Novel role for Sp1 in phorbol ester enhancement of human platelet thromboxane receptor gene expression.

Expression of platelet thromboxane receptors is transcriptionally increased during megakaryocytic differentiation stimulated by phorbol 12-myristate 13-acetate (PMA). We previously cloned and characterized the promoter region of the human thromboxane receptor gene and localized PMA-responsive elements to a region between 1.84 and 1.95 kilobase pairs (kb) 5′ of the transcription initiation site (D'Angelo, D. D., Davis, M. G., Houser, W. A., Eubank, J. J., Ritchie, M. E., and Dorn, G. W., II (1995) Circ. Res. 77, 466-474). Herein we report the localization of the PMA response element to a 14-nucleotide C-rich sequence, flanked by an octanucleotide inverted repeat, located −1.938 to −1.925 kb 5′ of the transcription start site of this gene. We further identify the PMA-responsive enhancer factor that binds to this C-rich sequence as Sp1. Heterologous thromboxane receptor gene promoter/thymidilate kinase reporter constructs transfected into K562 cells exhibited PMA responsiveness when the C-rich element was included with additional 3′ sequence from −1.924 to −1.84 kb. However, mutations of the C-rich element that disrupted a GC box located on the inverse strand eliminated PMA responsiveness and, in gel mobility shift assays, eliminated binding of Sp1. PMA treatment of K562 cells significantly increased, by 5-fold, Sp1 binding to the C-rich element and increased both phosphorylated and nonphosphorylated Sp1 protein levels by 2-fold. Furthermore, PMA treatment transiently increased Sp1 mRNA levels prior to increasing thromboxane receptor mRNA, suggesting that up-regulation of Sp1 contributes to up-regulation of thromboxane receptors. Finally, we have detected an unidentified K562 nuclear protein that binds specifically to the sense strand of the C-rich sequence overlapping the Sp1 binding site and that, by stabilizing a double stem-loop conformation of this DNA segment, may also play a role in Sp1 regulation of this gene. These studies are the first to describe regulatory and regulated roles for Sp1 in PMA-responsive gene expression and suggest that modulation of Sp1 levels controls thromboxane receptor expression during megakaryocytic differentiation.

Thromboxane A 2 is one of the most potent platelet-aggregating and vasoconstricting substances known and is crucial for maintenance of normal hemostasis (1). While considerable attention has been directed toward measuring increased thromboxane production in various disease states including acute coronary ischemic syndromes (2,3), relatively little is known about how target tissue responses to thromboxane may be modulated. Two intriguing studies have reported increased numbers of platelet thromboxane receptors, with enhanced platelet aggregation, in acute myocardial infarction and unstable angina pectoris (4,5). The observation that this thromboxane receptor abnormality was reversible after acute myocardial infarction (4) and the discovery of an identical abnormality in patients with unstable angina (the clinical precursor of completed myocardial infarct) (5), strongly suggest that unidentified factors can up-regulate platelet thromboxane receptors in vivo and that the resulting platelet hypersensitivity to thromboxane precedes, and can contribute to, thrombotic coronary occlusion.
Since platelets lack a significant capacity for protein synthesis, platelet thromboxane receptor number must be determined prior to platelet origination from platelet precursor megakaryocytes (6). Therefore, most investigations into the determinants of platelet thromboxane receptor expression have been carried out in cultured megakaryoblast-like cells. We previously reported that the platelet-like cultured human leukemic cell lines K562 and CHRF-288 -11 exhibited increased levels of thromboxane receptor protein and steady-state mRNA expression after megakaryocytic differentiation was stimulated with either thrombin or phorbol 12-myristate 13-acetate (PMA) 1 (7,8). These findings suggested that physiologic activation of protein kinase C could enhance thromboxane receptor gene expression in platelet progenitor cells and ultimately up-regulate platelet thromboxane receptors.
In an initial attempt to delineate the mechanism for phorbol ester-stimulated enhancement of thromboxane receptor gene expression, we cloned and characterized the promotor region of the human thromboxane receptor gene and localized PMA responsiveness to a region located between 1.84 and 1.95 kb upstream of the start of transcription (Ref. 8, GenBank TM accession number U30503). This region of the thromboxane receptor gene promotor specifically bound an unidentified K562 nuclear protein, the levels of which were increased in nuclear extracts from PMA-treated cells.
In the present study we have identified the PMA response element in the human thromboxane receptor gene promotor, determined the identity of the associated DNA binding factor, and have defined a mechanism for PMA enhancement of transcription of this gene. We describe a novel action for transcription factor Sp1 acting as a phorbol ester-responsive transactivating factor for the thromboxane receptor gene in K562 cells. These studies represent the first detailed examinations of transcriptional regulatory mechanisms for any eicosanoid receptor and the first report of Sp1 involvement in protein kinase C regulation of gene expression.

EXPERIMENTAL PROCEDURES
Materials-The K562 (chronic myelogenous leukemia) cell line was obtained from the ATCC (Rockville, MD). The Fast CAT chloramphenicol acetyltransferase kits were from Molecular Probes (Eugene, OR). All radiochemicals were purchased from Dupont NEN. Oligonucleotides were synthesized and purified at the University of Cincinnati Core DNA Facility. Double-stranded oligonucleotides corresponding to transcription factor binding sites (AP-1, AP-2, AP-3, Sp1, GRE, NF-1, NKB, OCT-1, and CRCB) were purchased from Stratagene, La Jolla, CA. Sp1 affinity-purified rabbit polyclonal antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Purified Sp1 was from Promega (Madison, WI). Unless otherwise indicated, other reagents were of the highest quality available from Sigma or Life Technologies, Inc.
Promoter Analysis-Promoter activity in genomic fragments was characterized using chimeric thromboxane receptor gene-chloramphenicol acetyltransferase (CAT) constructs in the heterologous thymidilate kinase promotor CAT expression vector pBLCAT5 (9). The chimeras were transfected in duplicate into K562 cells as described previously (8). Briefly, 10 g of DNA (total) plus 25 l of Lipofectamine (Life Technologies, Inc.) were incubated at room temperature for 45 min. K562 cells grown to a density of 10 5 cells/ml in RPMI containing 10% fetal calf serum were washed and resuspended in serum-free RPMI at a density of 3 ϫ 10 6 cells/ml. The DNA/Lipofectamine mixture was mixed with 800 l of cells, aliquoted in one well of a 6-well dish, and placed in a 95% air, 5% CO 2 tissue culture incubator at 37°C. After 5 hours, 4 ml of RPMI, 10% fetal calf serum was added. Unless otherwise indicated, 100 nM PMA or vehicle (95% ethanol) was added to cells 24 h after transfection. For basal activity studies, transfection efficiency was monitored by cotransfecting pCMV␤ (1 g) (Clontech) in which the ␤-galactosidase gene is driven by the cytomegalovirus promoter. The resultant CAT activity levels were then normalized for ␤-galactosidase activity.
Forty-eight hours following transfection, cells were pelleted by centrifugation and lysed in 0.25 M Tris-HCl, pH 7.4, plus three cycles of freeze-thawing followed by microcentrifugation to remove particulate matter. The supernatant was heated to 65°C for 7 min to destroy endogenous acetyltransferase activity and was stored at Ϫ70°C for up to 3 days prior to assay. CAT activity was measured using the Fast CAT chloramphenicol acetyltransferase assay kit and the manufacturer's recommended protocol. Fluorescent acetylated products were resolved by thin layer chromatography on silica gels, acetylated and nonacetylated products were separately pooled, and fluorescence was measured on a Photon Technologies spectrofluorometer (490 nm excitation and 512 nm emission). In all studies, transfection of promoterless pBLCAT5 served to measure background activity, and pcDNA 3 CAT (driven by the cytomegalovirus promoter) served as an index of maximal CAT expression by K562 cells.
Extract for ␤-galactosidase activity was prepared as described for CAT assays except that the 65°C incubation was omitted. Cell extract was incubated with 0.26 mg of o-nitrophenyl-␤-D-galactoside, 0.1 M MgCl 2 , 5 M ␤-mercaptoethanol at 37°C for 30 min to 1 h. The reaction was stopped with 1 M Na 2 CO 3 , and ␤-galactosidase activity was quantified by its absorbance at 410 nm.
Preparation of K562 Nuclear Extract-Nuclear extract from vehicle or PMA-treated K562 cells was prepared as described previously (8). Briefly, K562 cells were pelleted and washed in hypotonic buffer followed by Dounce homogenization. Nuclei were pelleted at 3300 ϫ g, extracted with 300 mM KCl for 30 min, and dialyzed against 100 mM KCl followed by centrifugation at 25,000 ϫ g. Aliquots of nuclear extract were stored at Ϫ70°C prior to use.
DNA Footprinting-DNA footprinting was performed using the Core Footprinting system (Promega). Briefly, promotor fragment Ϫ1.96/ Ϫ1.84 in the EcoRV site of pBluescript (Stratagene, La Jolla, CA) was excised with BamHI/XhoI. The fragment was 32 P-labeled with T4 polynucleotide kinase followed by EcoRI digestion to leave the 3Ј end labeled. Probe (50,000 cpm) was incubated with 50 ng of purified Sp1 with or without 25 ng of an oligonucleotide corresponding to a genuine Sp1 binding site in 50 l containing 25 mM Tris-HCl, pH 8.0, 50 mM KCl, 6.75 mM MgCl 2 , 0.5 mM EDTA, 10% glycerol, 0.5 mM dithiothreitol, 10 M ZnCl 2 for 1 h at room temp followed by 10 min on ice. DNase I (0.15 units) was added and incubated for 1 min at room temperature. The resultant products were size-separated on an 8 M urea, 6% polyacrylamide gel and visualized by autoradiography at Ϫ70°C with intensifying screen for 48 h.
Gel Mobility Shift Assays-Gel mobility shift assays were performed essentially as described previously (8). Briefly, 10,000 cpm of 32 P-labeled single-or double-stranded DNA was incubated with 1 g of K562 cell nuclear extract, with or without a 50-fold molar excess of a competing oligonucleotide for 1 h. DNA-protein complexes were resolved by electrophoresis through 5% polyacrylamide and visualized by autoradiography at Ϫ70°C with intensifying screen for 16 h. Relative amounts of shifted probes were quantitated with an LKB Ultro XL laser densitometer (Pharmacìa Biotech Inc.).
Double-stranded probes were constructed by combining equimolar amounts of sense and antisense strands and heating to 100°C for 2 min followed by slowly cooling to room temperature. Prior to most studies double-stranded probe was purified by electrophoresis through 15% polyacrylamide followed by elution into 10 mM Tris, 0.1 mM EDTA, pH 7.4.
Depletion of K562 cell nuclear extract of Sp1 was accomplished by immunoprecipitation of Sp1 protein. 400 g of K562 cell nuclear extract was included with 2 g of Sp1 polyclonal antibody for 1 h at 4°C. Protein A cell suspension (100 l of 10% solution) was added and incubated overnight at 4°C. Precipitate was removed by centrifugation, and supernatant was used in subsequent electrophoretic mobility shift assay (EMSA) studies.
Western Analysis-K562 cell nuclear extract (0.5 g) from vehicle or PMA-treated cells was size-separated by electrophoresis through a discontinuous 4 -10% SDS-PAGE gel followed by transfer to nitrocellulose. Sp1 was visualized using an affinity-purified rabbit polyclonal antibody following the manufacturer's recommended protocol using an anti-rabbit horseradish peroxidase-conjugated secondary antibody and a chemiluminescence detection system (Amersham Corp.). Relative amounts of Sp1 protein in vehicle and PMA-treated K562 cells were quantitated using a LKB Ultro XL laser densitometer (Pharmacia).
Statistical Methods-Data are reported as mean Ϯ S.E. unless otherwise stated. If normally distributed, multiple data sets were compared using a one way analysis of variance. Significant differences between individual means were determined using a two-tailed group t test and the Bonferroni procedure. If data were not normally distributed, multiple data sets were compared using one way analysis of variance on ranks followed by Dunn's test for comparison of individual groups. Statistical significance was assumed at p Ͻ 0.05.

The Upstream Region of the Human Thromboxane Receptor Gene Promotor Contains Phorbol Ester-responsive Regulatory
Elements-We have previously used human thromboxane receptor gene promotor-CAT chimeras transfected into K562 cells to localize a position-and orientation-independent phorbol ester-responsive element to the 110-bp fragment between 1.84 and 1.95 kb upstream of the start of transcription (8). These studies were hampered by low levels of reporter gene expression in the homologous constructs, possibly due to putative negative regulatory element(s) located near the start of transcription. Therefore, to facilitate more accurate delineation of phorbol ester-responsive elements, we engineered a series of heterologous promotor constructs by placing thromboxane receptor gene promotor fragments of interest upstream of the thymidilate kinase promotor that drives CAT expression (pBLCAT5). As depicted in Fig. 1, A and B responsiveness. Taken together, these results indicated that phorbol ester up-regulation of the thromboxane receptor gene required the C-rich element located between Ϫ1.938 and Ϫ1.925 kb upstream of the start of transcription. Furthermore, this element alone was not sufficient to confer phorbol ester responsiveness but appeared to work in concert with other elements located between 1.92 and 1.84 kb upstream of the transcription start site.
Identification of a K562 Nuclear DNA Binding Protein That Binds to the C-rich Phorbol Ester-responsive Element of the Human Thromboxane Receptor Gene Promotor as Sp1-Our functional studies indicated that a C-rich element between Ϫ1.938 and Ϫ1.925 was required for phorbol ester responsiveness of the thromboxane receptor gene. Computer analysis (12) suggested a potential AP-2 transcription factor binding site on the forward strand of this sequence, while the backward strand contained a GC box. In order to identify which transcription factor(s) from K562 cells bound to the C-rich element we used a double-stranded oligonucleotide corresponding to the C-rich sequence plus 6 bp of the flanking inverted repeat (C-rich probe) in EMSAs. C-rich probe bound to four major proteins from K562 nuclear extract. Competition for C-rich probe binding to the lowest and two higher molecular weight proteins was eliminated when a 50-fold molar excess of unlabeled probe was added to the reaction, while binding to an intermediate molecular weight protein was less efficiently displaced, suggesting that the intermediate protein bound in a less specific manner (Fig. 2). As a first approach to identifying the shifted K562 nuclear proteins, oligonucleotides corresponding to a number of known transcription factor binding sites were used as competitors. These studies revealed that only the Sp1 binding oligonucleotide was an effective competitor for C-rich probe binding, and the competition appeared to be most efficient for binding of the two higher molecular weight proteins (Fig. 2). Additional experiments were performed to determine if the high molecular weight K562 nuclear protein that bound to C-rich probe was Sp1. First, when EMSA was performed with K562 nuclear extract immunodepleted of Sp1, the two higher molecular weight DNA binding proteins were no longer observed (Fig. 3, lane 4). Second, when the anti-Sp1 antibody was included in the EMSA reaction, the two higher molecular weight bands "supershifted," indicating that the antibody recognized those DNA-protein complexes (Fig. 3, lane 5). Finally, when authentic Sp1 was used in EMSA studies with the C-rich probe, it shifted it in a manner identical to the higher molecular weight K562 nuclear protein (not shown). These studies confirmed that Sp1 from K562 cell nuclear extract bound to the C-rich element and was responsible for the high molecular weight bands observed on EMSA. To determine the exact site of Sp1 binding we performed DNase footprinting using promotor fragment Ϫ1.96/Ϫ1.84 and purified Sp1 protein (Fig. 4). Sp1 protected the region from Ϫ1.942 to Ϫ1.922 kb from DNase digestion. This region corresponds to the sequence 5Ј-GCTGC-CCCCGCCCCCACCCAG-3Ј, which includes the complete Crich element (underlined). Taken together with our functional studies of phorbol ester response (see above), the DNA binding studies suggested that a critical feature required for PMA enhancement of thromboxane receptor gene expression was Sp1 binding to the C-rich element.
Sp1 Is Up-regulated by Phorbol Ester in K562 Cells-Sp1 had not previously been implicated as mediating phorbol ester enhancement of gene expression. Therefore, we used EMSA to determine if Sp1 binding to the thromboxane receptor gene promoter C-rich region differed quantitatively in nuclear extracts from K562 cells treated with vehicle compared with 100 nM PMA. PMA treatment increased Sp1 binding to the C-rich probe 5.4 Ϯ 1.2-fold over vehicle (p ϭ 0.02, n ϭ 4) (Fig. 5A). In contrast, binding to the intermediate nuclear weight protein did not change, thus demonstrating equal loading of nuclear extract. In addition, Sp1 protein levels, assessed by Western blot analysis, increased 2.4 Ϯ 0.1-fold (p ϭ 0.002, n ϭ 3) in K562 cells stimulated with PMA (Fig. 5B). Interestingly, there was an equivalent increase in both the phosphorylated and nonphosphorylated forms of Sp1 after PMA treatment. Thus, increased Sp1 binding to the thromboxane receptor gene promotor appeared to be due to an increase in Sp1 protein levels, not phosphorylation of existing protein. Finally, Northern analysis showed that Sp1 mRNA levels increased following PMA treatment of K562 cells and that the increase in Sp1 mRNA preceded the phorbol ester-stimulated increase in thromboxane receptor mRNA (Fig. 5C). In aggregate, these results support the notion that phorbol ester treatment of K562 cells increases thromboxane receptor gene transcription in part by increasing Sp1 expression and DNA binding.
Specific Nucleotide Alterations in the Phorbol Ester-responsive Element of the Thromboxane Receptor Gene Affect DNA-Protein Interactions-To identify the nucleotide sequence requirements for Sp1 binding to the thromboxane receptor gene promoter C-rich element, several mutant double-stranded oligonucleotides were tested for their ability to bind K562 nuclear proteins (Fig. 6A). An oligonucleotide where A was substituted for C and T was substituted for G in the C-rich region (C-rich mut), did not bind Sp1 (Fig. 6B). Replacing the G and A within the C-rich region with cytosines (14C) eliminated binding to Sp1 but did not alter binding of the two lower molecular weight nuclear proteins. Extension of the C-rich probe flanking sequence to include the entire octanucleotide inverted repeat (C-richϩ) or mutation of the flanking sequence, while retaining an inverted repeat, (C-rich mut flank) did not affect K562 nuclear protein binding. These mutational studies are consistent with the idea that Sp1 binds to the C-rich element and depends on the presence of the GC box located on the inverse strand of C-rich probe.
During the EMSA studies we noted that one of the low molecular weight DNA binding proteins was observed in some experiments but not others (see Figs. 2 and 6B). We determined that this protein was detected primarily in studies where the double-stranded DNA probe had not been gel-purified prior to use but was not readily apparent when double-stranded probe was gel-purified. This suggested that the protein was binding to only one strand of the C-rich region. To investigate this point, studies were conducted using gel-cut double-stranded C-rich probe, sense strand C-rich probe, and antisense strand C-rich probe (Fig. 7). Interestingly, the low molecular weight nuclear protein specifically bound to the sense strand but not the antisense strand probe. This is confirmed by the appearance of Sp1 binding, with concurrent loss of the low molecular weight band, when an excess of unlabeled antisense strand was FIG. 6. A, sequences of mutant double-stranded oligonucleotides used in EMSA to examine nucleotide requirements for Sp1 binding to C-rich probe. B, results of EMSA examining K562 cell nuclear proteins binding C-rich mutant double-stranded oligonucleotides. Radiolabeled probes (denoted above lanes) were incubated with (lanes ϩN and ϩC) or without (lanes P) K562 cell nuclear extract. Lanes ϩC included a 50-fold molar excess of unlabeled competitor DNA. Any perturbation of the C-rich region (C-rich mut and 14C) interferes with Sp1 binding to the probe.

FIG. 7.
Results of EMSA examining binding of K562 cell nuclear proteins to double-stranded, sense, and antisense strands of C-rich probe. Radiolabeled probes (denoted above lanes) were incubated with (lanes NE, ϩds, ϩs, and ϩa) or without (lanes P) K562 cell nuclear extract. Lanes ϩds included a 50-fold molar excess of unlabeled double-stranded C-rich probe, lanes ϩs included a 50-fold molar excess of unlabeled C-rich sense strand probe, and lanes ϩa included a 50-fold molar excess of unlabeled C-rich antisense strand probe. Sp1 bound specifically to double-stranded C-rich probe, while the low molecular weight protein (ssCRBP) bound specifically to the sense strand of the C-rich probe. added to sense strand probe, thus forming the double strand form (Fig. 7, lane 10). We have designated this protein ssCRBP (single-stranded C-rich binding protein). EMSA studies using mutant oligonucleotide C-rich mut revealed that removing C residues from the element, in addition to preventing Sp1 binding also, eliminated ssCRBP binding, whereas mutation of the flanking sequence (C-rich mut flank) had no effect on either Sp1 or ssCRBP binding (data not shown). Thus, the ssCRBP binding site overlaps the Sp1 site but is specific for singlestranded DNA, while Sp1 only binds double-stranded DNA.
A putative functional role for ssCRBP competition with Sp1 at the thromboxane receptor gene promoter C-rich element would require 1) that the DNA strand could exist in both single-and double-stranded conformations and 2) that ss-CRBPs have an affinity for the C-rich element that approximates or exceeds that of Sp1. The presence of the octanucleotide inverted repeat flanking the C-rich element is permissive for a stem loop structure where the C-rich element would exist as a single-strand, thus satisfying the first condition. To address the second condition we characterized the DNA binding affinities of Sp1 and ssCRBP in EMSA experiments where binding to radiolabeled single-or double-stranded C-rich probe was competed for by increasing amounts of the respective unlabeled probe (Fig. 8). The shifted band and free probe were excised from the gel separately, and the radioactivity of bound and free probe were quantitated. Scatchard transformation of these data resulted in a K d of 14 nM for Sp1 (Fig. 8, right panel) and a K d of 1.9 Ϯ 0.7 nM (n ϭ 2) for ssCRBP (Fig. 8, left panel). Thus, ssCRBP's affinity for sense strand C-rich element is greater than that of Sp1 for double-stranded C-rich element.

DISCUSSION
These studies of transcriptional regulation of the human thromboxane receptor gene have localized and delineated a phorbol ester-responsive element to 14 base pairs in the promotor located from Ϫ1.938 to Ϫ1.925 kb 5Ј of the start of transcription. Functional studies indicated that this element acts synergistically with unidentified downstream sequences to confer PMA responsiveness upon this gene. The DNA-binding nuclear factor that binds to this element was identified as transcription factor Sp1, which we determined was itself upregulated by PMA. The critical role that Sp1 plays in PMAstimulated enhancement of thromboxane receptor transcription was demonstrated by mutation of the Sp1 binding site, which eliminated both Sp1 binding to, and PMA responsiveness of, the promotor.
Our laboratory has been exploring the regulatory mechanisms for the human platelet thromboxane receptor gene for several years. We and others have previously shown that phorbol esters, which both activate protein kinase C and stimulate megakaryocytic differentiation of cultured platelet-like leukemia cells, increase the level of thromboxane receptor mRNA transcripts and increase the binding capacity of membrane thromboxane receptors (7,8,13,14). Furthermore, our laboratory has shown that thrombin, a circulating platelet-stimulating factor, also activates protein kinase C and up-regulates thromboxane receptor mRNA in these cells (7,15). Thus, elucidation of the PMA (and presumably thrombin) response mechanism(s) for this gene may reveal regulatory mechanisms important in the pathogenesis of clinical syndromes, such as acute myocardial infarction or unstable angina pectoris, in which both thrombin and the number of platelet thromboxane receptors are increased (4,16,17).
Recently, we reported the initial characterization of the human (K562 cell) thromboxane receptor gene promoter (8). Although our 2-kb promoter constructs were phorbol ester-responsive, we could not identify a classic Fos-Jun (AP-1) binding element within that region of the promoter. However, several potential AP-2 binding sites were identified, which we postulated could transduce the phorbol-ester response of this gene. Our initial experiments with the heterologous thromboxane receptor gene promoter/thymidilate kinase constructs described herein seemed to indicate that one of the putative AP-2 sites, located between Ϫ1.94 and Ϫ1.92, was the PMA response element. However, subsequent studies indicated that AP-2 was not the responsible transcription factor. First, we could not detect AP-2 in immunoblots of K562 nuclear extract. 2 Second, performance of EMSA using this element and an excess of unlabeled AP-2-binding probe failed to compete for K562 nuclear protein binding (see Fig. 2). Finally, anti-AP-2 antibody did not supershift any of the bands noted on EMSA. 2 In contrast, Sp1 was identified as the transcription factor that bound to the PMA response element in the thromboxane receptor gene promoter, and interestingly, the opposite strand of the putative AP-2 site contained an Sp1-binding GC box. Sp1 is a widely expressed RNA polymerase II transcription factor, originally isolated from HeLa cells (18). The traditionally accepted role for Sp1 is that it binds to one or more GC boxes (consensus 5Ј-GGGCGG-3Ј) located near a transcription start site and, in concert with other members of the RNA polymerase complex, initiates transcription (19 -21). Although many genes regulated by Sp1 contain multiple GC boxes, a single Sp1 binding site is generally sufficient for transcriptional activation (11). While the TATA-less thromboxane receptor gene promotor studied herein contains GC boxes located in the proximity of its transcription start site (on forward or reverse strand at positions Ϫ176 to Ϫ171, 102-107, and 272-277 (8)), our studies determined that an Sp1 binding site located approximately 1.9 kb 5Ј of the start of transcription functioned as an enhancer of promotor activity after PMA stimulation. This unexpected finding implies that Sp1 has very different regulatory roles when it binds to upstream promoter regions compared with binding at or near the transcription start site. This notion is supported by a previous report that a CAT reporter construct containing an Sp1 binding GC box 1.7 kb from the transcription start site can act as a transcriptional activator in Drosophila SL2 cells, but only under conditions where Sp1 protein expression is increased (22). In short, under some circumstances Sp1 can act as a transcriptional enhancer independent of its actions in transcriptional initiation.
Our studies indicated that Sp1 binding alone is not sufficient for PMA enhancement of human thromboxane receptor gene expression but that additional downstream elements located between Ϫ1.91 and Ϫ1.84 are also required. It has previously been reported that Sp1 interacts with other transcription factors, including GATA-1, YY-1 and NF-B (23)(24)(25). In the case of GATA-1, this interaction synergistically enhances transcription of globin genes during erythroid cell differentiation and requires increased levels of Sp1 expression (25). In vitro studies of Sp1-regulated promoter activity have also indicated a requirement for coactivators that are thought to tether Sp1 to the basal transcriptional machinery (26). In a similar manner, Sp1 apparently requires an as yet unidentified co-factor to function as a PMA-responsive transcriptional enhancer of the thromboxane receptor gene.
Our finding of a single-strand-specific factor (ssCRBP) whose DNA binding site overlaps that of Sp1 raises an interesting possibility regarding the mechanism of PMA responsiveness in the thromboxane receptor gene. The region flanking the Sp1/ ssCRBP binding site contains an octanucleotide inverted repeat sequence that would allow the formation of a DNA stemloop structure such that the Sp1/ssCRBP binding site exists as a single strand permissive for ssCRBP binding, but not Sp1 binding, and therefore resulting in no transcriptional enhancement (Fig. 9). Single-and doubled-stranded DNA configurations may exist in a dynamic equilibrium, with the predominant form determined by the relative levels of Sp1 and ssCRBP. In this scenario, increased Sp1 expression following PMA treatment would result in a shift of this equilibrium toward the Sp1 binding linear form of the DNA with resultant transcriptional enhancement. Interestingly, we found that PMA treatment did not alter ssCRBP binding to the C-rich element. Thus, regulation of Sp1 levels appears to be the major factor influencing the equilibrium between Sp1 and ssCRBP. A similar model has been proposed for the rat preprotachykinin-A promotor where a region functionally shown to enhance transcription can form a stem-loop structure (27). This region can bind Sp1 and AP-2 when double-stranded, but the single sense strand binds an unidentified single-strand-specific protein. An element that regulates myoblast and fibroblast vascular smooth muscle ␣-actin gene expression has also been iden- tified in which double-stranded DNA binds transcription enhancer factor-1, while both the sense and antisense strands alone can bind other unidentified nuclear proteins (28).
The current paper and a number of other recent reports have modified our ideas regarding the role of Sp1 in gene expression, and it appears that an Sp1 mechanism of transcriptional enhancement, such as that demonstrated for the human thromboxane receptor gene, may be widespread. The regulatory effects of Sp1 for globin gene expression during erythroid cell differentiation were noted above (25). A similar requirement for Sp1 has been reported for induction of the skeletal ␣-actin promoter during cardiac myocyte hypertrophy (29). A regulatory role for Sp1 has now been reported for epidermal growth factor stimulation of the human gastrin promoter (30), cAMPinduced transcription of the bovine cyp11A gene (31) and glucose activation of the acetyl-CoA carboxylase promoter (32). We have performed preliminary studies that indicate that glucocorticoids, like PMA, increase K562 thromboxane receptor gene expression while increasing Sp1 mRNA expression, Sp1 protein levels, and Sp1 binding to the C-rich element. 3 The thromboxane receptor gene promotor contains no glucocorticoid response element (8), further suggesting that Sp1 expression can itself be regulated by glucocorticoids as well as phorbol esters and that Sp1 modulation is a regulatory determinant of the expression of this and other genes.
In conclusion, far from being a ubiquitously expressed "housekeeping" transcription factor, Sp1 is a dynamically regulated enhancer that appears to play a critical regulatory role for numerous genes and may be a final common pathway for the regulation of gene expression by a variety of independent stimuli. Further characterizing the roles of Sp1 and ssCRBP in thromboxane receptor gene expression and determining the identities of necessary co-factors is the object of ongoing investigation in our laboratory.