in A. Constructs were transfected into Cos-1 cells (A), or Jurkat and Cos-1 cells (B) by the method of Keller et al. (23). Luciferase activity was assayed by the dual luciferase reporter assay system as described under "Materials and Methods." The data are expressed in RLU, and are presented as a mean Ϯ S.E. of at least four separate experiments. All activities are presented as a percentage of the longest construct used, pGL3-MAT2A(Ϫ2374), which was set at 100% activity. or with probe G (Ϫ76 to Ϫ54 bp), which contains site Sp1-3 (Fig. 7). These data, together with the finding that an anti-Sp1 competitor reduced but did not abolish complex formation (Fig.  6c) suggested that factors other than Sp1, possibly other members of the Sp1 family, are binding to this region of the MAT2A promoter.
In contrast to the anti-Sp1 antibodies, anti-Sp2, -Sp3, and -Sp4 antibodies caused supershift with probe G (Ϫ76 to Ϫ54) that contains the Sp1-3 site only. The anti-Sp2 antibody induced a strong supershift when probes D-G were used, and in some experiments complex II appeared to be supershifted by the antibody. In the presence of the anti-Sp3 antibody, complexes I and III disappeared; complex III was either inhibited or shifted to comigrate with complex II. Thus, Sp3 appears to be involved in complex I and III formation. Supershift with the anti-Sp4 antibody was considerably less than with either anti-Sp2 or anti-Sp3 antibodies.
Together the above data suggest that Sp1 may not be directly involved in these complexes and that Sp2 and Sp3 (and possibly Sp4) are involved in complex formation on the region from Ϫ76 to Ϫ54. To further test this possibility, the area containing complexes I-III was excised from the EMSA gel, boiled in SDS buffer, and analyzed by Western blots using antibodies to various members of the Sp1 family of transcription factors. As shown in Fig. 8, Sp1 protein was not detected in complexes formed with probe E, whereas Sp2, Sp3, and Sp4 proteins were found in these complexes (Fig. 8). The anti-Sp3 antibody reacted with at least three major bands on the blot. Inasmuch as the nuclear extract included a mixture of protease inhibitors, it is less likely that the multiple bands are proteolytic products of Sp3; instead, it is possible that they represent internal isoforms of Sp3 that have been reported by others (33).
Effect of Mutating Sp1 Sites in the Proximal Region of the MAT2A Promoter on the in Vitro and in Vivo Promoter Activity-Taken together, the above data from EMSA and supershift assays indicate that the Sp1-3 site acts as a nucleation site for complex I-III formation and that Sp2 and Sp3 (and possibly Sp4), but not Sp1, are part of these complexes. To further investigate this notion, the three Sp1 sites contained within probe D were individually mutated as described under "Materials and Methods." Mutation of the Sp1-1 and Sp1-2 sites had little effect on complex formation, promoter activity, or supershift with anti-Sp1, -Sp2, -Sp3, or -Sp4 antibodies (Fig.  9). By contrast, mutation of the Sp1-3 site significantly diminished complex formation, and reduced promoter activity by 85% (Fig. 9). These results underscore the importance of the Sp1-3 site in controlling the expression of the MAT2A gene.

DISCUSSION
The level of expression of the MAT2A gene varies in different tissues and in different stages of differentiation or malignant transformation of the same tissue. Our previous studies have shown that MAT II is differentially expressed in normal and leukemic T lymphocytes (23)(24)(25). Variation in the expression of MAT II affects intracellular levels of AdoMet (24, 25, 34 -42), and this is likely to have an effect on a number of biological reactions involving this pivotal compound (9). To delineate the underlying mechanisms involved in regulating MAT2A gene expression, an in-depth characterization of the gene and its promoter was needed. Previous studies by Hiroki et al. (26) suggested that the first 343 bp of the 5Ј-flanking region of the rat MAT2A promoter contained consensus sequences for celltype-specific promoter elements; however, no studies of the interaction between specific nuclear factors and these elements were reported. Similarly, Mao et al. (27) reported a partial sequence of the human MAT2A promoter, but many of the FIG. 6. Competitive inhibition of complexes with various oligonucleotides and probes. Probe D was incubated with nuclear extracts from Jurkat T cells in the absence or presence of various competitors. Competition was conducted as detailed under "Materials and Methods." The competition pattern indicates that the major complexes are formed on the region between Ϫ76 and Ϫ54, which contains the Sp1-3 site (a and b). However, an Sp1 competitor reduced but did not abolish complex formation (c). regulatory aspects of this gene reported here were not included in their study.
In this study, we performed a detailed characterization of the proximal promoter of the MAT2A gene, and this has led to the identification of a site that is essential for its activity and can potentially be targeted for modulation of the expression of the MAT II ␣2 subunit in leukemic T cells. The organization of the human MAT2A gene was found to be quite similar to that of the human MAT1A gene (32) and the rat MAT2A gene (26). However, important differences in promoter activity and regulation were noted in this study. Sequences located 207 bases upstream from the translation start site or 87 upstream from the transcription start site were sufficient to drive promoter activity. Significant enhancement of activity was seen when sequences from Ϫ128 to Ϫ110, relative to the transcription start site, were included. In contrast to the rat gene and a recent report by Mao et al. (27), we found evidence for strong repressor elements contained between the region from Ϫ1281 and Ϫ325 of the human gene. This repressor activity was overcome by addition of sequences from Ϫ1881 to Ϫ1281.
Detailed analyses of the interaction of nuclear proteins from Cos-1 cells and the Jurkat human T cell leukemia cell line with the proximal promoter of the human MAT2A gene revealed a very similar pattern of promoter activity and complex formation in both cell types. Three complexes were formed with a probe representing the region from Ϫ285 to ϩ102. However, sequential deletions of this region showed that sequences between Ϫ76 and Ϫ54 are responsible for the major complexes seen.
A significant (p Ͻ 0.05) drop in activity was seen when sequences between Ϫ128 and Ϫ110 were deleted, suggesting that this region contains elements that enhance promoter activity. This region harbors a putative PEA3 site at Ϫ119, but antibodies to this factor did not affect complex formation and failed to cause a supershift. Similarly, the region from Ϫ110 to ϩ5 contained many sites for known transcription factors, including AP2, AP1, PEA3, AP-2, Sp1, GATA-1, CBF-C, c-Myb, and CAT-binding proteins, however, only antibodies to the Sp1 family of transcription factors affected the pattern of complex formation. Although this may indicate that putative sites for the other known factors are not functional, the possibility that other proteins are involved in these complexes or that functional interactions between known factors and their cognate sites may only take place in vivo should not be excluded and will be addressed in future studies.
Three Sp1 recognition sites are located within the region from Ϫ110 to ϩ5. However, complex formation was completely abolished either by competition with a probe representing sequences between Ϫ76 and Ϫ54 or by mutation of the Sp1-3 site (at Ϫ69). Mutation of either the Sp1-1 site (at Ϫ14) or the Sp1-2 site (at Ϫ47) did not affect complex formation and only slightly reduced promoter activity; whereas mutation of site Sp1-3 reduced promoter activity by ϳ85%. These data indicate that site Sp1-3 is important for driving MAT2A promoter activity.
Supershift with the anti-Sp1 antibody only occurred when all three Sp1 sites were present on the same DNA fragment and FIG. 8. Analysis of proteins bound to complexes I-III. 32 P-Labeled probe G representing sequences between Ϫ76 and Ϫ54 was incubated with nuclear extracts from Jurkat T cells and separated on native 4% acrylamide gels as described in the legend to Fig. 4. The region containing the radiolabeled protein-DNA complexes formed was excised from the gel and boiled in SDS sample buffer, loaded onto 10% SDSpolyacrylamide gel electrophoresis gels, and analyzed by Western blots using anti-Sp1, -Sp2, -Sp3, and -Sp4 antibodies. an Sp1 competitor reduced but did not abolish complex formation. These data suggested that Sp1 is probably not directly involved in binding to the Sp1-3 site. Sp1 is a member of a large family of related transcription factors (43,44). Sp3 and Sp4, and to a lesser extent Sp2, share extensive structural and sequence homology with Sp1 and to other members of the Sp family that share similar cognate DNA binding sites. Sp1 and Sp3 are ubiquitously expressed in cells, whereas the distribution of Sp2 and Sp4 appears to be more limited and cell-specific. The data presented here indicate that Sp2, Sp3, and possibly Sp4 are involved in complexes formed on the region between Ϫ76 and Ϫ54, which contains the Sp1-3 (GC box) site. Complex II appeared to be supershifted by anti-Sp2 antibodies, and complexes I and III disappeared in the presence of the anti-Sp3 antibody. Together, the findings that anti-Sp3 antibodies have the strongest effect on protein-DNA complexes and that this protein was found at a relatively high amount in these complexes suggest that it may play an important role in regulating MAT2A gene expression. Sp3 is a bifunctional protein that can both activate and repress transcription of genes (45,46). Internal isoforms of this protein containing activation and/or repressor domains have been described previously (33). Our data showed three major bands were recognized by the anti-Sp3 antibodies in Western blots of proteins bound to complexes formed on site Sp1-3. It is conceivable that several isoforms of Sp3 may be interacting at this site, but this remains to be determined. Studies are ongoing in our laboratory to further define the role of specific Sp proteins in regulating MAT2A gene expression, to identify other proteins that may be involved in complex formation at the Sp1-3 site, and to determine whether the DNA context of the Sp family cognate site confers specificity as well as affect the ability of the Sp proteins to regulate MAT2A gene expression (44).
In summary, detailed characterization of the MAT2A gene and its promoter revealed novel regions that exert positive and negative effects on gene expression. The Sp1-3 site located 69 bases upstream from the transcription start site acts as a nucleation site for complex formation. The DNA-protein complexes appear to involve several members of the Sp family of transcription factors, namely Sp2 and Sp3. The results of this study suggest that the Sp1-3 site is essential for promoter activity and can be targeted for potentiation of MAT2A expression. It should be mentioned that, in a recent study 2 the pattern of complex formation with nuclear extracts from normal resting and activated human T cells was starkly different from that seen with nuclear extracts from Jurkat leukemic T cells or from lymphocytes from a patient with acute lymphocytic leukemia. Interestingly, mutation of the Sp1-3 site abolished the formation of smaller complexes in normal cells while it resulted in the formation of a new major complex not seen in either fresh or established leukemic T cells. We believe that these findings support our hypothesis that differences in MAT2A gene regulation in normal and leukemic T cells exist and that this may account for the differential regulation of MAT II in these cells. Delineating the regulation of this key gene should shed light on the molecular basis for the differential expression of the ␣2 subunit of MAT II in normal and leukemic T cells.