Transcriptional activation of the human hematopoietic prostaglandin D synthase gene in megakaryoblastic cells. Roles of the oct-1 element in the 5'-flanking region and the AP-2 element in the untranslated exon 1.

The human hematopoietic prostaglandin D synthase (H-PGDS) gene is highly expressed in human megakaryoblastic cells, in which phorbol ester induces its expression. We characterized the promoter activity of the 5'-flanking region and the untranslated exon 1 (-1044 to +290) of the human H-PGDS gene in human megakaryoblastic Dami cells. Transient expression analysis using the luciferase reporter gene revealed that the 5'-flanking region and the untranslated exon 1 were sufficient for efficient expression of the H-PGDS gene in Dami cells, but not in monocytic U937 cells. Deletion and site-directed mutagenesis of the Oct-1 element in the 5'-flanking region decreased the promoter activity by approximately 30% compared with that of the entire region from -1044 to +290. An electrophoretic mobility shift assay demonstrated that Oct-1 specifically bound to the promoter region. Interestingly, even only untranslated exon 1 (+1 to +290) showed approximately 60% of the promoter activity of the entire region from -1044 to +290. Site-directed mutagenesis of the AP-2 element within the untranslated exon 1 abolished the basal promoter activity as well as its phorbol ester-mediated up-regulation. In AP-2-deficient HepG2 cells, the H-PGDS promoter activity was enhanced by coexpression with AP-2alpha. These findings indicate that the Oct-1 element in the 5'-flanking region acts as a positive cis-acting element and that the AP-2 element in the untranslated exon 1 is crucial for both basal and phorbol ester-mediated up-regulation of human H-PGDS gene expression in megakaryoblastic Dami cells.

Hematopoietic prostaglandin D synthase (H-PGDS) is responsible for the biosynthesis of PGD 2 in antigen-presenting cells (7), mast cells (8,9), megakaryocytes (10,11), and type 2 helper T lymphocytes (12). H-PGDS was originally purified from rat spleen as a cytosolic GSH-requiring enzyme (13,14). The cDNA for H-PGDS was cloned from rat (15), chicken (16), mouse, and human (17). H-PGDS was crystallized, and its tertiary structure was determined, with a resolution of 2.3 Å by x-ray diffraction analysis, to possess a unique catalytic site within a wide cleft near the GSH-binding site (15). Recently, the critical amino acid residues involved in the catalysis of H-PGDS were identified by a site-directed mutation study (18). On the basis of molecular evolutional study, H-PGDS was shown to be a member of the glutathione S-transferase (GST) family with significant homology to the invertebrate Sigma class GST. Since the Sigma class GST had been observed only in invertebrates, H-PGDS is the first identified vertebrate homolog of the Sigma class GST.
Recently, we isolated the human and mouse genes for H-PGDS and determined the chromosomal positions to be 4q21-22 and 3B-C, respectively (17). The positions of introns in the coding region were completely conserved between H-PGDSs and invertebrate Sigma class GSTs, indicating that H-PGDSs and invertebrate Sigma class GSTs evolved from a common ancestral gene (17). We also found that the expression profiles of the H-PGDS mRNAs were completely different among animal species (17). Furthermore, H-PGDS mRNA was actively expressed in human megakaryoblastic cell lines such as CMK11 and Dami (10), which had been established from the peripheral blood of a patient with acute megakaryoblastic leukemia (19,20). In these cell lines, the expression of H-PGDS mRNA was induced by treatment of the cells with phorbol 12-myristate 13-acetate (PMA) (10,11), which causes those megakaryoblastic cells to differentiate into mature megakaryocyte-like cells (19,20). However, little is known about the activation mechanisms, including PMA-mediated induction, of human H-PGDS gene expression.
To understand the regulatory mechanism for transcription of the human H-PGDS gene, we determined the nucleotide sequence of the 5Ј-flanking region and the untranslated exon 1 of the gene and investigated the cis-acting elements and the trans-acting factors responsible for the transcriptional activation, including the PMA-mediated up-regulation, of the H-PGDS gene in megakaryoblastic Dami cells. This is the first report about the characterization of the transcriptional regulation of a gene in the Sigma class GST family.
Primer Extension Analysis-Total RNA was prepared from CMK11-5 cells as described previously (17) and purified to poly(A) ϩ RNA with Oligotex-dT30 (Takara Shuzo, Kyoto, Japan). A 27-mer antisense primer was synthesized (5Ј-AGGCCAGTCAGCTTGTTCTATTCTGTG-3Ј, corresponding to the cDNA sequence spanning positions ϩ130 to ϩ104) and labeled with an infrared dye, IRD41 (Nisshinbo, Tokyo, Japan), at its 5Ј-end. One g of poly(A) ϩ RNA prepared from CMK11-5 cells or human placenta (CLONTECH, Palo Alto, CA) was dissolved in 20 l of a solution containing 50% (v/v) formamide, 400 mM NaCl, 1 mM EDTA, and 40 mM PIPES (pH 6.4) and mixed with the IRD41-labeled primer (1 pmol). The reaction mixtures were denatured at 80°C for 15 min and incubated at 25°C for 1 h. Hybridization products were precipitated with ethanol and dissolved in 50 mM Tris-HCl (pH 8.3) containing 75 mM KCl, 3 mM MgCl 2 , and 10 mM dithiothreitol, to which were added 1 mM dNTPs and 200 units of Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.). After incubation at 42°C for 1 h, the samples were extracted with phenol/chloroform and precipitated with ethanol. The precipitate was dissolved in gel loading buffer containing 95% (v/v) formamide, 10 mM EDTA (pH 7.6), 0.1% (w/v) xylene cyanol, and 0.1% (w/v) bromphenol blue and analyzed on a 6% Long Ranger and 7 M urea sequencing gel (FMC Corp. BioProducts, Rockland, ME) in 1.2ϫ Tris borate/EDTA (106.8 mM Tris borate and 2.4 mM EDTA). Signals were detected with an automated DNA sequencer system (Model 4000, LI-COR, Lincoln, NE).
Promoter-Luciferase Constructs, Site-directed Mutagenesis, Transient Transfection, and Luciferase Assay-An ϳ1.3-kilobase pair fragment (Ϫ1044 to ϩ290) consisting of the 5Ј-flanking region and the untranslated exon 1 of the human H-PGDS gene was amplified by PCR with the gene-specific primer sets with HindIII (sense) and BglII (antisense) sites at their respective 5Ј-ends. PCR fragments doubly digested with HindIII and BglII were cloned into the upstream region of the luciferase reporter gene of the pGL3-Enhancer vector (Promega, Madison, WI). The luciferase reporter construct carrying the promoter region from Ϫ1044 to ϩ290 is designated as Ϫ1044/ϩ290. A deletion series of the 5Ј-flanking region and the untranslated exon 1 was constructed in the same manner. Site-directed mutations were introduced into the putative cis-acting elements by PCR using mutated primers. All constructs were subjected to nucleotide sequencing to verify the correct sequence and orientation.
Cells grown to ϳ70% confluence were cotransfected with each construct (1 g) and pRL-SV40 (0.2 g; Promega) carrying the Renilla luciferase gene under the control of the SV40 early promoter as the internal control. Transfection was achieved with Effectene reagent (QIAGEN, Hilden, Germany) according to the manufacturer's instruction. Cells were harvested 48 h after transfection, and cell extracts were prepared. The luciferase activities were measured by a microplate luminometer (Lucy 1, Anthos, Salzburg, Austria) and a dual luciferase reporter assay kit (Promega). Each value was normalized to that of the Renilla luciferase control. The reporter activity of each construct was calculated relative to that of the pGL3-Enhancer vector, which was defined as 1. All transfections were performed in duplicate and repeated at least three times. The relative promoter activities are indicated as means Ϯ S.D.
Coexpression Study-The coding region of AP-2␣ cDNA (23) was amplified by PCR using AP-2␣F as the sense primer and 5Ј-TCACTT-TCTGTGCTTCTCCTCTTTGTC-3Ј as the antisense primer. The resultant PCR products were cloned into the pcDNA1.1/Amp expression vector (Invitrogen, San Diego, CA) through cloning into the pGEM-T Easy vector (Promega). HepG2 cells were cotransfected with the AP-2␣ expression vector and the Ϫ1044/ϩ290 construct; and after 72 h, the luciferase activities were measured as described above.

Analysis of the Regulatory Region of the Human H-PGDS
Gene-Previously, we isolated the gene for human H-PGDS (17). In this study, we determined the transcription initiation sites and the nucleotide sequence of the 5Ј-flanking region and the untranslated exon 1 of the human H-PGDS gene (Fig. 1, A  and B). The transcription initiation sites of the human H-PGDS gene were identified by primer extension analysis with poly(A) ϩ RNAs from human CMK11-5 cells and placenta. In both mRNAs, two predominant transcription initiation sites were detected, one at 299 base pairs and the other at 270 base pairs upstream from the A of the ATG initiation codon for translation. The transcription initiation site at 299 base pairs upstream from the ATG codon was defined as position ϩ1. Neither the typical TATA nor CAAT box was identified around the transcription initiation sites. The 5Ј-flanking region and the untranslated exon 1 of the human H-PGDS gene contained several potential cis-acting elements recognized by the transcription factors that may play roles in the regulation of the human H-PGDS gene, including one Oct-1 element (position Ϫ357), five GATA elements (positions Ϫ912, Ϫ534, Ϫ307, ϩ82, and ϩ204), two PEA3 elements (positions ϩ153 and ϩ172), one CRE-like element (position ϩ217), and one AP-2 element (position ϩ227) (Fig. 1B).
Previous studies have shown that human megakaryoblastic cell lines such as CMK11-5 and Dami abundantly express H-PGDS mRNA (10,11). In contrast, monocytic U937 cells did not express H-PGDS mRNA (data not shown). Since the transfection of Dami cells was more efficient than that of CMK11-5 cells (data not shown), we used Dami cells as the cell line expressing H-PGDS mRNA and U937 cells as one not expressing it for further studies of the transcriptional regulatory mechanisms of the human H-PGDS gene.
To define the regulatory elements for the cell type-specific expression of the human H-PGDS gene, we cloned a series of deletion constructs of the 5Ј-flanking region and the untranslated exon 1 into a luciferase reporter vector, pGL3-Enhancer vector, and transfected Dami or U937 cells with it. When the Ϫ1044/ϩ290 construct was used for transfection, efficient reporter activity was detected in Dami cells, but was negligible in U937 cells (Fig. 2), demonstrating that the promoter region from Ϫ1044 to ϩ290 contains the cis-acting element(s) for the Dami cell-specific expression of the human H-PGDS gene. No activity was detected when the longest region in the reverse orientation (ϩ290 to Ϫ1044) was used as the promoter (Fig. 2), indicating that this promoter activity is orientation-specific.
Function of the Oct-1 Element in the 5Ј-Flanking Region-Progressive deletion analysis of the 5Ј-flanking region and the untranslated exon 1 revealed that the deletion from Ϫ1044 to Ϫ527 led to an ϳ15% increase in promoter activity, suggesting the presence of a suppressor element(s) in this region. Further deletion from Ϫ1044 to Ϫ335 reduced the reporter activity by nearly 30% compared with that of the Ϫ1044/ϩ290 construct (Fig. 2), suggesting that a positive cis-acting element(s) lies in the region from Ϫ526 to Ϫ335. As we had found one Oct-1 element at position Ϫ357, we introduced a mutation into this element to verify whether Oct-1 was functional. Expression of the construct with the mutated Oct-1 element in Dami cells resulted in an ϳ30% decrease in the reporter activity compared with the activity of the Ϫ1044/ϩ290 construct, and this reduction was almost the same as that caused by the deletion from Ϫ1044 to Ϫ335 (Fig. 2). EMSA with the double-stranded H1 oligonucleotide containing the Oct-1 element and nuclear extracts prepared from Dami cells demonstrated the presence of one major DNA-protein complex (Fig. 3, lane 2). This DNA-protein complex was not formed in the absence of nuclear extracts (lane 1). The formation of the complex was inhibited by the addition of a 10or 50-fold excess amount of the unlabeled H1 oligonucleotide (lanes 3 and 4)  Roles of the Untranslated Exon 1 in Transcriptional Activation-In the analysis of the promoter activity as described above, deletion between ϩ86 and ϩ256 of the untranslated exon 1 completely abolished the reporter activity. To determine the promoter region and the putative cis-acting element(s) within the untranslated exon 1 of the human H-PGDS gene, we transfected Dami cells with a series of the deleted untranslated exon 1 fused with the luciferase reporter gene. As shown in Fig.  4, the untranslated exon 1 (the ϩ1/ϩ290 construct) as a pro-moter showed ϳ60% of the reporter activity compared with that of the Ϫ1044/ϩ290 construct. A lack of the region from ϩ222 to ϩ290 decreased this activity by ϳ90%. On the other hand, deletion of the region from ϩ257 to ϩ290 did not reduce this reporter activity. These results indicate that the region from ϩ222 to ϩ256 is crucial for basal expression. Since this region contains one each of the GATA, CRE-like, and AP-2 elements, a site-directed mutation was introduced into the GATA, CRE-like, and/or AP-2 elements to investigate the possible function(s) of these cis-acting elements (Fig. 4). The reporter activity upon introduction of the mutation into the AP-2 element was decreased by ϳ90% compared with that of the ϩ1/ϩ290 construct, whereas the mutation in the GATA or CRE-like element did not affect the reporter activity (Fig. 4). These results indicate that the AP-2 element is essential for the basal expression of the human H-PGDS gene in Dami cells.
Binding of the nuclear factor(s) specific for the region from ϩ222 to ϩ256 was demonstrated by EMSA in which the double-stranded H2 oligonucleotide containing CRE-like and AP-2 elements was incubated with nuclear extracts prepared from Dami cells. As shown in Fig. 5, one major DNA-protein complex was observed (lane 1). The specificity of this DNA-protein complex was shown by the competition experiment, in which the amount of the complex was significantly diminished by the addition of a 20-or 100-fold excess amount of the unlabeled H2 oligonucleotide (lanes 2 and 3) or of a 50-fold excess amount of the authentic AP-2 oligonucleotide (lane 5). However, the complex formation was not inhibited by a 100-fold excess amount of the unrelated pBluescript vector (lane 4) or by a 50-fold excess amount of the authentic AP-2 mutant oligonucleotide (lane 6). Therefore, AP-2 could bind to the untranslated exon 1 of the human H-PGDS gene.
Up-regulation by PMA through the AP-2 Element in the Untranslated Exon 1-To identify the region involved in the enhanced transcription by PMA, we measured the promoter activity of deleted or mutated promoter-luciferase constructs in the presence or absence of PMA in Dami cells (Fig. 4). The reporter activity of the ϩ1/ϩ290 construct was enhanced ϳ5fold by PMA treatment. Almost the same -fold induction was detected when the Ϫ1044/ϩ290 construct was used (data not shown). Deletion analysis demonstrated that PMA responsiveness was completely abolished when the region from ϩ222 to ϩ256 containing CRE-like and AP-2 elements was deleted ( Fig.   FIG. 3. EMSA shows that Oct-1 binds to the human H-PGDS promoter. The 32 P-labeled double-stranded H1 oligonucleotide containing the Oct-1 element was incubated for 20 min at room temperature with the nuclear extracts prepared from Dami cells. The incubation mixture was analyzed by electrophoresis on 5% polyacrylamide gels as described under "Experimental Procedures." Arrows a and b denote the shifted and supershifted complexes, respectively. Oct-1 WT, wild-type Oct-1; Oct-1 mu, mutant Oct-1; ori, origin. 4). Site-directed mutation of CRE-like and/or AP-2 element(s) demonstrated that mutation of the AP-2 element abolished the PMA responsiveness; in contrast, mutation of the CRE-like element did not affect PMA-mediated up-regulation in Dami cells (Fig. 4). These results clearly indicate that the AP-2 element within the untranslated exon 1 also contributes to the PMA-mediated up-regulation of the human H-PGDS gene in Dami cells.
Functional Expression of AP-2 in HepG2 Cells-The AP-2 subtype expressed in Dami cells was determined by reverse transcription-PCR using each gene-specific primer set for AP-2␣, AP-2␤, or AP-2␥ (Fig. 6A). In Dami cells, AP-2␣ mRNA expression was detected (second lane), but the mRNA expression of the other subtypes (AP-2␤ and AP-2␥) was below the detection limit under our experimental conditions (third and fourth lanes). For further examination of AP-2 as a transcriptional activator, the expression vector carrying the human AP-2␣ cDNA was constructed and used to transfect HepG2 cells, which have been shown to be deficient in AP-2 activity (23). The reporter activity of the Ϫ1044/ϩ290 construct in HepG2 cells increased upon coexpression with the expression vector for human AP-2␣, and the increase was AP-2␣ dose-dependent (Fig. 6B). These results, taken together, indicate that AP-2␣ activates human H-PGDS gene expression in Dami cells.

DISCUSSION
Previous studies have shown that the H-PGDS gene is highly expressed in human megakaryoblastic cells and that its tran-scription can be induced by PMA (10,11). We have recently reported that the expression profiles of H-PGDS genes are divergent among animal species (17). To date, however, there has been no study on the regulation of H-PGDS gene expression. In this study, we characterized, for the first time, the 5Ј-flanking region and the untranslated exon 1 of the human H-PGDS gene to elucidate its regulatory mechanisms. Our proposed model of the activation mechanism of human H-PGDS gene expression in megakaryoblastic Dami cells is shown in Fig. 7.
Two transcription initiation sites were identified by primer extension analysis to be 299 and 270 base pairs upstream from the ATG translation initiation codon (Fig. 1A). Neither the typical TATA nor CAAT box was found near either transcription initiation site. This finding is in agreement with the fact that multiple initiation sites are often observed in genes utilizing a TATA-less promoter (24,25). Like the human H-PGDS gene, the human cyclooxygenase-1 and thromboxane synthase genes involved in prostanoid biosynthesis also have TATA-less promoters with multiple transcription initiation sites (26,27).
Transient expression assays with a series of the human H-PGDS promoter-luciferase constructs demonstrated that an ϳ1.3-kilobase pair region covering the 5Ј-flanking region and the untranslated exon 1 of the human H-PGDS gene had effi- The double-stranded H2 oligonucleotide containing the AP-2 element labeled with [␥-32 P]ATP was incubated with the nuclear extracts prepared from Dami cells, and the mixtures were then subjected to EMSA. Unlabeled H2 oligonucleotide was added as the competitor. The shifted band is indicated by the arrow. AP-2 WT, wild-type AP-2; AP-2 mu, mutant AP-2; ori, origin.

FIG. 6. Expression of the AP-2 subtypes in Dami cells and functional expression in HepG2 cells.
A, the reverse transcription reaction was carried out using 1 g of total RNA prepared from Dami cells with oligo(dT) adapter primer, followed by nested PCR. The resultant PCR products were analyzed on 1.5% agarose gel. Lane M contains molecular size markers. B, the Ϫ1044/ϩ290 construct and 0, 0.2, or 2 g of AP-2␣ expression vector were used for cotransfection of HepG2 cells. After 72 h, luciferase activities were measured. Each value is expressed relative to the activity of an empty pGL3-Enhancer vector. cient promoter activity in megakaryoblastic Dami cells, but not in monocytic U937 cells (Fig. 2). In the human H-PGDS promoter, at least two regions (Ϫ526 to Ϫ335 and ϩ222 to ϩ256) are critical for cell type-specific expression in Dami cells. For the region from Ϫ526 to Ϫ335, the EMSA and site-directed mutation studies demonstrated that Oct-1 bound to ATG-CAAAT (Ϫ357 to Ϫ350), which is a perfect match with the Oct-1 consensus sequence (28), and was probably responsible for the trans-activation of this element. However, in the luciferase reporter assay, deletion of the region containing the Oct-1 element caused only a 30% decrease compared with the activity found with the Ϫ1044/ϩ290 construct (Fig. 2). Thus, Oct-1 can be considered to function as a minor transcriptional activator for the human H-PGDS gene. Efficient transcriptional activation through Oct-1 is dependent on the interaction with coactivators such as VP-16 (29) and Pit-1 (30). Therefore, additional coactivators might be needed for more efficient transcriptional activation of the human H-PGDS gene via the Oct-1 element.
The most important cis-acting element was located in the region from ϩ222 to ϩ256 containing CRE-like and AP-2 elements within the untranslated exon 1. Deletion and site-directed mutation of the AP-2 element, ACCCAAGGC (ϩ227 to ϩ235), whose consensus sequence is GCCNNNGGC (31), diminished the reporter activity to ϳ10% of that obtained with the ϩ1/ϩ290 construct (Fig. 4). EMSA clearly indicated the specific binding of AP-2 to this element (Fig. 5). Furthermore, the reporter activity of the ϩ1/ϩ290 construct in Dami cells was enhanced ϳ5-fold by PMA treatment (Fig. 4). A lack of the region from ϩ222 to ϩ256 containing CRE-like and AP-2 elements abolished the PMA responsiveness. Moreover, a mutation in AP-2 (but not in CRE-like) elements resulted in the loss of the responsiveness to PMA (Fig. 4), indicating that induction of human H-PGDS gene expression by PMA can be modulated only through the AP-2 element within the untranslated exon 1. Evidence supporting this result is that cotransfection of HepG2 cells with the Ϫ1044/ϩ290 construct and the AP-2 expression vector increased the reporter activity (Fig. 6B). These results indicate that the untranslated exon 1 of the human H-PGDS gene is essential for the transcriptional activation and that especially the AP-2 element is crucial for both basal and PMA-mediated upregulation of gene expression in Dami cells. Such an activation mechanism of the human H-PGDS gene resembles that of the human tissue-type plasminogen activator gene (32), whose basal and PMA-mediated expressions are regulated through the AP-2 element within the untranslated exon 1. It is interesting that even only the untranslated exon 1 of the human H-PGDS gene showed ϳ60% of the promoter activity of the Ϫ1044/ϩ290 construct, unlike the untranslated exon 1 of the human tissue-type plasminogen activator gene, which showed only 20 -30% of the promoter activity found upon constitutive expression (32).
The AP-2 family consists of three members, AP-2␣, AP-2␤, and AP-2␥ (33), which are involved in the regulation of a number of cellular events, including growth and differentiation (34), and in the transcriptional activation from the signal derived from PMA (35) and retinoic acid and cAMP (36,37). Dami cells expressed AP-2␣, but not the AP-2␤ and AP-2␥ subtypes (Fig. 6A). Therefore, the expression of the human H-PGDS gene was considered to be activated by AP-2␣ in Dami cells. However, it is known that AP-2 is expressed in U937 cells (38), but actually the U937 cells used in this study expressed AP-2␥ (data not shown). AP-2␥ might suppress human H-PGDS gene expression in U937 cells, or transcriptional activation of the human H-PGDS gene by AP-2 might require additional factors.
In summary, this study has, for the first time, directly examined, by luciferase reporter assay, site-directed mutation study, and EMSA, the roles of specific transcription factors in the regulation of the human H-PGDS gene, which belongs to the Sigma class GST family. We have suggested that the Oct-1 element in the 5Ј-flanking region acts as the cis-acting element and that the AP-2 element in the untranslated exon 1 plays critical roles in the basal and PMA-mediated induction of human H-PGDS gene expression in megakaryoblastic Dami cells. A functional study using HepG2 cells indicated that AP-2␣ can activate human H-PGDS gene expression.