Journal list menu
Regulation of transcription of the Dnmt1 gene by Sp1 and Sp3 zinc finger proteins
Abstract
The Sp family is a family of transcription factors that bind to cis-elements in the promoter regions of various genes. Regulation of transcription by Sp proteins is based on interactions between a GC-rich binding site (GGGCGG) in DNA and C-terminal zinc finger motifs in the proteins. In this study, we characterized the GC-rich promoter of the gene for the DNA methyltransferase (Dnmt1) that is responsible for methylation of cytosine residues in mammals and plays a role in gene silencing. We found that a cis-element (nucleotides −161 to −147) was essential for the expression of the mouse gene for Dnmt1. DNA-binding assays indicated that transcription factors Sp1 and Sp3 bound to the same cis-element in this region in a dose-dependent manner. In Drosophila SL2 cells, which lack the Sp family of transcription factors, forced expression of Sp1 or Sp3 enhanced transcription from the Dnmt1 promoter. Stimulation by Sp1 and Sp3 were independent phenomena. Furthermore, cotransfection reporter assays with a p300-expression plasmid revealed the activation of the promoter of the Dnmt1 gene in the presence of Sp3. The transcriptional coactivator p300 interacted with Sp3 in vivo and in vitro. Our results indicate that expression of the Dnmt1 gene is controled by Sp1 and Sp3 and that p300 is involved in the activation by Sp3.
Abbreviations
-
- ChIP
-
- chromatin immunoprecipitate
-
- ODN
-
- oligodeoxynucleotides
-
- HAT
-
- histone acetyltransferase.
Transcription is regulated by the combinational actions of proteins that bind to distinct promoter and enhancer elements. In general, a limited number of cis-acting DNA elements is recognized, not that by a single transcription factor exclusively but, rather, by a set of different proteins that are structurally related [1]. The promoter regions of many eukaryotic genes contain GC-rich sequences [2] and some of the most widely distributed promoter elements are GC boxes and related motifs [2].
The Sp family of transcription factors includes the proteins Sp1, Sp3 and Sp4, which recognize and bind to GC boxes as well as to GT/A-rich motifs with similar affinity, and Sp2, which binds preferentially to GT/A-rich sequences [2,3]. Sp1 and Sp3 are expressed in a wide variety of mammalian cells whereas Sp4 has been detected predominantly in neuronal tissues. The regulation of gene expression by Sp transcription factors is complex. Although certain promoters can be activated by either Sp1 or Sp3 in assays in vivo and are occasionally activated by both Sp1 and Sp3 that act in a synergistic manner [4–6], there are other promoters that show a definite preference for Sp1 or Sp3 [7]. Furthermore, Sp3 can function as an activator or a repressor of transcription, depending on the gene in question [8,9].
The genes for several mammalian activators and repressors of transcription have been cloned. The gene for p300 was first cloned as the gene for an E1A-associated protein with properties of a transcriptional adapter [10]. The protein was found later to possess intrinsic histone acetyltransferase (HAT) activity and to function as a coactivator in MyoD-, p53-, and SRC-1-mediated transcription [11,12]. Furthermore, p300 appears to play a critical role in progression of the cell cycle and the differentiation of cells [11,12].
The methylation of DNA plays a role in the regulation of gene expression [13,14], genomic imprinting [15] and inactivation of the X chromosome [16] and it has been shown to be essential for mammalian development [17,18]. Altered patterns of DNA methylation has been implicated in tumorigenesis [19]. However, the mechanisms by which DNA methylation is regulated during development and tumorigenesis remain largely unknown. Five distinct families of gene for DNA methyltransferases, designated Dnmt1, Dnmt2, Dnmt3a, Dnmt3b and Dnmt3L, have been identified in mammalian cells [20]. Dnmt1 is expressed constitutively in proliferating cells, it is associated with foci of DNA replication [21] and methylates CpG dinucleotides [22]. These findings are consistent with the hypothesis that Dnmt1 is a maintenance methyltransferase that restores appropriate patterns of DNA methylation to the genome shortly after DNA replication [23]. Representative sites for initiation of transcription have been found in the promoter of the Dnmt1 gene, namely, an oocyte-specific site, a somatic cell-specific site and a spermatocyte-specific site. In adult somatic cells, most of the available data indicate that the identification by Bigey et al. of many sites of initiation of transcription of Dnmt1 [24] might have been a mistake and that there is a single site only [23].
To understand the regulation of expression of the Dnmt1 gene in somatic cells, it is necessary to identify and characterize the promoter region of this gene. Previous studies showed that the 5′ end region of the Dnmt1 gene is typical of a TATA-less and GC-rich promoter [24]. However, specific cis- or trans-acting elements involved in the regulation of that promoter remain to be identified.
In this study, we identified a cis-element located between nucleotides −161 and −147 that appeared to be activated independently by Sp1 and Sp3. Moreover, the p300 coactivator appeared to be involved in the Sp3-mediated activation of the mouse Dnmt1 promoter in somatic cells.
Materials and methods
Cells, plasmids and materials
Mouse NIH3T3 cells and Drosophila SL2 cells were obtained from the JCRB Cell Bank (Tokyo, Japan) and from S. Kojima at RIKEN (Tsukuba, Japan). pCMV-Sp1 and pGEX-Sp1 were provided by R. Chiu of the UCLA School of Medicine (Los Angeles, CA, USA). Plasmids pPac, pPacSp1, pPacUSp3, pGEX-Sp3 and pCMV-Sp3 were gifts from G. Suske at Philipps-Universität (Marburg, Germany). GST–p300 (amino acids 1–596), GST–p300 (amino acids 744–1571) and GST–p300 (amino acids 1,572–2414) were obtained from Y. Shi of Harvard Medical School (Boston, MA, USA). Plasmid pCi-p300 was a gift form Y. Nakatani of the Dana Farber Cancer Research Institute (Boston, MA, USA). DNA fragments of the mouse Dnmt1 promoter were excised with appropriate restriction enzymes (D1 to D5 in Fig. 1A). Each DNA fragment was inserted into the Nhe1 and Xho1 sites of pGL3-basic (Promega Co., Madison, WI, USA). The integrating of all of the above recombinant plasmids was verified by sequence analysis. Antibodies against Sp1 (PEP2), Sp3 (D-20), AP-2 (C-18) and p300 (C-20) were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA).
Characterization of the promoter of the mouse Dnmt1 gene. (A) Transcriptional activity of the mouse Dnmt1 promoter in NIH3T3 cells. The top diagram shows the mouse Dnmt1 gene and the lower diagram shows the variously deleted promoters fused to a gene for luciferase (gray rectangle). The corresponding luciferase activities are also shown. Numbering is relative to the first site of initiation of transcription (+1). Reporter plasmid D5 has the deletion of nucleotides −173 to −120 of D2 fusion construct. NIH3T3 cells were transfected with plasmids that encoded the various constructs and luciferase activity was measured as described in the text. Promoter activities are expressed relative to the activity associated with reporter plasmid D4, 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. (B) Nucleotide sequence of the 5′ flanking region of the mouse gene for Dnmt1. Bases are numbered relative to the site of initiation of transcription (+1). P1, P2 and P3 indicate the probes used for EMSAs. (C) Binding of transcription factors to the Dnmt1 promoter. Binding of transcription factors to the region between nucleotides −173 and −117 of the Dnmt1 promoter was analyzed by EMSAs with 32P-labeled probes P1, P2, and P3 (1 × 104 c.p.m) and 5 µg protein of nuclear extract (NE) from NIH3T3 cells. Three shifted protein-DNA complexes are indicated by arrows (B1–B3).
Site-directed mutagenesis
Site-directed mutagenesis was performed with a QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) using various oligodeoxynucleotide primers (n1 to n10 in Fig. 2D) and a fragment of the Dnmt1 promoter (nucleotides −220 to +79) as template. The mutated DNA fragments were subcloned into the Nhe1 and Xho1 sites of pGL3-basic (Promega Co.). The integrity of all of the vectors was verified by sequence analysis.
Identification of a cis-element in the Dnmt1 promoter. (A) EMSAs were performed with mutated oligodeoxynucleotides as competitors and the wild-type ODN probe. (B) Oligodeoxynucleotides used as competitors are listed. Wt, Sequence of the wild-type probes. M1 through M11, the GAATTC sequence was introduced into the wild-type motif to generate the mutant sequences. (C) Assays of luciferase reporter activity in NIH3T3 cells using mutated reporter constructs. NIH3T3 cells were transfected with the reporter plasmid D2 or with plasmids that included the sequences n1 through n10 and luciferase activities were measured, and compared with that obtained with NIH3T3 cells that had been transfected with reporter plasmid D4 (500 ng). These assays were repeated at least three times and the standard deviation for each average value is indicated. (D) Nucleotide sequences obtained by site-directed mutagenesis. Wt, wild-type sequence; n1 through n10, TT di-nucleotides were introduced, as indicated, to generate the mutants.
Recombinant proteins
Glutathione S-transferase (GST), GST–Sp1, GST–Sp3, GST–p300 (amino acids 1–596), GST–p300 (amino acids 744–1571) and p300 (amino acids 1572–2414) were prepared as described previously [25]. 35S-Labeled Sp1 and Sp3 were synthesized in a TNT Reticulocyte Lysate System (Promega Co.), according to the manufacturer's protocol.
Cell culture, transfections and assays of promoter activity
Mouse NIH3T3 and F9 cells and human HeLa and 293 cells were grown in Dulbecco's modified Eagle's medium (Nissui Pharmaceutical Co., Tokyo, Japan) with 10% fetal bovine serum (Invitrogen BV, Groningen, the Netherlands) at 37 °C in a humidified atmosphere of 5% CO2 in air. Schneider's Drosophila SL2 cells were maintained in Shields and Sang M3 insect medium (Sigma–Aldrich Japan Co., Tokyo, Japan) supplemented with 10% fetal bovine serum at 22 °C in air. One day prior to transfection, mammalian cells were seeded in 12-well plates at 8 × 104 cells per well and SL2 cells were seeded at 1 × 106 cells per well. Cells were transfected with the indicated amount of reporter plasmid using Lipofectamine™ 2000 (Invitrogen BV). Cell extracts were prepared 24 or 48 h after transfection. Promoter activity was determined with a Luciferase Assay System (Promega Co.) as described by the manufacturer.
Electrophoretic mobility shift assays (EMSAs)
Nuclear extracts were prepared from NIH3T3, HeLa, 293 and F9 cells and EMSAs were performed as described previously [26]. DNA probes were radiolabeled with T4 polynucleotide kinase (New England BioLabs Inc., Beverly, MA, USA) and [γ-32P]ATP (Amersham Pharmacia Biotech., Uppsala, Sweden). The binding reaction was performed in 20 µL of buffer that contained 25 mmN-2-Hydroxyethylpiperazine-N′-2-ethanesulfonic acid/KOH (Hepes/KOH, pH 7.9), 25 mm KCl, 5 mm MgCl2, 50 mm ZnSO4, 1 µg poly(dI-dC), and a nuclear extract or purified GST-fusion protein. In some cases, competitors or antibodies were added to reaction mixtures which were then incubated on ice for 20 min. After addition of the 32P-labeled DNA probe, the mixture was incubated for a further 20 min on ice. Products of reactions were resolved on a 5% polyacrylamide gel in 0.5 × Tris/borate/EDTA buffer. Electrophoresis was performed at 180 V for 3 h at 4 °C.
Immunoprecipitation of chromatin (ChIP)
NIH3T3 cells were fixed in 1% formaldehyde at room temperature for 15 min. Chromatin was prepared with a kit from Upstate Biotechnology (Lake Placid, NY, USA) according to the recommendations of the manufacturer, with eight 10-s pulses of sonication at 10-s intervals, which yielded chromatin fragments of ≈ 1.0-kb in length. Equivalent amounts of chromatin were immunoprecipitated with the indicated antibody or with irrelevant IgG, as a control, at 4 °C for 5 h. Immunocomplexes were then recovered by addition of protein A/G PLUS-agarose beads (Santa Cruz Biotech., Santa Cruz, CA, USA) with incubation at 4 °C for 2 h. After the beads had been washed extensively, DNA was eluted and cross-linking was reversed by addition of 200 µL of elution buffer (1% SDS/0.1 m NaHCO3) and incubated overnight at 65 °C. DNA was extracted with phenol/chloroform (1 : 1, v/v), precipitated in ethanol and then analyzed by PCR with primers that corresponded to the cis-element (nucleotides −220 to +79), namely, 5′-AAGGCTAGCCAGAGTCA TCCTCTGC-3′ (forward direction) and 5′-GCGCTCG AGCTTGCAGGTTGCAGAC-3′ (reverse direction). PCR was performed for 35 cycles and products were analyzed by agarose gel electrophoresis.
Immunoprecipitation and Western blotting
Cell pellets were lysed in RIPA buffer (1× NaCl/Pi, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 100 µg·mL−1 phenylmethanesulfonyl fluoride, 1 mm sodium orthovanadate and 2 µg·mL−1 aprotinin]. Whole-cell extracts (500 µg proteins) were incubated with 2 µg of antibody [anti-Sp1 Ig, anti-Sp3 Ig, anti-(AP-2) Ig, anti-p300 Ig or nonimmunized rabbit IgG] for 1 h, and then 30 µL of a suspension of protein A/G PLUS-agarose beads (Santa Cruz) were added. After incubation for 1 h at 4 °C, immunoprecipitates were gently washed three times with NaCl/Pi, boiled and subjected to SDS/PAGE (7% acrylamide gel). Proteins were electroblotted onto a poly(vinylidene difluoride) membrane filter and blocked for 1 h in Blotto A [10 mm Tris/HCl (pH 8.0), 150 mm NaCl, 5% skim milk, 0.05% Tween-20], and then incubated with 2 µg·mL−1 of primary antibody (anti-Sp1 Ig or anti-Sp3 Ig) in BlottoA for 1 h. Finally, 30 µL of a solution of horseradish peroxidase conjugated-secondary antibody (New England Biolabs) in Blotto A were added. Antibody–HRP complexes were detected with ECL Western blotting detection reagents (Amersham Pharmacia Biotech.) according to the instructions from the manufacturer.
`GST-pull down' assay
Two micrograms of GST–protein and 35S-labeled Sp3 (3 × 103 c.p.m) that had been in vitro translated in a final volume of 500 µL of binding buffer [250 mm NaCl, 50 mm Hepes/KOH (pH 7.5), 0.5 mm EDTA, 0.1% Nonidet P-40, 0.2 mm phenylmethanesulfonyl fluoride, 1 mm dithiothreitol and 500 µg·mL−1 BSA], were incubated at 4 °C for 1 h and then 30 µL of a suspension of glutathione–Sepharose 4B (Amersham Pharmacia Biotech) were added. After incubation for 1 h at 4 °C, samples were gently washed three times with NaCl/Pi, boiled and fractionated by SDS/PAGE (7% acrylamide gel).
Results
Identification of cis-elements that control expression of the Dnmt1 gene
We subcloned a 2.0-kb DNA fragment that contained the Dnmt1 promoter region from a mouse genomic clone into pGL3-basic, a vector that includes a gene for luciferase without any eukaryotic promoter or enhancer elements. We then generated a series of 5′-serial-deletion constructs of the promoter–luciferase gene for transfections and subsequent assays of luciferase activity (Fig. 1A). Constructs D1 to D5, together with the pGL3-basic vector, were used for transient transfection of NIH3T3 cells in an attempt to identify the cis-elements of the gene for Dnmt1 and to delineate the 5′ boundary of the promoter. As shown in Fig. 1A, weak control of promoter activity was associated with a region of 1827 bp in the upstream region between nucleotides −2000 and −173, with no more than 40% variation in activity. A large reduction of approximately sixfold in promoter activity was detected when we removed nucleotides −173 to −120 (Fig. 1A, D3 to D5), suggesting that cis-acting elements that are critical for Dnmt1 promoter activity might be located in this region. We detected a single site for initiation of the transcription of Dnmt1 at +1 in this promoter (data not shown), a result that is consistent with previous reports [20,23]. Thus, it appeared that the region from nucleotides −173 to −120 contained critical cis-elements for transcriptional activation of the Dnmt1 gene.
To examine DNA-binding proteins, we prepared DNA probes from the promoter of the mouse gene for Dnmt1 (nucleotides −173 to −117), which contained putative binding sites for Sp1, and performed gel shift assays as described previously [26]. Three DNA probes, P1 (nucleotides −173 to −138), P3 (nucleotides −160 to −131) and P2 (nucleotides −141 to −117) were prepared (Fig. 1B). EMSAs of nuclear extracts from NIH3T3 cells with the P1 probe revealed shifts of three bands on the gel (B1 to B3 in Fig. 1C). The intensity of these bands was significantly enhanced with P3 probe and no DNA–protein complexes were evident when we analyzed nuclear extracts of NIH3T3 cells with the P2 probe. We observed three similar shifted bands when we used nuclear extracts from 3T6 cells, HeLa cells, 293 cells and F9 cells (data not shown). To define the binding site in this region more precisely, we synthesized 11 mutant oligodeoxynucleotides (ODNs; M1 to M11 in Fig. 2B) and used them as competitiors in EMSAs (Fig. 2A). The DNA–protein complexes B1, B2, and B3 were detected with the wild-type P1 probe. After addition of a 100-fold excess of unlabeled mutant ODN (M1 to M11) to the reaction mixture, we found that shifted bands were not eliminated by the unlabeled mutant ODNs M5 through M8. The nucleotide sequences of these ODNs were compared and the consensus-binding site was defined as the 15-bp sequence 5′-GGCAAGGGGGAGGTG-3′ (Fig. 2B), which we designated the GA motif. To examine whether or not this sequence is critical for the transcriptional activity of the Dnmt1 promoter, we synthesized 10 mutant ODNs (n1 to n10 in Fig. 2D) and introduced them into luciferase constructs to generate the respective reporter plasmids. We examined the activity of these luciferase reporters in NIH3T3 cells and found that only the n5 construct lacked transcriptional activity (Fig. 2C). Thus, the central GG dinucleotide in the 15-bp motif seemed to be critical for the transcriptional activity of the Dnmt1 promoter.
Sp1 and Sp3 bind to the GA motif in the Dnmt1 promoter
To identify the transcription factors that bind to the GA motif in the Dnmt1 promoter, we looked for transcription factors in the TRANSFAC database [27]. We failed to identify known factors that bind to this sequence. However, we found that both AP-2 and Sp1 bound to sequences that exhibited strong similarity, namely 0.859 and 0.856, respectively, to the GA motif. To determine whether Sp1 and/or AP-2 could bind to the GA consensus sequence, we performed EMSAs in the presence of antibodies against Sp1, Sp3 and AP-2 (Fig. 3A). The retarded band designed B1 was shifted even further upon addition of antibodies against Sp1 (lane 3), while the retarded bands corresponding to B2 and B3 were shifted further upon addition of antibodies against Sp3 (lane 4). Both antibodies against Sp1 and against Sp3 affected the migration of all retarded bands. In contrast, anti-(AP-2) Ig did not affect the migration of shifted bands (lane 6). Thus, it appeared that both Sp1 and Sp3 bound to the 5′-GGCAAGGGGGAGGTG-3′ sequence in the Dnmt1 promoter in vitro. We also performed a chromatin immunoprecipitation experiment to assess the interaction of Sp1 and Sp3 in the transcription of the Dnmt1 gene at the chromatin level (Fig. 3B). In contrast to the absence of any effect of IgG, immunoprecipitates obtained with antibodies both against Sp1 and against Sp3 yielded a 318-bp DNA fragment after PCR that included the 15-bp sequence. Antibodies against AP-2 did not generate the 318-bp DNA (data not shown). These results supported our hypothesis that both Sp1 and Sp3 bind to cis-elements in the gene for Dnmt1 that include the GA motif.
Sp1 and Sp3 bound to a cis-element in the promoter of the mouse gene for Dnmt1. (A) 32P-Labeled probe P3 (1 × 104 c.p.m) was incubated with 5 µg protein of nuclear extract (NE) from NIH3T3 cells in the presence and absence of antibodies against Sp1, Sp3 or AP-2. Lane 3, Sp1-specific antibody (2 µg); lane 4, Sp3-specific antibody (2 µg); lane 5, Sp1-specific and Sp3-specific antibodies (2 µg each); lane 6, AP-2-specific antibody (2 µg). (B) Chromatin immunoprecipitation assays. Chromatin immunoprecipitation assays were performed as described in the text. DNA and proteins were cross-linked with formaldehyde, and DNA was sheared and immunoprecipitated with Sp1- or Sp3-specific antibody (2 µg·mL−1 each). After reversal of cross-links, DNA was amplified with primers specific for the promoter region of the Dnmt1 gene. Products of PCR were resolved by agarose gel electrophoresis. The arrowhead indicates the amplified DNA fragment (318 bp).
Stimulation of transcription from the Dnmt1 promoter by Sp1 and Sp3 in SL2 cells
Both Sp1 and Sp3 bound to the Dnmt1 promoter. Therefore, we next examined the effects of Sp1 and Sp3 on the expression of the gene for Dnmt1. We transfected Drosophila SL2 cells with a luciferase reporter construct in the presence and absence of the expression plasmids pPacSp1 and pPacUSp3, respectively. SL2 cells lack endogenous Sp proteins, and, thus, the cis-element-dependent activation of the Dnmt1 promoter is dependent on the gene products of exogenously introduced genes that encode Sp1 or Sp3. The activity associated with the luciferase reporter construct D2 was stimulated in the presence of pPacSp1 (Fig. 4A) and in the presence of pPacUSp3 (Fig. 4B). Moreover, the promoter activity of D2 was further enhanced in the presence of both pPacSp1 and pPacUSp3 (Fig. 4E). In contrast, the activity associated with the luciferase reporter construct n5 was not stimulated by either pPacSp1 or pPacUSp3 (Fig. 4C,D). These results indicate that both Sp1 and Sp3 enhanced transcription from the Dnmt1 promoter.
Activation of transcription from the Dnmt1 promoter by Sp1 and Sp3 in Drosophila SL2 cells. SL2 cells were transfected with 500 ng of the Dnmt1 reporter plasmid D2 (A,B,E) and the reporter plamid n5 without GA motif (C,D) and the indicated amounts of pPacSp1 (A,C), pPacUSp3 (B,D) or both pPacSp1 and pPacUSp3 (E) (each 500 ng) and then luciferase activities were measured. The total amount of the plasmid DNA (pPacSp1 or pPacUSp3) was adjusted to 1 µg with pPac (no insert). Assays were repeated at least three times and the standard deviation for each mean value is indicated.
Independent activation of the Dnmt1 promoter by Sp1 and Sp3 through the same GA motif
Sp1 and Sp3 bound to the same cis-element in the promoter of the gene for Dnmt1. Therefore, we next examined whether activation by Sp1 or by Sp3 affect expression of the gene for Dnmt1. In an attempt to identify whether Sp1 and Sp3 could bind to the same GA motif, we prepared appropriate GST fusion proteins and performed EMSAs with the P3 probe that included the GA motif. Specific and different shifted bands were detected with GST–Sp1 and GST–Sp3. The shifted band due to GST–Sp3 gradually disappeared when increasing amounts of GST–Sp1 were added as competitor (Fig. 5A,B). In contrast, the intensity of the shifted band that included GST–Sp1 gradually increased. Conversely, the addition of GST–Sp3 resulted in a similar reduction in the binding of Sp1 to the GA motif (data not shown). These results imply that the binding of Sp1 and of Sp3 to the GA motif were independent phenomena and that each competed for binding to DNA with the other through the same cis-element.
Binding of Sp1 and Sp3 to the cis-element in the promoter of the Dnmt1 gene. (A) EMSAs of GST-Sp1 and GST-Sp3 fusion proteins with the P3 probe. Indicated amounts of GST-Sp1 (onefold to 50-fold excess) were added to a reaction mixture that included 0.1 µg of GST-Sp3 and the P3 probe (0.2 nmol·mL−1). (B) Relative binding of Sp1 and Sp3 to P3 in EMSA. The results in (A) are summarized as relative DNA-binding activity (%). The total intensity of P3 probe on a reaction mixture for EMSAs was taken arbitrarily as 100%. (C) Sp1 and Sp3 do not form a stable complex in NIH3T3 cells. Immunoblotting of Sp1 in the immunoprecipitates derived from 500 µg protein of whole-cell extracts of NIH3T3 cells with Sp3-specific antibody (2 µg·mL−1). Sp1 did not form a complex with Sp3 in NIH3T3 cells. Input: this lane was loaded 50 µg protein of whole-cell extracts.
We next examined whether Sp1 might be included in transcriptional complexes with Sp3 in vivo. Western blotting analysis with antibodies against Sp1 of immunoprecipitates obtained with antibodies against Sp3 indicated that Sp1 was not immunoprecipitated with Sp3 from extracts of NIH3T3 cells (Fig. 5C). Complementary studies of immunoprecipitates obtained with Sp1-specific antibodies and Western blotting with antibodies against Sp3 also indicated that Sp1 and Sp3 did not interact with each other (data not shown). We also examined the molecular association of GST–Sp1 and GST–Sp3 fusion proteins and found no evidence of any association in vitro (data not shown).
Enhancement by p300 of transcription from the Dnmt1 promoter that is induced by Sp3
The transcriptional coactivator p300 mediates growth arrest by catalyzing histone acetylation and the subsequent rearrangement of chromatin [11,12]. Recent reports indicate that p300 also collaborates with Sp1 or Sp3 to regulate the expression of the promoter of the gene for p21Waf1/Cip1[28,29]. Therefore, we examined the effect of p300 on the promoter activity of the Dnmt1 gene in the presence of pCMV-Sp1 and of pCMV-Sp3 in NIH3T3 cells. As shown in Fig. 6, cotransfection with pCi-p300 and pCMV-Sp3 enhanced the reporter activity controlled by the Dnmt1 promoter, but cotransfection of pCi-p300 with pCMV-Sp1 did not. The extent of activation was much higher than that obtained with pCi-p300 and with pCMV-Sp3, indicating that p300 enhanced the promoter activity of the Dnmt1 gene that was induced by Sp3, but not by Sp1. Further studies of transactivation using a GAL4 fusion with p300 and the dominant negative form of p300 are required for a full understanding of the molecular mechanism of this phenomenon.
Enhancement of the Dnmt1 promoter activity upon cotransfection with p300- and Sp3-expression plasmids. NIH3T3 cells were transfected with 0.5 µg of the Dnmt1 gene reporter plasmid D2 (see Fig. 1A) plus 0.5 µg of pCMV-Sp1, pCMV-Sp3 or/and pCi-p300 and luciferase activity was measured in each case. Total amounts of DNA (pCMV-Sp1, pCMV-Sp3 and pCi-p300) were adjusted to 2 µg with pBSII KS(+). The assay was repeated at least three times and the standard deviation for each value is indicated.
Sp3 interacts with the C-terminal region of p300
To determine whether Sp3 associates directly with p300, we performed immunoprecipitation and Western blotting assays using antibodies against p300 and Sp1 or Sp3 and extracts of NIH3T3 cells. We immunoblotted immunoprecipitates obtained with p300-specific antibodies with antibodies against Sp1 or against Sp3. As shown in Fig. 7A, a p300-specific band was detected with antibodies against Sp3 but not against Sp1. Antibodies specific for AP-2 (data not shown) and control IgG did not yield any evidence of interactions with Sp3. To determine which regions of p300 interacted with Sp3, we prepared 35S-labeled Sp3 by translation in vitro and investigated the binding to various GST–p300 fusion proteins by the GST-pull down assay [25]. Only the C-terminal region of p300 (amino acids 1572–2414), which included the C/H region and E1A-binding region of p300, was found to associate with Sp3 (Fig. 7B). We performed similar complementary experiments to determine which parts of p300 interacted with Sp3 and found that the [35S]Met-labeled carboxyl region of p300 interacted with GST–Sp3 (data not shown). These results indicated that Sp3 was able to bind to the C-terminal domain of p300.
p300 was associated with Sp3. (A) p300 associated with Sp3 and not with Sp1 in NIH3T3 cells. p300 was immunoprecipitated with anti-p300 Ig (2 µg) from 500 µg protein of whole-cell extract of NIH3T3 cells and the immunoprecipitate was immunoblotted with antibodies against Sp3 (2 µg·mL−1; upper panel) or Sp1 (2 µg·mL−1; lower panel) as described in the text. (B) Direct interaction of Sp3 with various GST–p300 deletion mutants. 35S-Labeled Sp3 (3 × 103 c.p.m) was incubated with 2.0 µg of each GST–p300 mutant, as indicated (lanes 3–5), or with 2.0 µg GST alone (lane 2). Lane 1, one tenth of input 35S-labeled Sp3. After electrophoresis, radiorabeled protein was detected by autoradiography. The shaded boxes of p300 indicate the C/H1 domain, C/H2 domain and C/H3 domain [10].
Discussion
The promoters of many housekeeping genes have a number of common characteristics, such as the presence of multiple sites for initiation of transcription which, presumably, compensate for the absence of a TATA box and a CAAT box, and they often have an unusual high GC-content [1]. The Dnmt1 gene is also a housekeeping gene [24]. However, details of the cis-elements in its promoter and the factors that regulate expression of the Dnmt1 gene from the minimum promoter in somatic cells remain to be determined. In this report, we identified a cis-element in the promoter of the Dnmt1 gene in somatic cells and showed that both Sp1 and Sp3 bound to this regulatory element, the GA motif, independently. Moreover, both Sp1 and Sp3 stimulated the promoter activity of the Dnmt1 gene and Sp3 was found to associate with p300 through the C-terminal region of the latter protein and to enhance its activity.
In order to identify the minimal promoter in a 2.0-kb region of the Dnmt1 gene, we generated a series of promoter–luciferase deletion constructs for reporter assays and then determined the luciferase activity associated with each respective construct (Fig. 1A). The results indicated that a cis-acting region of 53-bp was located between nucleotides −173 and −120. Precise dissection of the region by EMSAs demonstrated that a minimal element, from nucleotides −161 to −147, was critical for the activation of transcription of the Dnmt1 gene (Fig. 2A). We identified a GA motif in this region required for the binding of SP1 and SP3 to the DNA (Fig. 3A,B) in order to activate transcription of the Dnmt1 gene (Fig. 4). EMSAs in the presence of anti-Sp1 Ig and anti-Sp3 Ig confirmed these results (Fig. 3A). The GA motif is different from the typical sequence of Sp1-binding sites, GGGCGG (Fig. 2). Thus, Sp1- and Sp3-binding sites might be affected by sequences adjacent to GC-rich or GA-rich elements that influence maximum binding. It has been reported that Sp1 and Sp3 might be involved in the activation of a very large number of genes, such as housekeeping genes for tissue-specific and cell cycle-regulated proteins [3]. Moreover, Sp proteins are involved not only in activation but also in repression. We studied the effects on the expression of the Dnmt1 gene by Sp1 or Sp3, which bound to the similar elements (Fig. 3). The two proteins are found in the same cells and are indistinguishable in terms of DNA-binding specificity (Fig. 5A,B), and both proteins bound independently to the GA motif (Fig. 5C). All Sp proteins contain three zinc fingers close to the C-terminus, with glutamine-rich domains adjacent to serine/threonine structures in the N-terminal region. Sp1, Sp3 and Sp4 are more closely related to each other than they are to Sp2 [3]. The homology among the zinc fingers of all known Sp proteins is close to 90%, but, the homology among entire sequences is close to 40%.
In the transcriptional activation of the Dnmt1 gene, both Sp1 and Sp3 play a critical role (Fig. 4) via binding to the same cis-element (GA motif) and the binding of each is independent of the other (Fig. 5). A competition experiment in vitro with the GA motif demonstrated that each factor is able to replace the other in terms of binding to DNA.
The growth characteristics of Sp1-deficient embryonic stem cells (ES cells) are normal and such cells can be induced to differentiate [30]. Nevertheless, Sp1 is essential for normal mouse embryogenesis and the development of Sp1-knockout embryos is severely retarded, with death occurring around day 11 of gestation. Thus, Sp1 appears to be a transcription factor whose function is essential after day 10 of development. Other Sp proteins, such as Sp3, might be able to compensate, at least in part, for the loss of Sp1 activity at early embryonic stages. Sp3 is expressed ubiquitously and has the potential to activate transcription. Moreover, its DNA-binding activity is indistinguishable from that of Sp1. A recent study of Sp3 null-mice suggested that Sp1 and Sp3 might have similar and therefore redundant functions during early development but might have distinct and highly specific functions at later stages of development [31]. Thus, Sp1 and Sp3 might have a wide range of redundant functions and might be able to replace each other in Sp1-and Sp3-knockout mice [31]. It was reported that three isoforms of Sp3 exist and that these different isoforms play different roles in the activation and repression of transcription [9]. In our hands, Sp1 generated a single band and Sp3 generated two bands in EMSAs (two of three bands migrated to the same position; B3 in Fig. 3A). It has been suggested that two small isoforms of Sp3 might act as repressor molecules with full-length Sp3 acting as an activator [3,9]. Sp3 acts as a transcriptional activator at many promoters, as does Sp1 [5,32]. In studies of other promoters such as the uteroglobin gene [8], monocyte chemoattractant protein-1 gene and ornithine decarboxylase gene [7,33], however, Sp3 was found to be inactive or to act as only a weak activator or as a repressor of Sp1-mediated transcription.
The most obvious differences between Sp1 and Sp3 are the presence of a potent inhibitory domain in Sp3 [34]. The relative abundance of Sp1 and Sp3 allows the fine tuning of the regulation of gene activities. In endothelial cells that contain high levels of both Sp1 and Sp3, the ratio of Sp1 to Sp3 is higher than in nonendothelial cells [35]. In primary keratinocytes, levels of Sp3 exceed those of Sp1. The ratio of Sp3 to Sp1 is inverted when these cells differentiate in vitro. In differentiating keratinocytes, only Sp3 enhances the activation of the promoter of the gene for p21Waf1/Cip1[36]. A change in the ratio of Sp1 to Sp3 also occurs when C2C12 myocytes are cultivated under hypoxic conditions. Hypoxia causes the progressive depletion of Sp3, whereas the level of Sp1 remains unchanged [37]. It has been demonstrated that the expression of the gene for Dnmt1 is regulated in a cell-cycle-dependent manner. The expression of Dnmt1 is enhanced during the S phase and then declines with the approach of the M phase. In a preliminary study, we found that levels of expression of Sp1 and Sp3 differed during the cell cycle; Sp1 was expressed predominantly at the G1 phase and Sp3 was expressed at S phase (data not shown). Therefore, it is quite plausible that Sp1 and Sp3 might control the expression of the Dnmt1 gene. Further studies are required to clarify the distinct roles of Sp1 and Sp3 at different phases of the cell cycle.
The transcriptional cofactor p300 is coprecipitated in complexes with Sp1 [38]. The activation of the promoter of the gene for p21Waf1/Cip1 by butyrate and nerve growth factor requires functional collaboration between Sp1 and Sp3 [28,38]. However, although p300 and Sp1 are components of the complex that activates the promoter of the gene for p21Waf1/Cip1, the interaction is indirect. Thus, p300 is required for the tricostatin A-induced (TSA-induced), Sp1-mediated transcription of the gene, but details of the interaction between Sp1 and p300 in this phenomenon are unknown. It is possible that Sp3 might bind to a GC or GA motif in the promoter of the gene for p21Waf1/Cip1 through binding with p300, as such motifs might bind Sp1 and Sp3. Thus, Sp3 might regulate the TSA-dependent activity of the promoter. We showed that Sp3 might associate with the C-terminal region of p300 by direct binding (Fig. 7B). This region is similar to the region to which GATA-1, E2F and p53 also bind [10]. Thus, regulation by Sp1 and Sp3 of target genes might be involved in the cell cycle. Furthermore, the activity of p300 is dependent on gene dosage during early embryogenesis. Thus, it is possible that Sp3 might compete for binding to DNA with Sp1-like proteins and Krüppel-like factors via interactions with p300 at different phases of the cell cycle in somatic cells. In conclusion, we propose that the nucleotide sequences to which Sp1 and Sp3 bind are similar and binding of these factors is independently regulated by different coactivators in a cell-cycle-dependent manner. The distinct functions of Sp1 and Sp3 in the regulation of expression of the Dnmt1 gene during the cell cycle remain to be clarified.
Acknowledgements
The authors thank Drs G. Suske, R. Chiu, Y. Shi, Y. Nakatani, S. Kojima, H. Ugai, C. Jin, J. Song and A. Wolff for plasmids, reagents and valuable discussions. This work was supported by the Special Coordination Funds of RIKEN, by grants from the Ministry of Education, Science, Sports, Culture and Technology of Japan (to K. K. Y) and a grant from the Program for Promotion of Basic Research Activities for Innovative Biosciences (to K. S).
