The human POLD1 gene. Identification of an upstream activator sequence, activation by Sp1 and Sp3, and cell cycle regulation.

The promoter of the human POLD1 gene encoding the catalytic subunit of DNA polymerase δ is G/C-rich and does not contain a TATA box. Transient transfection analysis in HeLa cells employing POLD1-luciferase chimeric plasmids revealed a core promoter region extending 328 base pairs (bp) from the major transcription initiation site. Multiple elements in this region including two 11-bp direct repeats located between nucleotide positions −92 and −22, play an important role in POLD1 promoter activity. Deletion or linker-replacement mutations of the repeats drastically reduced the promoter activity. A 70-bp DNA fragment containing the two repeats could stimulate the expression of the POLD1 or a heterologous promoter in an orientation-independent manner. DNase I footprinting and band-shift assays showed that HeLa nuclear extracts contained proteins specifically binding to the repeat sequences. Southwestern blot and UV cross-linking analyses identified Sp1 and two 85-kDa proteins that bound to the repeats. Additionally, screening of HeLa cDNA expression libraries for the sequence-specific DNA-binding protein using the 11-bp repeat sequences as the probe, identified a cDNA that corresponds to Sp3, a member of the Sp1 family. Cotransfection studies in Drosophila SL2 cells showed that both Sp1 and Sp3, but not Sp2, could activate the POLD1 promoter through the repeat sequences. The POLD1 promoter activity was induced about 4-fold at the late G1/S boundary in serum-stimulated cells. The 11-bp repeats together with an E2F-like sequence, located adjacent to the major transcription initiation site, were important for the stimulation. Taken together, this study provides a direct evidence for transcriptional regulation of the human POLD1 gene.

The promoter of the human POLD1 gene encoding the catalytic subunit of DNA polymerase ␦ is G/C-rich and does not contain a TATA box. Transient transfection analysis in HeLa cells employing POLD1-luciferase chimeric plasmids revealed a core promoter region extending 328 base pairs (bp) from the major transcription initiation site. Multiple elements in this region including two 11-bp direct repeats located between nucleotide positions ؊92 and ؊22, play an important role in POLD1 promoter activity. Deletion or linker-replacement mutations of the repeats drastically reduced the promoter activity. A 70-bp DNA fragment containing the two repeats could stimulate the expression of the POLD1 or a heterologous promoter in an orientation-independent manner. DNase I footprinting and band-shift assays showed that HeLa nuclear extracts contained proteins specifically binding to the repeat sequences. Southwestern blot and UV cross-linking analyses identified Sp1 and two 85-kDa proteins that bound to the repeats. Additionally, screening of HeLa cDNA expression libraries for the sequence-specific DNA-binding protein using the 11-bp repeat sequences as the probe, identified a cDNA that corresponds to Sp3, a member of the Sp1 family. Cotransfection studies in Drosophila SL2 cells showed that both Sp1 and Sp3, but not Sp2, could activate the POLD1 promoter through the repeat sequences. The POLD1 promoter activity was induced about 4-fold at the late G 1 /S boundary in serum-stimulated cells. The 11-bp repeats together with an E2F-like sequence, located adjacent to the major transcription initiation site, were important for the stimulation. Taken together, this study provides a direct evidence for transcriptional regulation of the human POLD1 gene.
The DNA replication of eukaryotic chromosomes is a complex but highly regulated process. Through the cooperation of multiple protein factors and enzymes including DNA polymerases, each chromosome replicates once during the S phase of the cell cycle (reviewed in Ref. 1). Presently, the mechanisms underlying this cell cycle regulation of DNA replication are not completely understood.
DNA polymerase ␦ (pol ␦) 1 is one of the major enzymes involved in the synthesis of mammalian nuclear DNA (2,3). It was reported as a new type of DNA polymerase with an intrinsic 3Ј to 5Ј exonuclease activity (4), suggesting that it possesses exonucleolytic proofreading ability (5,6). Purified pol ␦ is composed of a 125-kDa catalytic subunit and an associated 48-kDa small subunit whose function has not been defined (7). In addition, a 36-kDa factor was shown to convert the pol ␦ activity from low to high processivity (8). This factor was subsequently shown to be identical to the proliferating cell nuclear antigen (9,10).
Previously, we, in collaboration with Lee's group (11), and others (12) isolated the full-length cDNA for the catalytic subunit of human pol ␦. In addition, the coding sequences of the pol ␦ catalytic subunit have been isolated from bakers' yeast (13), fission yeast (14), malaria parasite (15,16), calf thymus (17), and mouse cells (18). Sequence analysis showed that the catalytic subunit of human pol ␦ contains 1,107 amino acids and is related to other eukaryotic DNA polymerases (11). Recently, we also cloned the gene for the catalytic subunit of human pol ␦ (POLD1) and its 5Ј-flanking sequences (19). The human POLD1 gene contains 27 exons and 26 introns. Transcription of the gene appears to be initiated at multiple sites with the major initiation site 53 nucleotides upstream of the ATG start codon. As found in the promoters of many mammalian housekeeping genes (for reviews, see Refs. 20 -22), the POLD1 promoter is G/C-rich and does not contain a TATA box. Several potential binding sites for transcription factors AP2, CTF, Ets1, GCF, MBF-1, NF-E1, and Sp1 are present in the 5Ј-flanking region of the POLD1 gene (19). However, the significance of these elements in regulating POLD1 promoter activity is presently not known.
Studies on the expression of the genes involved in nucleotide metabolism and DNA synthesis reveal the presence of multiple regulatory elements including the GC boxes, which are required for maximal promoter activity (23)(24)(25)(26)(27)(28)(29)(30). The transcription factor Sp1 (31,32) has been shown to regulate the transcription through the GC boxes. Recently, several Sp1-related proteins, Sp2, Sp3, and Sp4, that bind to the GC box or GT motif were identified (33)(34)(35); however, their role in promoter regulation is less defined. Sp3 and Sp4 share a high degree of amino acid sequence homology with Sp1 (33). The high degree of structural conservation among Sp1, Sp3, and Sp4 suggests that these transcriptional regulators act through binding to similar DNA elements. Like Sp1, both Sp3 and Sp4 can bind to the GC box or GT motif, and Sp4 can activate several Sp1responsive promoters (34,36,37). However, the Sp4 mRNA is detected only in brain (33). On the other hand, Sp3 is ubiquitously expressed and can either activate or repress transcription (34, 36, 38 -40). In addition, Sp3 can repress Sp1-mediated activation (34,36,39). Another Sp1-family member, Sp2, appears to share less homology with other members and has not been shown to act as an activator or a repressor (35). In addition, the binding affinity of Sp2 to the GT motif is much weaker than that of Sp1 and Sp3. Nonetheless, it has been hypothesized that these Sp1-family transcription factors may play a concerted role in regulating different GC or GT sequence in the promoter (36). Since the POLD1 promoter contains potential Sp1-binding sites, it would be important to examine if the Sp1-family proteins can regulate POLD1 transcription.
POLD1 expression is induced in serum-stimulated cells, consistent with its involvement in DNA replication (11). Northern blot analysis of RNA from populations of human Molt 4 cells separated by counterflow centrifugal elutriation showed that POLD1 mRNA content is regulated during the cell cycle (41). Similarly, expression of the gene for the human pol ␣ catalytic subunit is stimulated upon serum addition to starved cells, and multiple elements in its promoter appear to mediate the full serum response (30). Presently, the response of the POLD1 promoter to serum induction has not been studied. No serumresponsive element (42) or DNA replication-related element, identified in the Drosophila genes for pol ␣ and proliferating cell nuclear antigen (43), is present in the POLD1 upstream region. Thus, it is not known which elements in the POLD1 gene are responsible for serum induction and cell cycle regulation.
To understand the regulation of the human POLD1 gene, we examined the cis-acting sequences controlling its expression in cycling and serum-stimulated cells. We identified two 11-bp direct repeats located in the 5Ј promoter proximal region, which could function as an activator sequence. We also found that Sp1, Sp3, and two 85-kDa proteins could bind to the repeats, and showed that both Sp1 and Sp3 could activate the POLD1 promoter. In addition, we reported that expression of the POLD1 promoter was induced at the G 1 /S boundary and the induction was mediated through the 11-bp repeat sequences and an E2F-like sequence located near the major transcription initiation site.

MATERIALS AND METHODS
Plasmid Construction-The pGL2-␦(Ϫ1758) plasmid (previously named pGL2-␦P1), containing the 1.8-kb 5Ј-flanking DNA of the POLD1 gene linked to the luciferase expression cassette from pGL2-basic (Promega), has been described (19). A series of 5Ј-unidirectional deletion mutations within the POLD1 promoter DNA of pGL2-␦(Ϫ1758) were created using the Erase-a-Base system (Promega) or the Bal31 nuclease (44). Some 5Ј and 3Ј deletion mutants of the POLD1 promoter were constructed by deleting appropriate restriction fragments or by polymerase chain reaction (PCR; Ref. 45). The linker replacement mutants within the 11-bp repeat region were constructed by ligating an XhoI linker to the proper deletion clones created by Bal31 deletion or by PCR. To construct the pGL2-HSVtk basal plasmid, the EcoRI-PstI fragment of the minimal herpes simplex virus thymidine kinase (HSVtk) promoter containing the TATA sequence and 60-bp upstream DNA from pMC1 (Stratagene) was isolated and ligated to the luciferase expression cassette. The chimeric promoter plasmids, pGL2-␦R2R1-HSVtk basal and pGL2-␦R1R2-HSVtk basal , were created by inserting the 70-bp PstI-SacII fragment of the POLD1 promoter DNA 5Ј to the HSVtk basal promoter in the sense and antisense orientation, respectively. The exact locations of the deletion end points or appropriate fusions in all of these plasmids were determined by sequencing with POLD1-specific primers or primers derived from the pGL2 vector (46).
The Sp1 expression plasmid, pPacSp1 (47), was generously provided by Dr. Robert Tjian of University of California at Berkeley. It contains a 2.6-kb Drosophila actin 5C promoter, a 700-bp Ubx cDNA fragment consisting of the 5Ј untranslated region (UTR) plus the first eight codons of the Ubx open reading frame (ORF), a 2.1-kb Sp1 cDNA fused with the Ubx ORF, and a 1.1-kb 3Ј UTR of the actin 5C gene providing the polyadenylation (poly(A)) signal sequence. Expression plasmids for Sp2 and Sp3 were generated based on pPacSp1 (47). Briefly, to construct the Sp2 expression plasmid, a BamHI linker was inserted immediately 5Ј to the Sp2 translation initiation codon in the Sp2 cDNA plasmid, Sp2-pKS(Ϫ) (American Type Culture Collection; Ref. 35), and an XhoI linker was inserted into the SacI site at nucleotide position 2062. The BamHI-XhoI fragment containing the entire Sp2 coding region was excised and used to replace the BamHI-XhoI Sp1 fragment in pPacSp1 to generate the pPacSp2 plasmid. To construct the Sp3 expression plasmid, an Sp3 cDNA clone was isolated from the screening of a cDNA expression library (see below). A 2.1-kb EcoRI-XhoI Sp3 cDNA fragment was excised and used to replace the EcoRI-XhoI Sp1 cDNA in pPacSp1 to generate the pPacSp3 plasmid.
Cell Culture, Transfection, Promoter Activity, and Western Blot Analysis-Human HeLa cells and mouse C3H10T 1/2 cells (American Type Culture Collection) were routinely grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS). Drosophila Schneider line (SL2) cells (48) were grown in Schneider medium (Life Technologies, Inc.) supplemented with 10% FBS at 25°C. Transfection assays were performed by the calcium phosphate technique (49) with slight modifications (50). Each POLD1-luciferase plasmid (20 g) was transfected into a 100-mm dish of actively growing HeLa cells. To standardize the transfection efficiency, 2 g of the pCH110 plasmid (Pharmacia Biotech Inc.), containing the SV40 early promoter driving ␤-galactosidase gene, were also included as an internal control in each transfection. Calcium phosphate-DNA coprecipitate mixtures were prepared and added to cells in the presence of medium. Following overnight incubation, the medium was replaced with fresh growth medium. At 48 h after transfection, cells were harvested in reporter lysis buffer and the luciferase and ␤-galactosidase activities were measured using the reporter assay system (Promega). For all transfection assays, at least three independent experiments were performed. The luciferase activity expressed from each promoter construct was normalized to the ␤-galactosidase activity. For transfection of Drosophila SL2 cells, 5 g of each POLD1-luciferase plasmid and various amounts of the Sp1, Sp2, or Sp3 expression plasmid were used. The total amounts of DNA were compensated to 20 g with the control pPacU vector (47). An equal amount of cell extracts from each transfection was used for measuring the luciferase activity. To detect the expression of the Sp1-family protein, the same amount (80 g) of protein extracts from each transfection of SL2 cells was electrophoresed in an 8% SDS-polyacrylamide gel and electroblotted onto a PVDF-Plus filter (Micron Separations) using a Hoefer Transphor electrophoresis unit (44). The filter slice containing extracts from cells transfected with each Sp1-family expression plasmid was stained with the corresponding antibody against each expressed protein (anti-Sp1 (IC6), anti-Sp2 (K-20), or anti-Sp3 (D-20) antibody from Santa Cruz), followed by a secondary antibody conjugated with alkaline phosphatase using the ProtoBlot AP system (Promega).
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts from HeLa spinner cells (American Type Culture Collection) were prepared as described (51). Briefly, HeLa cell nuclei were extracted with 0.35 M KCl and nuclear extracts were dialyzed against the buffer containing 10 mM HEPES, pH 7.9, 100 mM KCl, 20% glycerol, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, and 0.5 mM dithiothreitol. The protein concentration of the extract was determined by using the Bio-Rad protein assay dye reagent and was approximately 5 mg/ml. For preparing nuclear extracts from transfected SL2 cells, a rapid micropreparation method was used (52). SL2 cells were transfected with 5 g of the Sp1, Sp2, or Sp3 expression plasmid, and nuclear extracts were prepared 48 h after transfection. EMSA (53) was performed according to the Gel Shift Assay System (Promega) with minor modifications. Briefly, various amounts of nuclear extracts were incubated with 1 g of poly(dI-dC)⅐poly(dI-dC) (Pharmacia) in 10 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM DTT, 50 mM NaCl, and 4% glycerol in a total volume of 18 l. After incubation at room temperature for 5 min, 2 l of 32 P-labeled double-stranded DNA probe (0.8 ng) was added and the mixture incubated for an additional 20 min. The reaction products were electrophoresed in a 5% polyacrylamide gel (acrylamide:bisacrylamide ratio ϭ 60:1). The probes used were the 70-bp PstI-SacII fragment (position Ϫ92 to Ϫ23 relative to the major transcription initiation site) containing the two 11-bp repeats of the POLD1 promoter or the linker replacement mutants with one or both copies of the repeats replaced by the XhoI linker. For antibody supershift experiments, 1 g of the anti-Sp1 or anti-Sp3 antibody (Santa Cruz) was added to the binding reaction described above, and the reaction mixture was incubated overnight at 4°C prior to electrophoresis.
DNase I Footprinting Analysis-The POLD1 promoter DNA containing the sequence between Ϫ261 and ϩ49 relative to the major transcription start site was isolated from pGL2-␦(Ϫ261) by digestion with KpnI and HindIII. This substrate DNA was treated with calf intestinal alkaline phosphatase (Boehringer Mannheim), gel-purified, and 5Ј endlabeled with T4 polynucleotide kinase or 3Ј end-labeled with Klenow fragment of E. coli DNA polymerase I (44). The end-labeled DNA was digested with XhoI to remove 21-bp DNA from the distal end of the POLD1 promoter. The resulting DNA fragment, labeled only at the proximal end, was purified using a quick-spin G50 column (Boehringer Mannheim). The purified probe (0.3 pmol) was incubated with 100 g of HeLa nuclear extracts in a volume of 50 l containing 25 mM Tris-HCl, pH 8.0, 50 mM KCl, 6.25 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM DTT, and 10% glycerol, at room temperature for 10 min. 50 l of a solution containing 5 mM CaCl 2 and 10 mM MgCl 2 were added and incubated at room temperature for another 1 min. Subsequently, the reaction mixture was treated with DNase I at the concentration between 1.5 and 6 units (or 0.15 and 0.6 unit for the control DNA without nuclear extracts) at room temperature for 1 min. The reaction was terminated by adding 90 l of the stop solution containing 200 mM NaCl, 30 mM EDTA, 1% SDS, and 100 g/ml yeast tRNA. The DNase I-digested DNA was purified by phenol extraction and ethanol precipitation, and then electrophoresed in an 8% polyacrylamide gel containing 8 M urea. Sequencing G ladders were generated by methylating the probe with dimethyl sulfate (Aldrich), followed by cleavage with piperidine (Sigma).
Southwestern Blot Analysis-Nuclear extracts from HeLa (75 g) or transfected SL2 cells (80 g) were electrophoresed in an 8% SDSpolyacrylamide gel. Fractionated proteins in the gel were then electrotransferred to a PVDF-Plus filter (Micron Separations). After transferring, the filter was prehybridized with a 100-ml hybridization buffer containing 10 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM DTT, 50 mM NaCl, and 5% nonfat dry milk at room temperature for 30 min, and then rinsed twice with the hybridization buffer containing 0.25% nonfat dry milk. Subsequently, the filter was hybridized with an end-labeled probe in 25 ml of hybridization buffer containing 0.25% nonfat dry milk and 250 g of poly(dI-dC)⅐poly(dI-dC) at room temperature for 2 h. The hybridized filter was washed four times with the hybridization buffer containing 0.25% nonfat dry milk, blot-dried, and exposed overnight to an x-ray film.
UV Cross-linking Analysis-The 70-bp POLD1 promoter DNA fragment containing the two 11-bp repeats was end-labeled using Klenow fragment in the presence of bromodeoxyuridine triphosphate (Sigma) and [␣-32 P]dCTP (44). The labeled probe was incubated with 12.5 g of HeLa nuclear extracts in a volume of 20 l to allow binding to reach equilibrium as described above for EMSA. Each reaction mixture was then pipetted onto a parafilm and irradiated for various times at a distance of 5 cm using an Ultraviolet Transilluminator with a 310 nm filter (Ultra-Lum). Samples were not treated with nucleases since the substrate DNA was labeled at its 3Ј ends (32). Cross-linked complexes were analyzed by electrophoresis in an 8% SDS-polyacrylamide gel and visualized by autoradiography.
Screening of HeLa cDNA Expression Libraries-Two HeLa cell cDNA expression libraries constructed in either the Uni-ZAP™ (Stratagene) or EXlox R (Clontech) vector were screened for clones encoding proteins that can bind to the POLD1 DNA containing the 11-bp repeats as described by Singh et al. (54) and Vinson et al. (55). The probe used was a trimer of the 70-bp POLD1 fragment (Ϫ92 to Ϫ23 relative to the major transcription initiation site), which was 32 P-labeled using the nick translation kit (Boehringer Mannheim). About 10 6 phages from each library were plated out. After filter lift, without the treatment to denature and renature the proteins, the filters were prehybridized, hybridized, and washed using the same conditions as those for the Southwestern blot analysis. One positive clone was obtained from the HeLa cDNA library in Uni-ZAP™. The cDNA insert from the positive clone was rescued according to the In Vivo Excision Protocol (Stratagene) and sequenced.
Synchronization and Flow Cytometry Analysis-For synchronization experiments, C3H10T 1/2 cells were plated out at a density of 5 ϫ 10 5 cells/100-mm dish and transfected the next day with 20 g of each POLD1-luciferase plasmid plus 2 g of the control pCH110 plasmid as described previously (50). Calcium phosphate-DNA coprecipitates were added to cells in the presence of medium overnight. Transfected cells were then washed twice with Ca 2ϩ -and Mg 2ϩ -free phosphate-buffered saline (PBS) and changed to DMEM supplemented with 0.1% FBS. After arresting at low serum for 3 days, transfected cells were restim-ulated with growth medium containing 10% FBS. At various times after serum stimulation, cells were harvested and the same amount of protein extracts was used for reporter enzyme assays.
For flow cytometry, synchronized cells were harvested by trypsinization at various time points, washed in PBS containing 1 mM EDTA, pH 8.0, and 1% bovine serum albumin, and then fixed and permeabilized with ice-cold 70% ethanol. After washing with PBS containing 1% bovine serum albumin, fixed cells were incubated in propidium iodide (PI) staining solution (10 g/ml of PI and 0.5 mg/ml of RNase A) at 37°C for 30 min. and then 4°C in the dark for at least 1.5 h. Flow cytometry analysis was performed on a Coulter EPICS Elite flow cytometer. For detection of PI, stained cells were excited at 488 nm with a 15-milliwatt air-cooled argon laser and fluorochrome emission was absorbed with a 610-nm long pass filter. Extended analysis of DNA content was performed on Multicycle software (Phoenix Flow System).

RESULTS
The Core Promoter Region of the Human POLD1 Gene Is Located within the 328-bp 5Ј-Flanking DNA-Previously, we showed that the 1.8-kb 5Ј-flanking DNA of the human POLD1 gene could direct the expression of the luciferase reporter gene when transfected into actively growing HeLa cells (19). In order to define the cis-acting elements required for the transcription of the POLD1 promoter, a series of unidirectional deletion mutants were created within the POLD1 promoter segment of the pGL2-␦(Ϫ1758) plasmid ( Fig. 1). The promoter activity of each deletion mutant was then evaluated by luciferase assay of extracts from transfected HeLa cells. The experiment was repeated more than three times, and the results are shown in Fig.  1. By reference to the major transcription start site as ϩ1, deletions of the 5Ј upstream sequence between nucleotide positions Ϫ1758 and Ϫ328 did not affect or slightly reduced the promoter activity (compare pGL2-␦(Ϫ328) to pGL2-␦(Ϫ1758) in Fig. 1). On the other hand, a 3Ј deletion to position Ϫ93 (pGL2-␦(Ϫ1758 ϳ Ϫ93)) reduced the promoter activity to about 20% of the 1.8-kb promoter activity. Another 3Ј deletion to position Ϫ247 (pGL2-␦(Ϫ1758 ϳ Ϫ247)) further reduced the promoter activity to about 14%. Further 3Ј deletion to position Ϫ322 (pGL2-␦(Ϫ1758 ϳ Ϫ322)) almost completely abolished the promoter activity. These results indicate that the core promoter region of the human POLD1 gene is located within the Ϫ328 region of its 5Ј-flanking DNA.
Further 5Ј deletion analysis showed that deletions between positions Ϫ328 and Ϫ92 gave rise to slightly reduced promoter activities (Fig. 1). Deletion to position Ϫ92 (pGL2-␦(Ϫ92)) decreased the promoter activity to about 34% of that of the 1.8-kb promoter. Further deletion to position Ϫ22 (pGL2-␦(Ϫ22)) drastically reduced the promoter activity to less than 2% of that produced by the 1.8-kb promoter. Deletion to position Ϫ10 (pGL2-␦(Ϫ10)) reduced the promoter activity to 1% of the 1.8-kb promoter, while deletion to position Ϫ2 (pGL2-␦(Ϫ2)) reduced the promoter activity to 0.3%, similar to that of pGL2-Basic (Fig. 1). These results suggest that while multiple regions are requires for full promoter activity, the sequence between nucleotides Ϫ92 and Ϫ22 plays an essential role in the expression of the human POLD1 gene.
Two 11-bp Direct Repeat Sequences Are Essential for the POLD1 Promoter Activity-To examine the regulatory element(s) within the Ϫ92 to Ϫ22 region, a series of fine deletions within the POLD1 promoter segment of the pGL2-␦(Ϫ92) plasmid were constructed and used to test their promoter activities as described above ( Fig. 2A). By comparison to the promoter activity expressed from pGL2-␦(Ϫ92), deletion to position Ϫ81 did not affect the promoter activity. In contrast, deletion to position Ϫ75 or Ϫ68 decreased the promoter activity to about 46% and 28% of that expressed by pGL2-␦(Ϫ92), respectively. However, further deletion to position Ϫ58 resulted in partial recovery of the promoter activity from 28% to 50%. These result suggest that a sequence located between positions Ϫ81 and Ϫ68 positively regulates the POLD1 promoter expression, while there is a negative regulatory sequence located between positions Ϫ68 and Ϫ58. Further analysis showed that an additional deletion to position Ϫ48 resulted in a similar level of promoter activity as that of the Ϫ58 deletion ( Fig. 2A). In contrast, deletion to position Ϫ40 led to a large reduction (7%) of the promoter activity and deletion to position Ϫ22 decreased the promoter activity to about 5% of that produced by pGL2-␦(Ϫ92). These results suggest the existence of another positive regulatory sequence located around position Ϫ48 to Ϫ40. Interestingly, examination of the sequences within the two positive regulatory regions revealed the presence of two identical 11-bp sequences (5Ј-GGGGCGTGGCC-3Ј, located at position Ϫ81 to Ϫ71 and position Ϫ49 to Ϫ39), arranged in the same orientation in each region ( Fig. 2A).
To test if the two 11-bp repeats are essential for POLD1 promoter activity, several linker replacement mutations, with an XhoI linker replacing part of the 11-bp repeat sequence in pGL2-␦(Ϫ92), were constructed and tested (Fig. 2B). As compared to the promoter activity expressed from pGL2-␦(Ϫ92), mutants with a linker replacing either the left (␦(Ϫ92)M2R1-1) or right half (␦(Ϫ92)M2R1-2) of the 5Ј copy of the 11-bp repeat sequence, reduced the promoter activity by about 2-fold. Similarly, a mutant with a linker replacing the middle portion of the 3Ј copy of the 11-bp repeat (␦(Ϫ92)R2M1) also reduced the promoter activity by about 2-fold. Interestingly, mutations in both copies of the 11-bp repeats resulted in drastic reduction of the promoter activity by 10 -12-fold (␦(Ϫ92)M2M1-1 and ␦(Ϫ92)M2M1-2 in Fig. 2B). Additionally, the promoter activities of these double replacement mutants were similar to that of the deletion mutant without both repeats (compared to pGL2-␦(Ϫ40) in Fig. 2A). All of these results indicate that the two 11-bp repeats are important transcriptional regulatory elements for the POLD1 promoter.
The POLD1 11-bp Direct Repeats Can Enhance Promoter Activity in an Orientation-independent Manner-The role of the two 11-bp direct repeats as a positive regulatory sequence was further investigated by placing them upstream of a heterologous promoter. A 70-bp POLD1 DNA containing the two repeats was inserted in both orientations into the plasmid containing the luciferase reporter gene controlled by the HSVtk basal promoter (Fig. 3). The resulting plasmids were transfected into HeLa cells and tested for their promoter activities. As expected, the HSVtk basal promoter expressed relatively low activity since it contained only the TATA sequence and 60-bp upstream DNA (compare pGL2-HSVtk basal to pGL2-␦(Ϫ1758) in Fig. 3). Insertion of the two 11-bp repeats in the sense orientation upstream of the HSVtk basal promoter resulted in stimulation of the expression by about 16-fold (compare pGL2-␦R2R1-HSVtk basal to pGL2-HSVtk basal ). Interestingly, insertion of the two repeats in the antisense orientation upstream of the HSVtk basal promoter also stimulated the expression by about 7-fold (compared pGL2-␦R1R2-HSVtk basal to pGL2-HSVtk basal ). These results indicate that the two 11-bp direct repeats can function as an activator on a heterologous promoter in an orientation-independent manner. Consistently, the 70-bp POLD1 DNA containing the two direct repeats also stimulated the POLD1 promoter activity when inserted in the antisense orientation (compare pGL2-␦(Ϫ92)R1R2 to pGL2-␦(Ϫ22)). However, it should be noted that the two 11-bp direct repeats, when placed in the original orientation, appeared to activate the POLD1 or HSVtk basal promoter to a higher extent than when inserted in the opposite orientation (Fig. 3).
Specific Nuclear Proteins Bind to the POLD1 Promoter DNA-To localize the protein binding sites on the POLD1 promoter DNA, DNase I footprint analysis was performed us- ing crude HeLa nuclear extracts and a 310-bp POLD1 DNA (nucleotide position Ϫ261 to ϩ49) as the substrate (Fig. 4). When the sense strand of the POLD1 DNA was examined, the nuclear extracts protected three regions (Fig. 4A). Protected region I extended from position Ϫ117 to Ϫ97. Analysis of the sequence in this region revealed the presence of a putative Sp1-binding site. Protected region II extended from position Ϫ85 to Ϫ66 and protected region III extended from position Ϫ52 to Ϫ34. Remarkably, the 11-bp activator sequence is located at the center of the protected region II and III (designated R2 and R1; see Fig. 4A). Analysis of the antisense strand of the POLD1 DNA demonstrated protection of the same three regions (Fig. 4B). These results indicate that the specific nuclear proteins bind to the POLD1 promoter DNA, including the two 11-bp repeats.
To corroborate the results of the DNase I footprint analysis and to examine the specificity of the binding proteins, electrophoretic mobility shift assays were carried out using HeLa nuclear extracts and the 70-bp POLD1 DNA (position Ϫ92 to Ϫ23, designated R2R1), containing the two 11-bp repeat sequences (but not the upstream Sp1 site detected by the DNase I footprint analysis) from pGL2-␦(Ϫ92), as the probe (Fig. 5). With low amounts of nuclear extracts, one major (designated C I ) and one minor (designated C IV ) protein-DNA complexes were observed (Fig. 5A, lanes 2 and 3). With increasing amounts of nuclear extracts, four complexes, designated C I -C IV , were identified. With excess amount (12 g) of nuclear extracts, the C IV complex became the predominant species, while the C I complex became the minor species (lane 4). The specificity of the protein-DNA interaction was indicated by the disappearance of the shifted bands in the presence of a 50-fold molar excess of unlabeled R2R1 DNA (lane 6). In contrast, incubation with a 50-fold excess of a DNA fragment containing the POLD1 initiator region (position Ϫ22 to ϩ49, designated DI) did not result in any decrease of the intensity of the shifted species (lane 7).
To further examine the characteristics of the four complexes, the 70-bp DNA fragments containing one or two mutated copies of the 11-bp repeat from the linker replacement mutants (␦(Ϫ92)M2R1-2, ␦(Ϫ92)R2M1, and ␦(Ϫ92)M2M1-2 in Fig. 2B) were used in the EMSA. When the M2R1 DNA with the linkerreplacement mutation in the 5Ј copy of the 11-bp repeat sequence was used as the probe, the C I complex was detected as the major species even when excess amounts of HeLa nuclear extracts were used (Fig. 5A, lanes 8 -11). Similarly, when the R2M1 DNA with a mutation in the 3Ј copy of the 11-bp repeat sequence was used as the probe, the C I complex remained as the major species, while some slow-migrating complexes could also be detected when excess amounts of nuclear extracts were used (lanes 12-15). These results suggest that the C I complex was formed only when one copy of either repeat was occupied by the protein. In contrast, no protein-DNA complex was detected when the M2M1 DNA containing mutations in both copies of the repeat sequences was used (lanes 16 -19). All of these results indicate that the 11-bp repeats in the POLD1 promoter are bound by specific nuclear proteins.
Since the left half of the 11-bp repeat sequence (5Ј-GGGGCGTGGCC-3Ј) resembles the consensus Sp1-binding site (5Ј-GGGCGG-3Ј; Refs. 32 and 56), we also conducted an antibody supershift experiment to determine if the specific protein-DNA complexes detected in EMSA contained the Sp1 protein. By adjusting the amount of HeLa nuclear extracts and the R2R1 DNA as the probe, four protein-DNA complexes similar to those observed above (Fig. 5A) were detected (Fig. 5B). Addition of an anti-Sp1 antibody resulted in supershifting of most, but not all, of these complexes. When either M2R1 or R2M1 DNA was used as the probe, addition of the anti-Sp1 antibody also led to supershifting of most of the C I complex (Fig. 5B). The addition of an anti-E2F antibody as a control (50) had no effect on the complexes (data not shown). Taken to- gether, the results indicate that specific nuclear proteins, including Sp1, can bind to the 11-bp repeat sequence in the POLD1 promoter DNA.
A Doublet of the 85-kDa Nuclear Proteins Can Also Bind to the Two 11-bp Repeats-To further characterize the 11-bp repeat-binding proteins, Southwestern blot analysis of HeLa nuclear extracts was carried out using the same 70-bp R2R1 DNA containing the two repeats as the probe (Fig. 6). In addition, a double-stranded 22-bp oligonucleotide with an Sp1 site (5Ј-ATTCGATCGGGGCGGGGCGAGC-3Ј; Ref. 57) was also used as a probe in a comparative analysis. When the R2R1 DNA was used, a doublet of major DNA-binding proteins with molecular masses of around 85 kDa and a minor DNA-binding protein of 105 kDa were detected (Fig. 6A, lane 1). However, when the Sp1 site-containing oligonucleotide was used, a major DNAbinding protein of 105 kDa, which corresponds to the size of the reported Sp1 protein (57), was identified (Fig. 6A, lane 2). Note that this Sp1 protein co-migrated with the 105-kDa band detected by the R2R1 probe (compare lane 2 to lane 1 in Fig. 6A), suggesting that the 105-kDa band is the Sp1 protein. In addition, the Sp1 oligonucleotide also detected a minor band of around 53 kDa in the analysis.
We also conducted a Southwestern blot experiment using the mutant POLD1 DNA probes containing one or two mutated copies of the 11-bp repeat from the linker replacement mutants, like those used in the previous EMSA (Fig. 5). In contrast to R2R1, the M2R1 DNA with the linker-replacement mutation in the 5Ј copy of the 11-bp repeat sequence or the R2M1 DNA with the mutation in the 3Ј copy only detected the 105-kDa protein (Fig. 6B). In addition, the intensity of this 105-kDa band detected by these single-replacement mutant DNAs was much stronger than that detected by the wild-type R2R1 DNA. When the M2M1 DNA containing mutations in both copies of the repeats was used, only the 105-kDa band with reduced intensity, was observed (Fig. 6B). These results suggest that the binding of the 85-kDa proteins requires both copies of the 11-bp repeat sequences.
Because Southwestern blot employs a denaturation step, some DNA-binding proteins sensitive to this treatment might not be detected in this analysis. To exclude this possibility and to confirm the results obtained from the Southwestern blot analysis, UV cross-linking experiments were conducted. The same 32 P-labeled R2R1 DNA containing the two 11-bp repeats was mixed with HeLa nuclear extracts and binding was allowed to reach equilibrium. The mixture was then subjected to UV irradiation for various time and the cross-linked material analyzed by electrophoresis. A doublet of major bands migrating at about 110 kDa and a minor band with slightly higher molecular mass were detected (Fig. 7). The pattern of these bands is similar to that observed in the Southwestern blot analysis (Fig. 6), albeit with slower mobilities due to the presence of the probe DNA in the cross-linked complexes. A band with molecular mass of around 64 kDa was also seen. The intensity of all of these bands increased linearly with increasing time of irradiation, consistent with a simple interaction between the activated DNA and closely associated proteins. In addition, production of all of these bands could be inhibited by competition with the unlabeled homologous DNA (data not shown). Taken together, these results indicate that in addition to Sp1, the 85-kDa proteins in HeLa nuclear extracts can bind specifically to the 11-bp repeat sequences of the POLD1 promoter DNA. Sp1 and Sp3, but Not Sp2, Can Activate the POLD1 Promoter-To identify the gene encoding the protein that binds to the 11-bp repeat sequences, we screened HeLa cDNA expression libraries with a concatenated 70-bp R2R1 DNA, which contains the two 11-bp repeats as the probe. One positive clone containing a 2.1-kb cDNA insert was obtained, and sequence analysis showed that the cDNA sequence is identical to that of Sp3, a member of the Sp1 family (33,35). The sequence of the 2.1-kb cDNA covers almost the entire Sp3 ORF (33,35), starting at position ϩ46, relative to the first nucleotide of the published Sp3 cDNA sequence designated as ϩ1 (33), and ending at ϩ2109, 15 bp downstream of the translation termination codon.
To test if the Sp1-family proteins can activate the POLD1 promoter, expression plasmids for Sp1, Sp2 (35), and Sp3 were FIG. 5. Interaction of the POLD1 promoter DNA with HeLa nuclear proteins. A, detection of multiple protein-DNA complexes. EMSA was conducted as described under "Materials and Methods." A 70-bp POLD1 DNA containing the two 11-bp repeats was used as the probe. R2R1 contains two wild-type copies of the repeats, M2R1 contains one linker-replacement mutation in the 5Ј copy of the repeats, R2M1 contains one linker-replacement mutation in the 3Ј copy of the repeats, and M2M1 contains two mutated copies of the repeats. DI containing the sequence from the POLD1 initiator region (nucleotide position Ϫ22 to ϩ49) was used as the nonspecific competitor. The protein-DNA complexes formed, are marked as C I, C II, C III, and C IV. The location of the free probe is also indicated. B, antibody supershift experiments. The anti-Sp1 antibody was added to the binding reaction as described above, and the reaction mixture was incubated overnight at 4°C prior to electrophoresis. The amount of the antibody used is indicated. constructed (Fig. 8A) and used in cotransfection of Drosophila SL2 cells, which lack endogenous Sp factors and have been useful in studying the function of the Sp1-family proteins (34,47,58). When the POLD1-luciferase construct pGL2-␦(Ϫ92) containing the two 11-bp repeats was cotransfected with increasing amounts of each Sp1-family expression vector, both Sp1 and Sp3, but not Sp2, stimulated the POLD1 promoter activity (Fig. 8B). Maximal stimulation was detected when 5 g of the Sp1-or Sp3-expression plasmid were used. To determine if the stimulation of the POLD1 promoter activity by Sp1 and Sp3 was mediated through the two 11-bp repeats, the POLD1luciferase plasmid with (pGL2-␦(Ϫ92)) or without (pGL2-␦(Ϫ22)) the two repeats was cotransfected with 5 g of each Sp1-family expression plasmid into SL2 cells. In addition, the pGL2-Basic plasmid was used as the vector control. As expected, only the Sp1-and Sp3-expression plasmid stimulated the promoter activity of pGL2-␦(Ϫ92) containing the two repeats (Fig. 8C). Deletion of the two 11-bp repeats (pGL2-␦(Ϫ22)) drastically reduced the ability of Sp1 or Sp3 to stimu-late the POLD1 promoter activity. Intriguingly, both Sp1 and Sp3 also stimulated the luciferase activity expressed by the control pGL2-Basic plasmid (Fig. 8C), suggesting that this reporter plasmid contains cryptic promoter elements responsive to Sp1 and Sp3 stimulation in Drosophila SL2 cells. Nonetheless, by comparing to the level of stimulation for each reporter construct, the luciferase activity of the pGL2-␦(Ϫ22) plasmid without the two 11-bp repeats was stimulated to the extent similar to that of the pGL2-Basic plasmid by Sp1 or Sp3, while in the presence of the two repeats (pGL2-␦(Ϫ92)) a much larger stimulation of the luciferase activity was observed (Fig.  8C). It should be pointed out that all three Sp1-family proteins were expressed at the similar level in transfected SL2 cells (Fig. 8D). Taken together, these results indicate that Sp1 and Sp3, but not Sp2, can activate the POLD1 promoter through the two 11-bp repeats.
To examine if the stimulation of the POLD1 promoter activity by Sp1 and Sp3 correlates with their ability to bind to the 11-bp repeat sequence, nuclear extracts from transfected SL2 cells expressing each Sp1-family protein were tested for the binding ability to the POLD1 R2R1 DNA. As compared with the protein-DNA complexes detected in HeLa extracts (Fig. 9A,  lanes 2 and 3), extracts from SL2 cells transfected with the control pPacU vector gave rise to two fast-migrating complexes with relatively low intensity (Fig. 9A, lane 4), representing some endogenous binding activities from the Drosophila cells.
On the other hand, extracts from Sp1-or Sp3-expressing SL2 cells produced multiple slow-migrating complexes, which migrated to similar positions as those seen in HeLa extracts (Fig.  9A, lanes 5 and 6). In contrast, extracts from Sp2-expressing SL2 cells did not give rise to any slow-migrating complexes and only showed endogenous binding activities (Fig. 9A, lane 7).
Two additional experiments were carried out to confirm that the slow-migrating complexes are resulted from the binding of the POLD1 R2R1 DNA by Sp1 or Sp3 expressed in transfected SL2 cells. First, an antibody supershift experiment using the anti-Sp1 or anti-Sp3 antibody was conducted. Similar to those observed in HeLa extracts (Fig. 5A), multiple protein-DNA complexes were detected with increasing amounts of extracts prepared from Sp1-or Sp3-expressing SL2 cells (Fig. 9B). Addition of the anti-Sp1 antibody to the binding reaction containing Sp1-expressing SL2 extracts led to supershift all of the slow-migrating complexes to form higher-ordered complexes (compare lane 4 with lane 5 in Fig. 9B). Similarly, addition of the anti-Sp3 antibody to the binding reaction containing Sp3-expressing SL2 extracts also led to supershift all of the slowmigrating complexes to form higher-ordered complexes (compare lane 8 with lane 10 in Fig. 9B). Second, a Southwestern blot using the R2R1 DNA as the probe was also conducted. When extracts from Sp1-expressing SL2 cells were used, a 75-kDa DNA-binding protein was detected (Fig. 9C), consistent with the calculated molecular mass from the Sp1 ORF in the expression plasmid. Note that the molecular mass of this SL2 cell-expressed Sp1 is smaller than that of the reported Sp1   FIG. 8. Trans-activation of the POLD1 promoter by the Sp1-family proteins. A, schematic diagram of the expression plasmids for Sp1, Sp2, and Sp3. The cDNA of Sp1, Sp2, or Sp3 was inserted into the pPacU vector containing the Drosophila actin 5C promoter, the Ubx 5Ј UTR (leader), and the actin 5C poly(A) signal sequence (47). B, activation of the POLD1 promoter by Sp1 and Sp3, but not Sp2. The POLD1-luciferase plasmid pGL2-␦(Ϫ92), which contains the two 11-bp repeats but not the upstream Sp1 site, were cotransfected with various amounts of the expression plasmid for Sp1 (छ), Sp2 (E), or Sp3 (Ç) into Drosophila SL2 cells. The pPacU vector was also used as a background control. The level of activation of the POLD1 promoter by each Sp1-family protein is calculated as the luciferase activity expressed from cells cotransfected with pGL2-␦(Ϫ92) and the Sp1-family expression plasmid divided by that from cells transfected with pGL2-␦(Ϫ92) and pPacU. C, the POLD1 11-bp repeats are required for the activation by the Sp1-family proteins. The POLD1-luciferase plasmid pGL2-␦(Ϫ92) containing the two repeats (represented as the solid bar), the pGL2-␦(Ϫ22) plasmid without the two repeats (represented as the shaded bar), or the control pGL2-Basic plasmid (represented as the open bar) was cotransfected with each Sp1-family expression vector into SL2 cells, and the luciferase activity was measured as before. D, expression of the Sp1-family proteins in Drosophila SL2 cells. Western blot analysis for detecting the Sp1, Sp2 or Sp3 protein was conducted as described under "Materials and Methods." Extracts from cells transfected with the pPacU vector were stained with a mixture of antibodies against Sp1, Sp2, and Sp3, serving as a background control. The positions of Sp1 (ä), Sp2 (E), and Sp3 (Ä) relative to the molecular size markers are indicated. protein, presumably due to N-terminal 82-amino acid truncation of the Sp1 ORF in the expression plasmid (47). Similarly, when extracts from Sp3-expressing SL2 cells were used, a 90-kDa DNA-binding protein was detected (Fig. 9C), consistent with the calculated molecular mass from the Sp3 ORF (33,35) in the expression plasmid. In addition, the molecular mass of the Sp1 or Sp3 protein detected here is identical to that observed in the Western blot analysis (Fig. 8D). Intriguingly, although both Sp1 and Sp3 bind well to the R2R1 DNA in EMSA (Fig. 9, A and B), the intensity of the Sp3 band detected in this Southwestern blot is much weaker than that of Sp1 (Fig.  9C). In contrast, when extracts from SL2 cells transfected with the pPacSp2 or the control pPacU plasmid were used, only an endogenous 50-kDa protein with weak intensity was detected (Fig. 9C). From all of these analyses, we conclude that Sp1 and Sp3, but not Sp2, can bind to the 11-bp repeat-containing R2R1 DNA.
The POLD1 Promoter Is Regulated during the Cell Cycle-Yang et al. (11) previously showed that the human POLD1 mRNA level is increased 10-fold in serum-stimulated IMR90 cells. To test if the POLD1 promoter can respond to serum stimulation, the POLD1-luciferase plasmid pGL2-␦(Ϫ1758) containing the 1.8-kb POLD1 promoter was transfected into C3H10T 1/2 cells. After transfection, cells were growth-arrested and then restimulated into the cell cycle with 10% serum. The stimulated cells were harvested for reporter enzyme assays at various time points after serum addition. As shown in Fig. 10A, the 1.8-kb POLD1 promoter was induced during the cell cycle and the increase in the POLD1 promoter activity began at 8 -12 h after serum stimulation and reached to the highest level (about 4-fold) at 24 -28 h, corresponding to the late G 1 /S phase as indicated by flow cytometry analysis (Fig. 10B).
To define the cis-acting elements responsible for serum stimulation of the POLD1 promoter during the cell cycle, several promoter deletion mutant derivatives of pGL2-␦(Ϫ1758) were studied. Deletion of the 5Ј upstream sequence between nucleotide positions Ϫ1758 and Ϫ92 appeared to have little or no effect on serum stimulation of the POLD1 promoter activity. Further deletion to position Ϫ22 (pGL2-␦(Ϫ22)), which removes the two 11-bp repeats, reduced but still showed some response to serum stimulation (about 2-fold; Fig. 10B). These results indicate that the sequence between position Ϫ92 and Ϫ22 containing the two repeats is important for serum stimulation of the POLD1 promoter. To confirm the role of the two 11-bp repeats in serum response, linker-replacement mutations in both copies of the repeats were introduced into pGL2-␦(Ϫ1758) (Fig. 11A). The resulting pGL2-␦(Ϫ1758)M2M1 mutant was studied for the serum response as described before. Similar to the deletion mutant pGL2-␦(Ϫ22) without the two repeats, pGL2-␦(Ϫ1758)M2M1 also showed reduced response to serum stimulation (Fig. 11B).
Since the POLD1 promoter without the two repeats (pGL2-␦(Ϫ22)) still gave rise to some response to serum stimulation (Fig. 10B), we examined if there was any additional element involved in the regulation. Examination of the POLD1 sequence between position Ϫ22 and ϩ49 in pGL2-␦(Ϫ22) revealed two sequences that resemble the binding site for the transcription factor E2F, known to be involved in cell-cycle regulation of many S-phase genes (21,59,60). One such sequence is located adjacent to the major transcription initiation site at position Ϫ13 to Ϫ6 with the sequence of 5Ј-TTG CGCGC-3Ј (one nucleotide (underlined) differs from the consensus sequence of the E2F-binding site 5Ј-TTTSSCGC-3Ј, where S ϭ G/C). The other is located in the 5Ј UTR at position ϩ37 to ϩ30 with the sequence of 5Ј-TTTCCCGC-3Ј, which is arranged in the antisense orientation and completely matches with the consensus sequence of the E2F-binding site (Fig. 11B). To examine if these two sequences are important for serum stimulation of the POLD1 promoter, mutation in each sequence was introduced into pGL2-␦(Ϫ1758) (Fig. 11A). Mutation of the putative E2F-binding sequence in the leader region (pGL2-␦(Ϫ1758)Lm) did not affect the response of the POLD1 promoter to serum stimulation, while mutation of the E2F-like sequence at the initiator region (pGL2-␦(Ϫ1758)Im) reduced the response to about 2-fold, suggesting that the E2F-like sequence at the initiator region is also important for serum stimulation.
To examine if both the 11-bp repeats and the E2F-like sequence at the initiator region can confer full serum response of the POLD1 promoter, the double mutant pGL2-␦(Ϫ1758)M2M1/Im containing mutations in both of these sequences was constructed (Fig. 11A) and analyzed. As shown in Fig. 11B, this mutant plasmid did not show any response to serum stimulation. Taken together, these results indicate that the human POLD1 promoter is induced during serum stimulation, and both the 11-bp repeats and an E2F-like sequence adjacent the major transcription initiation site are important for the regulation. DISCUSSION In this paper, we have analyzed the promoter of the gene for the catalytic subunit of human DNA polymerase ␦ (POLD1). We have demonstrated that the core promoter of the POLD1 gene is located within 328-bp DNA upstream from the major transcription initiation site. Multiple elements including two 11-bp direct repeats located between nucleotide position Ϫ92 and Ϫ22, play an important role in POLD1 promoter activity.
Several lines of evidence indicate that the two 11-bp repeats FIG. 10. The POLD1 promoter is induced at the late G 1 /S boundary in serum-stimulated cells. A, serum induction of POLD1 promoter-luciferase constructs. Various reporter constructs with different lengths of the POLD1 promoter were tested for their response to serum induction. The luciferase activity expressed from each promoter construct was normalized to the ␤-galactosidase activity and is shown as the relative activity to that expressed at time 0 prior to serum addition. The experiment was repeated multiple times and the representative data are shown. The constructs used are pGL2-␦(Ϫ1758) (ࡗ), pGL2-␦(Ϫ366) (E), pGL2-␦(Ϫ92) (Ⅺ), and pGL2-␦(Ϫ22) (Ç). B, flow cytometry analysis of synchronized C3H10T 1/2 cells at various time points during serum stimulation. At least 10,000 gated cells were analyzed as described under "Materials and Methods." function as an activator sequence for the POLD1 promoter. First, deletion of one copy of the two repeats reduced the POLD1 promoter activity by about 2-4-fold, while deletion of both copies nearly abolished the promoter activity. Second, linker-replacement mutations in either one or both copies of the two repeats gave rise to a similar effect on the promoter activity as the deletion mutants. Third, a 70-bp POLD1 DNA containing the two 11-bp repeats can enhance the expression of both the POLD1 and the heterologous HSVtk basal promoters in an orientation-independent manner.
The sequence of the 11-bp repeat (5Ј-GGGGCGTGGCC-3Ј) does not show complete identity with any known transcription factor-binding sites (61)(62)(63). However, the left half of the repeat sequence (5Ј-GGGCGT-3Ј) resembles the consensus sequence of the Sp1-binding site (5Ј-GGGCGG-3Ј; 31, 32). Indeed, several experiments indicated that Sp1 can bind to the 11-bp repeat sequence of the POLD1 promoter DNA. Southwestern blot analysis showed that a 105-kDa protein, comigrating with the Sp1 protein, could bind to the repeat sequence. An anti-Sp1 antibody could supershift most of the protein-DNA complexes containing the 11-bp repeats in EMSA. In addition, the Sp1 protein expressed in Drosophila SL2 cells could bind to the repeat sequence. Similar to Sp1, another Sp1-family protein, Sp3, could also bind to the 11-bp repeats. A cDNA clone encoding Sp3 was obtained by screening the HeLa cDNA expression library for the sequence-specific DNA-binding protein using the 11-bp repeats as the probe. Also, the Sp3 protein expressed from this cDNA could bind to the repeat sequence. The role of these Sp1-family proteins in regulating the POLD1 promoter activity was demonstrated by the fact that both Sp3 and Sp1 can activate the POLD1 promoter through the two 11-bp repeats. In contrast, Sp2 can not bind to the repeat sequence and as a consequence, has no effect on the POLD1 promoter activity.
In addition to the Sp1-family proteins, a doublet of 85-kDa protein was found to bind to the 11-bp repeats in both the Southwestern blot and UV cross-linking analyses. Consistently, although a linker-replacement mutation in the left half of the 5Ј copy of the repeat sequence that resembles the Sp1binding site reduced the POLD1 promoter activity, mutation in the right half produced a similar effect. In addition, deletion of the entire 5Ј copy of the 11-bp repeat resulted in a greater reduction of the promoter activity, suggesting the requirement of the entire 11-bp repeat sequence and the possible involvement of the 85-kDa proteins for POLD1 promoter regulation. It should be pointed out that these 85-kDa proteins are not degradation products of Sp1, since they were not detected by the Sp1 site-containing oligonucleotide DNA. Additionally, the molecular weights of these proteins are not the same as those of the Sp1-family members (33)(34)(35) or the small GT/GC box-binding proteins BTEB1, BTEB2, and EKLF (64 -66).
Intriguingly, the 85-kDa binding proteins were the predominant species that bind to the 11-bp repeats in Southwestern blot analysis, while most, but not all, of the protein-DNA complexes detected by EMSA could be supershifted by the anti-Sp1 antibody. Similarly, Sp3 appears to bind well to the 11-bp repeat sequence in EMSA but only binds weakly in Southwestern blot analysis as compared to Sp1, which shows strong binding in both assays. It is possible that Sp1 may have stronger binding affinity to the 11-bp repeat sequence in EMSA, while the conditions in Southwestern blot analysis allow the detection of other subtle DNA-binding proteins such as the FIG. 11. The 11-bp repeats and an E2F-like sequence are required for full serum response. (A) Diagram of the pGL2-␦(Ϫ1758) construct and its mutant derivatives. The locations of the two 11-bp repeats (R2R1) are indicated. The major transcription initiation site is designated as ϩ1. The mutations (M2M1) in the two repeats were created as described in Materials and Methods. The mutation in the E2F-like sequence at the initiator region (Im) or the putative E2F-binding sequence in the leader region (Lm) were introduced by the PCR-based method (44). At the bottom is the POLD1 sequence from nucleotide position Ϫ20 to ϩ40 containing the two potential E2Fbinding sites (underlined). The nucleotides changed in the Im and Lm mutations are also indicated. (B) Effect of various site-specific mutations to serum response of the POLD1 promoter. The promoter activities were measured as described in Fig. 10. The constructs used are pGL2-␦(Ϫ1758),ࡗ; pGL2-␦(Ϫ1758)Lm,छ; pGL2-␦(Ϫ1758)Im,Ç; pGL2-␦(Ϫ1758)M2 M1,Ⅺ; and pGL2-␦(Ϫ1758)M2M1/Im,E. 85-kDa proteins. Importantly, mutation analysis showed that binding of the 85-kDa proteins to the POLD1 promoter DNA required the presence of both copies of the 11-bp repeats. The molecular masses and the specific binding property of these 85-kDa binding proteins suggest that they are potentially novel transcription factors for the expression of the POLD1 gene. Further characterization of these DNA-binding proteins and their encoding genes should allow us to better understand the transcriptional regulation of the gene encoding this major DNA replication enzyme.
The result from DNase I footprinting analysis showed that, in addition to the 11-bp repeat region, another protein-binding site was detected in the upstream region of the POLD1 promoter. Sequence analysis identified two Sp1-binding sites in the POLD1 core promoter region (19), and one of these sites coincides with the protein-binding site detected in the footprinting analysis. It appears that these Sp1 sites play a positive role in regulating the human POLD1 gene, since deletion of these Sp1 site-containing regions gave rise to some reduction of the promoter activity. Besides the positive elements identified in the human POLD1 promoter, a negative element located within the region from nucleotide position Ϫ68 to Ϫ58 appears to exist since deletion of this region resulted in an increase of promoter activity. Interestingly, a sequence within this region (5Ј-CACTTG-3Ј at position Ϫ63 to Ϫ58) shares similarity to the binding sites of the basic region helix-loop-helix and basic region helix-loop-helix-zipper proteins, defined as the E-box sequence motif (5Ј-CAXXTG-3Ј in which the central 2 bp are specified by each protein; Refs. [67][68][69][70][71][72][73]. At present we have not identified the protein that binds to this element using crude HeLa nuclear extracts. Nonetheless, the presence of both positive and negative elements and their interactions may play an important role in regulating the expression of the POLD1 gene in the proliferating state (11).
Our previous analysis showed that the human POLD1 mRNA level is increased 10-fold in serum-stimulated IMR90 cells (11). In this study, the POLD1 promoter is induced about 4-fold during serum stimulation, suggesting that regulation of the POLD1 gene during the cell cycle is not limited to the transcriptional level. Posttranscriptional regulation has been found to be important for growth stimulation of several genes involved in nucleotide metabolism and DNA synthesis (20,22). It should be noted that the POLD1 gene contains 26 introns, and seven of them are smaller than 100 bp. It is possible that correct processing of the pre-mRNA and, perhaps, the sequence within the intron may contribute to the regulation of the POLD1 gene. In addition, Zeng et al. (41) showed that the catalytic subunit of human pol ␦ is a phosphoprotein that is most actively phosphorylated during the S phase. Thus, regulation of POLD1 expression appears to be mediated at multiple levels.
Interestingly, both the 11-bp repeats and an E2F-like sequence located immediately upstream to the major transcription initiation site are responsible for serum stimulation of POLD1 promoter activity. The E2F transcription factor is involved in the transcription of several cellular genes necessary for proliferation and is one of the targets of the retinoblastoma family of growth suppressors (reviewed in Ref. 74). The retinoblastoma protein, which negatively regulates cell cycle progression from G 1 into S phase, binds to E2F, and the functional consequence of this interaction is to prevent E2F from activating transcription. The presence of the E2F site at or adjacent to the transcription initiation site has been found in a few growthregulated genes (for reviews, see Refs. 20, 21, 59, and 60). Although the sequence of this E2F-like element in the POLD1 promoter has one nucleotide difference from the consensus sequence of the E2F-binding site, it is possible that E2F may regulate the POLD1 gene through this element for its timely expression during the cell cycle. Experiments are in progress to investigate if indeed, E2F can bind and activate this POLD1 element. Recently, Lin et al. (75) and Karlseder et al. (76) found that E2F1 and Sp1 both functionally and physically interact with each other, and suggested that through this interaction, Sp1 and E2F1 may regulate transcription of genes containing binding sites for either or both factors. Given the fact that Sp1 can bind and activate the POLD1 promoter through the 11-bp repeat sequence, and both the 11-bp repeats and the E2F-like sequence near the initiator region are involved in serum stimulation of the POLD1 promoter activity, we are presently conducting experiments to examine if Sp1 and E2F can cooperatively regulate POLD1 expression. Nevertheless, the present study provides direct evidence for transcriptional regulation of the human POLD1 gene.