Prostaglandin E2 stimulates bone sialoprotein (BSP) expression through cAMP and fibroblast growth factor 2 response elements in the proximal promoter of the rat BSP gene.

Bone sialoprotein (BSP), an early marker of osteoblast differentiation, has been implicated in the nucleation of hydroxyapatite during de novo bone formation. Prostaglandin E2 (PGE2) has anabolic effects on proliferation and differentiation of osteoblasts via diverse signal transduction systems. Because PGE2 increases the proportion of functional osteoblasts in fetal rat calvarial cell cultures, we investigated the regulation of BSP, as an osteoblastic marker, by PGE2. Treatment of rat osteosarcoma UMR 106 cells with 3 microm, 300 nm, and 30 nm PGE2 increased the steady state levels of BSP mRNA about 2.7-, 2.5-, and 2.4-fold after 12 h. From transient transfection assays, the constructs including the promoter sequence of nucleotides (nt) -116 to +60 (pLUC3) were found to enhance transcriptional activity 3.8- and 2.2-fold treated with 3 microm and 30 nm PGE2 for 12 h. 2-bp mutations were made in an inverted CCAAT box (between nt -50 and -46), a cAMP response element (CRE; between nt -75 and -68), a fibroblast growth factor 2 response element (FRE; nt -92 to -85), and a pituitary-specific transcription factor-1 motif (between nt -111 and -105) within pLUC3 and pLUC7 constructs. Transcriptional stimulation by PGE2 was almost completed abrogated in constructs that included 2-bp mutations in either the CRE and FRE. In gel shift analyses an increased binding of nuclear extract components to double-stranded oligonucleotide probes containing CRE and FRE was observed following treatment with PGE2. These studies show that PGE2 induces BSP transcription in UMR 106 cells through juxtaposed CRE and FRE elements in the proximal promoter of the BSP gene.

Prostaglandins are considered important local factors that modulate bone metabolism through their effects on osteoblastic cells and osteoclasts (1,2). Prostaglandin E 2 (PGE 2 ), 1 a major eicosanoid produced by osteoblasts, is a potent stimulator of bone resorption (3) that can stimulate the formation of osteoclast-like multinuclear cells in mouse bone marrow cultures (4,5). The effects of PGE 2 on osteoclastogenesis are, at least in part, mediated by osteogenic cells, which express macrophage colony-stimulating factor (6) and receptor activator of nuclear factor B ligand (RANKL) (7) that promote, and osteoprotegerin, a decoy receptor for RANKL (8), that suppresses osteoclast formation. PGE 2 has been shown to stimulate RANKL and inhibit osteoprotegerin production (7,9) and also increases production of interleukin-6, which can further enhance osteoclastogenesis (10 -12). In contrast, studies have revealed that PGE 2 also has bone-forming activity (2,13). Treatment of male, female, and overiectomized mice with PGE 2 increases bone mass in vivo (14), whereas PGE 2 stimulates collagen and DNA synthesis and induces bone growth in calvarial organ (15) and cell cultures in vitro (16,17). However, PGE 2 can either stimulate or inhibit cellular growth and differentiation of osteoblastic cells depending on PGE 2 concentration (15,18,19).
To explain the diverse effects of PGE 2 , the presence of multiple receptors for PGE 2 in osteoblasts was postulated. Recent cloning of four subtypes of the PGE receptor has made it possible to analyze the PGE receptor subtypes (EP1-EP4) on osteoblasts (3,13). EP1 is coupled to Ca 2ϩ mobilization, EP2 and EP4 activate adenylate cyclase, whereas EP3 inhibits adenylate cyclase (20 -22). An EP1 agonist stimulated cell growth and inhibited alkaline phosphatase activity, whereas an EP4 agonist reduced cell growth and increased alkaline phosphatase activity in MC3T3-E1 osteoblast-like cells (23). These studies indicate that osteoblasts express multiple subtypes of the PGE receptor and that each subtype is might be linked to different aspects of PGE 2 action. Thus, activation of the EP4 receptor stimulates bone formation and prevents bone loss (24), whereas bone resorption by lipopolysaccharide is impaired in EP4 knockout mice (25). Collectively, these results show that PGE 2 has anabolic effects on bone formation.
Bone sialoprotein (BSP) is a highly sulfated, phosphorylated, and glycosylated protein that is characterized by its ability to bind to hydroxyapatite through polyglutamic acid sequences and to mediate cell attachment through an RGD sequence (26 -28). The temporospatial deposition of BSP into the extracellular matrix (29,30) and the ability of BSP to nucleate hydroxyapatite crystal formation (31) indicate a potential role for this protein in the initial mineralization of bone, dentin, and cementum. Recent studies have shown that BSP is also expressed by osteotropic cancers, suggesting that BSP might play a role in the pathogenesis of bone metastases (32,33). Thus, regulation of the BSP gene appears to be important in the differentiation of osteoblasts, in bone matrix mineralization, and in tumor metastasis. The rat, human, and mouse BSP genes have been cloned and partially characterized (34 -37). These promoters include a functional inverted TATA element (nt Ϫ24 to Ϫ19) (38), which overlaps a vitamin D response element (39), and an inverted CCAAT box (Ϫ50 to Ϫ46), which is required for basal transcription (40,41). In addition, a fibroblast growth factor 2 (FGF2) response element (FRE; Ϫ92 to Ϫ85) (42), a cAMP response element (CRE; Ϫ75 to Ϫ68) (43), a transforming growth factor-␤ activation element (Ϫ499 to Ϫ485) (44), a pituitary-specific transcription factor-1 (Pit-1) motif (Ϫ111 to Ϫ105) that mediates the stimulatory effects of parathyroid hormone (45), and a homeodomain binding element (Ϫ199 to Ϫ192) (46) have been characterized. Further upstream, a glucocorticoid response element overlapping an AP-1 site (27,47) has also been identified.
Because BSP is a marker of osteoblastic differentiation and bone formation, we have analyzed the effects of PGE 2 on BSP expression in UMR 106 cells. Our studies show that PGE 2 increases transcription of the BSP gene through cAMP-dependent protein kinase, tyrosine kinase, and MAP kinase pathways and that the effects are mediated via CRE and FRE transcriptional elements in the proximal promoter of the rat BSP gene.
Cell Culture-The rat clonal cell line, UMR 106 cells (generously provided by Dr. T. J. Martin) were cultured at 37°C in 5% CO 2 air in ␣-minimum essential medium (␣-MEM) supplemented with 10% fetal bovine serum and used in these studies as an osteoblastic cell line that synthesizes BSP (27,48). Rat stromal bone marrow cells (49), were kindly provided by Dr. S. Pitaru (Tel Aviv University, Tel Aviv, Israel). The cells were first grown to confluence in 60-mm tissue culture dishes in ␣-MEM medium containing 10% fetal bovine serum, then cultured in ␣-MEM without serum, and incubated with prostaglandin E 2 . RNA was isolated from triplicate cultures and analyzed for expression of BSP mRNA by RT-PCR and real time PCR as described below.
RT-PCR and Real Time PCR-Following treatment total RNA was extracted from UMR 106 cells with guanidium thiocyanate at different times, as described previously (44), and 1 g was used as a template for one-step RT-PCR and cDNA synthesis. cDNA was prepared using random hexamer and Moloney murine leukemia virus reverse tran-scriptase RNase H Ϫ . Conventional one-step RT-PCR was performed using a SuperScript one-step RT-PCR kit. The primers were synthesized on the basis of the reported rat cDNA sequences for BSP and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Sequences of the primers used for PCR were as follow: BSP forward, 5Ј-CTGCTTTA-ATCTTGCTCTG-3Ј; BSP reverse, 5Ј-CCATCTCCATTTTCTTCC-3Ј; GAPDH forward, 5Ј-CCATGTTTGTGATGGGTGTG-3Ј; and GAPDH reverse, 5Ј-GGATGCAGGGATGATGTTCT-3Ј. cDNA synthesis and predenaturation was performed for 1 cycle at 50°C for 30 min and 94°C for 2 min, and amplification was carried out for 30 (BSP and GAPDH) cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and final extension was 94°C for 10 min in a 50-l reaction mixture. After amplification, 10 l of each reaction mixture was analyzed by 2% agarose gel electrophoresis, and the bands were then visualized by ethidium bromide staining. The expected size of the PCR products for BSP and GAPDH were 211 and 264 bp, respectively. Quantitative real time PCR was performed using the following primer sets: BSP-R-T forward, 5Ј-TCCTCCTCTGAAACGGTTTCC-3Ј; BSP-R-T reverse, 5Ј-CGAACTATCGCCATCTCCATT-3Ј; GAPDH-R-T forward, 5Ј-AGATG-GTGAAGGTCGGTGTC-3Ј; and GAPDH-R-T reverse, 5Ј-ATTGAACTT-GCCGTGGGTAG-3Ј using the SYBR Green qPCR Kit in a DNA Engine Opticon 2 continuous fluorescence detection system (MJ Research Inc.). The expected size of the PCR products for BSP and GAPDH were 73 and 167 bp, respectively. The amplification reactions were performed in 20 l of final volume containing 1ϫ SYBR Green Master Mix, 0.25 M primer mixture, and 10 ng of cDNA. To reduce variability between replicates, PCR premixes, which contain all of the reagents except for cDNA, were prepared and aliquoted into 0.2-ml thin wall strip tubes (MJ Research Inc.). The thermal cycling conditions were 40 cycles of the following protocol: 15 s of denaturation at 95°C, 50 s of annealing at 64°C, followed by 12 s of extension at 77°C. Post-PCR melting curves confirmed the specificity of single-target amplification, and fold expression of BSP relative to GAPDH was determined in triplicate (50).
Transient Transfection Assays-Exponentially growing UMR 106 cells were used for transfection assays. 24 h after plating, the cells at 50ϳ70% confluence were transfected using a LipofectAMINE reagent. The transfection mixture included 1 g of a luciferase (LUC) construct (45) and 2 g of pSV-␤-galactosidase vector as an internal transfection control. Two days post-transfection, the cells were deprived of serum for 12 h, and 3 M or 30 nM PGE 2 or 3 M of the respective EP agonists were added for 12 h prior to harvesting. The luciferase assay was performed according to the supplier's protocol (picaGene; Toyo Inki) using a Luminescence reader BLR20 (Aloka) to measure the luciferase activity. The protein kinase inhibitorsH89 (5 M) and H7 (5 M) were used to inhibit protein kinase A and C. Herbimycin A (1 M) and PP1 (10 M) were used for tyrosine kinase and Src tyrosine kinase inhibition, respectively (42,51). U0126 (5 M) was used to inhibit MAP kinase kinase activity (52). Sodium orthovanadate (50 M) and okadaic acid (50 nM) were used for tyrosine phosphatase and serine-threonine phosphatase inhibition, respectively (53,54). Forskolin (1 M) was used for activation of adenylate cyclase (45). Oligonucleotide-directed mutagenesis by PCR was utilized to introduce dinucleotide substitutions using the QuikChange site-directed mutagenesis kit (Stratagene). All of the constructs were sequenced as described previously (42) to verify the fidelity of the mutagenesis.
[Ca 2ϩ ] i was calculated from the measurement of the ratio of fluorescence intensities (57,58). All of the experiments were performed three times with different cell batches.
Statistical Analysis-Triplicate samples were analyzed for each experiment, and the experiments were replicated to ensure consistency of the responses to PGE 2 . Significant differences between control and PGE 2 treatment were determined using Student's t test.

Stimulation of BSP mRNA Expression in UMR 106
Cells-BSP gene expression was investigated at 6 and 12 h after PGE 2 stimulation by conventional (Fig. 1A) and real time PCR (Fig.  1B). When osteoblastic UMR 106 cells were exposed to 3 M, 300 nM, and 30 nM PGE 2 , expression of BSP mRNA was increased 2.3-, 2.0-, and 2.2-fold at 6 h and 2.7-, 2.5-, and 2.4-fold at 12 h, respectively, as shown by conventional RT-PCR (Fig.  1A). To further confirm the PGE 2 effects on BSP transactivation, we applied real time PCR to examine the mRNA expression level of BSP. As moter constructs ligated to a luciferase reporter gene were transiently transfected into UMR 106 cells, and their transcriptional activity was determined in the presence of PGE 2 . With the construct pLUC3, encompassing BSP promoter nucleotides Ϫ116 to ϩ60, transcription was increased 3.8-fold with 3 M PGE 2 and 2.2-fold with 30 nM PGE 2 after 12 h of treatment (Fig. 2, A and B). PGE 2 also increased transcription of pLUC4 (Ϫ425 to ϩ60), pLUC5 (Ϫ801 to ϩ60), pLUC6 (Ϫ938 to ϩ60), and pLUC7 (Ϫ1149 to ϩ60). In shorter constructs (pLUC1, Ϫ18 to ϩ60; pLUC2, Ϫ43 to ϩ60), luciferase activities were not increased by PGE 2 (data not shown). When transcriptional activity in response to 30 nM PGE 2 was analyzed in normal rat stromal bone marrow cells (49), the transcriptional activity of pLUC3 was increased 2-fold (Fig. 2C). Within the DNA sequence that is unique to pLUC3 (between nt Ϫ116 and Ϫ43), an inverted CCAAT box (ATTGG; between nt Ϫ50 and Ϫ46), CRE (between nt Ϫ75 and Ϫ68), FRE (between nt Ϫ92 and Ϫ85), and the Pit-1 motif (between nt Ϫ111 and Ϫ105) are present (Fig. 3).
Because PGE 2 signaling activities are mediated by different protein kinases, we investigated the effects of the protein kinase C inhibitor H7 (5 M), the cAMP-dependent protein kinase inhibitor H89 (5 M), the tyrosine kinase inhibitor herbimycin A (1 M), the Src kinase inhibitor PP1 (10 M), and the MAP kinase kinase inhibitor U0126 (5 M) on PGE 2 -mediated transcription to determine the signaling pathway. Whereas PGE 2induced pLUC3 promoter activation was inhibited by H89, herbimycin A, PP1, and U0126, no effect was observed for H7 (Fig. 4), indicating an involvement of cAMP-dependent protein kinase, tyrosine kinase (Src), and MAP kinases in mediating the effects on BSP transcription. To assay for the responsiveness of the BSP promoter to serine-threonine phosphorylation, tyrosine phosphorylation, or elevated intracellular cAMP level, we used the serine-threonine phosphatase inhibitor okadaic acid, tyrosine phosphatase inhibitor sodium orthovanadate, and also forskolin, which is known to stimulate an adenylate cyclase. Vanadate (50 M) and forskolin (1 M) stimulated pLUC3 promoter activity ϳ1.6and ϳ1.7-fold, respectively, whereas okadaic acid (50 nM) was without effect (Fig. 5). Simultaneous stimulation with vanadate (50 M) and PGE 2 (3 M) up-regulated pLUC3 promoter activity synergistically. However, a combination of forskolin and PGE 2 increased pLUC3 transcription to the same level observed for PGE 2 stimulation (Fig. 5).
To determine which PGE 2 receptor subtype transduced the PGE 2 effects on BSP transcription, the following receptor agonists were used in the transcription assays: 3 M 17-phenyl trinor PGE 2 (for EP1 and EP3), ONO-AP-324-01 (for EP3), butaprost (for EP2), and prostaglandin E 1 alcohol (for EP2 and EP4) (Fig. 6). Butaprost and prostaglandin E 1 alcohol stimulated pLUC3 promoter activity to a similar extent as PGE 2 , whereas no significant increase in transcription was observed with either 17-phenyl trinor and ONO-AP-324-01, indicating that PGE 2 activates BSP transcription by a mechanism involving cAMP stimulation through EP2 and EP4 receptors.
To determine the regulatory element(s) between nt Ϫ116 and Ϫ43 that is utilized by PGE 2 , a series of 5Ј deletion constructs were prepared. Transcription by constructs Ϫ116BSPLUC and Ϫ108BSPLUC was increased by PGE 2 , but no increase was seen with Ϫ84BSPLUC. These results indicated that the element responding to PGE 2 was present between nt Ϫ108 and Ϫ85 in the BSP promoter (Fig. 7). Next we introduced mutations in the possible response elements encoded within nt Ϫ116 to ϩ60 of pLUC3. In addition, we examined whether these sites function in the large promoter construct (nt Ϫ1149 to ϩ60; pLUC7), as shown in Fig. 8. Whereas mutations in the Pit-1 had little effect on PGE 2 stimulation and mutation of the CCAAT box essentially abolished basal expression, mutations of the CRE and especially the FRE significantly reduced the PGE 2 effects on the transcriptional activities (Fig. 8). Furthermore, when both CRE and FRE sites were mutated, PGE 2induced luciferase activity was completely abolished (Fig. 8). These results suggest that the FRE and possibly the CRE are required as functional cis-elements for up-regulation of BSP transcription by PGE 2 .
Gel Mobility Shift Assays-To identify nuclear proteins whose binding to the CRE and FRE elements might be modulated by PGE 2 , double-stranded oligonucleotides were end-la- FIG. 9. PGE 2 up-regulates a nuclear protein that recognizes the CRE and FRE. Radiolabeled double-stranded CRE ( Ϫ84 CCCA-CAGCCTGACGTCGCACCGGCCG Ϫ59 ) and FRE oligonucleotides ( Ϫ98 TTTTCTGGTGAGAACCCACA Ϫ79 ) were incubated for 20 min at 21°C with nuclear protein extracts (3 g) obtained from UMR 106 cells incubated without (lanes 1 and 4) or with PGE 2 at 30 nM for 6 h (lanes 2 and 5) and 12 h (lanes 3 and 6). DNA-protein complexes were separated on 5% polyacrylamide gel in low ionic strength Tris borate buffer, dried under vacuum, and exposed to an imaging plate for quantitation using an image analyzer.  lanes 11 and 12). DNA-protein complexes were separated on 5% polyacrylamide gel in low ionic strength Tris borate buffer, dried under vacuum, and exposed to an imaging plate for quantitation using an imaging analyzer.  lanes 11 and 12). DNA-protein complexes were separated on 5% polyacrylamide gel in low ionic strength Tris borate buffer, dried under vacuum, and exposed to an imaging plate for quantitation using an imaging analyzer. beled and incubated with equal amounts (3 g) of nuclear proteins extracted from confluent UMR 106 cells that were either not treated (control) or treated with 30 nM PGE 2 for 6 and 12 h. When the CRE and FRE were used as probes, the formation of FRE-protein complexes (Fig. 9, lanes 4 -6) and slowly migrating CRE-protein complexes were increased by PGE 2 (Fig. 9, lanes 1-3). That these DNA-protein complexes represent specific interactions was demonstrated by competition experiments in which 20-and 40-fold molar excess of CRE and consensus CRE (Fig. 10, lanes 3, 4, 7, and 8) and FRE double-stranded oligonucleotides (Fig. 11, lanes 3 and 4) reduced by the amount of complex formation dose-dependently. In contrast, mutated CRE, FRE, and inverted CCAAT (Fig. 10,  lanes 5, 6, and 9 -12) and mutated FRE, CRE, consensus CRE, and inverted CCAAT oligonucleotides (Fig. 11, lanes 5-12) did not compete with CRE-protein and FRE-protein complex formation. To verify that the PGE 2 was operating through CRE and FRE, we also used gel mobility shift analyses to evaluate the potential effects of PGE 2 on the nearby inverted CCAAT and consensus CRE sites. When we used the inverted CCAAT sequence as a probe, the CCAAT-NF-Y protein complex (40,41,59) did not change after PGE 2 stimulation (Fig. 12, lanes 1-3). In comparison, CRE binding was increased by PGE 2 (Fig. 12,  lanes 4 -6). Notably, a stronger shift was obtained with the concensus CRE compared with the CRE in the proximal promoter of the BSP gene.
To further characterize the proteins in the complexes formed with the CRE and FRE, we used antibodies for several transcription factors. The addition of antibody to CREB disrupted the formation of the CRE DNA-protein complexes (Fig. 13, lane  4), whereas incubation of nuclear extracts with anti-phospho-CREB antibody produced a visible supershift complex (Fig. 13,  lane 5). FRE-nuclear protein complex was not disrupted or supershifted by antibodies to CREB, c-Jun, c-Fos, Pit-1, Oct-1, NFB p65, and NFB p50 (data not shown).

DISCUSSION
Prostaglandins are among the most potent regulators of bone cell function (4,17). Extensive studies have demonstrated that PGE 2 has both anabolic and catabolic effects on osteoblastic cells (13,18). Although the effects on bone resorption are indirect, involving the expression of cytokines such as RANKL, which promote osteoclast formation (7), prostaglandins can directly stimulate osteoblastic cells to differentiate and form bone (2,13). The expression of BSP, which is essentially specific to mineralized tissues and is expressed by newly formed osteoblasts coincident with mineralization, provides a valuable marker for osteogenic differentiation and bone formation (28). Our studies show that PGE 2 , consistent with its promoting osteogenesis, increases expression of BSP by activation of EP2 and EP4 receptors in UMR 106 cells. Transduction of the PGE 2 signaling is mediated by cyclic AMP-dependent protein kinase A, Src tyrosine kinase, and MAP kinase, which target nuclear proteins that bind to CRE and FRE elements in the proximal promoter of the BSP gene.
Prostaglandins, acting through different cell surface receptors on osteoblastic cells, stimulate bone remodeling by promoting both anabolic and catabolic responses, the relative responses being dependent on the target cell population and the concentration of PGE 2 . In bone marrow cells, which are targets for the anabolic actions of PGE 2 (60), PGE 2 can stimulate both phospholipase C and adelylate cyclase pathways in osteoblasts (2,10). The stimulation of phospholipase C results in the breakdown of phospholipid to form diacylglycerol, which activates protein kinase C (61), and inositol phosphates, which cause the release of intracellular concentration of free calcium ([Ca 2ϩ ] i ) (58). Although 3 and 6). DNA-protein complexes were separated on 5% polyacrylamide gel in low ionic strength Tris borate buffer, dried under vacuum, and exposed to an imaging plate for quantitation using an imaging analyzer. Control, nuclear extract from control confluent cells. naling. In contrast, stimulation of BSP transcription appears to utilize the cAMP-dependent protein-tyrosine kinase pathway because transcription is inhibited by herbimycin A and stimulated by vanadate and forskolin. Moreover, BSP transcription is mediated by EP2 and EP4 receptors (Fig. 6), through which cAMP production is stimulated (21). That transcription is suppressed by Src inhibitors to Src kinase and MAP kinase (Fig. 4) further implicates these enzymes in the signaling pathway.
BSP has been characterized as a unique marker of osteogenic differentiation that can regulate the formation of mineral crystals (28). In this study, we have identified response elements in the BSP gene promoter that mediate the PGE 2 action on BSP transcription. In UMR 106 cells, PGE 2 (3 M and 30 nM) stimulated BSP promoter activity (pLUC3) ϳ3.8and 2.2-fold (Fig.  2, A and B), comparable with the increase in BSP mRNA levels of ϳ2.7and 2.4-fold by conventional RT-PCR (Fig. 1A) and ϳ3.6and 3.7-fold by real time PCR (Fig. 1B). PGE 2 also induced BSP transcription in stromal bone marrow cells (Fig.  2C), indicating that the increased BSP expression occurs in normal osteoprogenitor cells and is not a specific feature of transformed UMR 106 cells. From transient transfection assays we initially located the PGE 2 -responsive region to the proximal promoter (nt Ϫ116 and Ϫ43; Fig. 2) of the BSP gene, which encompasses an inverted CCAAT box (nt Ϫ50 and Ϫ46), a putative CRE (nt Ϫ75 and Ϫ68), a FGF2 response element (FRE; nt Ϫ92 and Ϫ85), and a Pit-1 (nt Ϫ111 and Ϫ105) motif (Fig. 3). The results of luciferase analyses using fine 5Ј deletion constructs between nt Ϫ116 to Ϫ43 in the BSP promoter show that the PGE 2 effects are targeted to a region encompassed by nt Ϫ108 and Ϫ43 (Fig. 7). Whereas mutation of the Pit-1 element was without effect, mutation of the CCAAT element resulted in the loss of basal transcriptional activity, as reported previously (40,43). As a consequence the involvement of the inverted CCAAT was difficult to ascertain. However, the lack of PGE 2 -induced transcription with constructs Ϫ84BSPLUC and Ϫ60BSPLUC (Fig. 7) indicate that the CCAAT is not a target of PGE 2 regulation. In comparison, mutations in the CRE and FRE sites suggest that they are required for the induction of BSP expression by the PGE 2 . The involvement of the FRE and CRE elements is further supported by EMSA analyses in which proteins from nuclear extracts formed complexes with the FRE and CRE elements that were increased by PGE 2 (Fig. 9). However, although the luciferase assays show a much reduced PGE 2 -stimulated transcription when the individual CRE and FRE sites are mutated and the combined mutations show total abrogation, the formation of CRE-nuclear factor complexes is weak compared with results obtained with a concensus CRE (Figs. 9 and 12). Moreover, there is no significant increase in transcription with the Ϫ84BSPLUC construct (Fig. 7), which omits the FRE but retains the CRE element. In comparison, the FRE clearly shows increased binding of the nuclear protein in response to PGE 2 . Thus, our studies suggest that transcriptional activation is mediated by the juxtaposed FRE and CRE elements, with the FRE being the predominant target of the PGE 2 effects.
Although the CRE element binding of nuclear protein was not strong, the binding protein was, nevertheless up-regulated by PGE 2 (Fig. 9) and could be identified as CREB by antibody interference (Fig. 13). Moreover, phosphorylation of the CREB was induced by PGE 2 (Fig. 13). Although cAMP-dependent protein kinase signaling does not affect CREB binding to its cognate CRE element, it can direct phosphorylation of CREB, which is required for transcriptional activation (62,63). In comparison, the nuclear factor binding to the FRE element, which is regulated by tyrosine kinase, has yet to be characterized and is the focus of current studies because of its potential role in regulating basal and FGF2-induced transcription of BSP in osteoblasts (42), as well as mediating the PGE 2 effects.
The molecular pathways of PGE 2 regulation of BSP gene transcription are identified in these studies. PGE 2 acting through EP2 and EP4 prostaglandin receptors on osteoblastic cells activates signaling pathways involving cAMP generation and tyrosine phosphorylation (protein-tyrosine kinase), which activate MAP kinase to phosphorylate CREB and thereby transactivate BSP transcription through the CRE. Whether this same pathway also activates the FRE response through the same or a linked pathway or whether there is a concerted action on the CRE and FRE through a single pathway is difficult to discern at this time because the FRE-binding nuclear protein is yet to be characterized.
In conclusion, our study has identified CRE and FRE elements in the rat BSP proximal promoter that mediate BSP transcription induced by PGE 2 and that the PGE 2 increases the nuclear protein binding activities of CRE and FRE and enhances CREB phosphorylation. Because BSP is expressed by differentiated osteoblasts and PGE 2 is a crucial factor for bone metabolism, it is conceivable that these two response elements may contribute to the cell-specific expression of the BSP gene during the formation of the mineralized extracellular matrix of bone.