Biglycan Gene Expression in the Human Leiomyosarcoma Cell Line SK-UT-1

In this study we demonstrate that the gene encoding the small leucine-rich proteoglycan biglycan is expressed in human myometrial tissue and in the human leiomyosarcoma cell line SK-UT-1. Treatment of SK-UT-1 cells with forskolin or 8-bromo-cAMP strongly increased biglycan mRNA and this effect was transcriptional as shown by transient transfection experiments with biglycan promoter-luciferase reporter fusion genes. The cAMP-mediated induction of the transfected biglycan promoter in SK-UT-1 cells was abolished by coexpression of a specific protein kinase A inhibitor, and was mimicked by overexpression of the catalytic subunit (Cβ) of protein kinase A. By 5′ deletion analysis, part of the cAMP response was localized to the segment from residues −78 to −46 of the biglycan promoter. This region conferred strong cAMP responsiveness to a heterologous promoter. Electrophoretic mobility shift and antibody supershift assays identified two specific complexes that contained nuclear proteins antigenically related to the ubiquitous transcription factors Sp1 and Sp3, respectively. The binding site of these proteins was mapped to a CT-rich sequence extending from −59 to −49 in the biglycan promoter. Mutating this sequence eliminated complex formation and markedly reduced basal and cAMP-dependent promoter activity of transfected reporter genes. In vitro binding studies using recombinant Sp1 revealed that the nuclear factor binding to the CT element was not Sp1 but a Sp1-like protein(s). Western blot analysis of SK-UT-1 nuclear proteins confirmed expression of Sp3, Sp1 and nuclear proteins that crossreacted with Sp1 antibody but according to their molecular weight were not Sp1. These results indicate that all cAMP-dependent as well as some basal biglycan transcription in SK-UT-1 cells is mediated through activated protein kinase A and that both functions are conferred at the promoter level through the interaction of Sp1-like/Sp3 factors with the CT element at −59 in the biglycan promoter.

Biglycan (BGN) 1 (1) is a prototype member of a growing family of proteoglycans (PGs), the small leucine-rich proteoglycans (SLRPs) (reviewed in Refs. [2][3][4]. In most tissues BGN exists as a classical PG in which the core protein is substituted with two chondroitin/dermatan sulfate chains. As an extracellular PG, BGN is involved in the regulation of matrix assembly and growth factor activity due to its ability to bind collagen type I (5) and transforming growth factor-␤ (6), respectively. The cell surface-associated or pericellular localization of BGN (7), its binding to fibronectin (8,9), and the observation that changes in cell morphology and an enhanced substratum detachment are associated with a decrease in BGN synthesis in tumor necrosis factor-treated endothelial cells (10) suggest that this PG might interfere with cell adhesion. Recently, evidence has been presented that it also has a role in cellular migration processes, e.g. those of endothelial cells after wounding (11). However, the precise function(s) of BGN is not known at present. BGN is expressed ubiquitously, but within a given tissue constitutive gene activity is generally restricted to only one or a few cell types; among them are chondrocytes and osteoblasts (1,7,12), endothelial cells (7,13,14), fibroblasts (15,16), myofibroblasts (7,17,18), vascular smooth muscle cells (13,14,18), and some tumor cells (10,19). Furthermore, in the liver, BGN mRNA is specifically induced during trans-differentiation of Ito into myofibroblast-like cells (17). Despite considerable progress that has been made in describing the expression pattern of BGN under normal and various pathological conditions, little is known of how the information conveyed by extracellular stimuli leads to changes in BGN gene activity. Also, the cisand trans-acting regulatory elements involved in basal, tissue/ cell-specific, and differentiation-dependent expression have not been functionally characterized yet.
We (19) and others (15,20) have recently presented a structural and functional characterization of the human BGN 5Јflanking region. As expected for a protein with a constitutive and specific spatial and temporal expression pattern, multiple consensus sequences for known DNA-binding proteins were identified that when bound to their cognate trans-acting factors may support basal transcriptional activity or relay information from cytokine and hormone stimulation into changes of gene activity. For instance, the presence of several interleukin-6 response elements correlated with transcriptional activation by interleukin-6 in the breast cancer cell line T47D (19). Recently, a short region was identified in the BGN promoter that was differentially regulated in individuals with sex chromosome anomalies and that bound to the transcription factor c-Krox in vitro (20). Clustered in the proximal BGN promoter are putative binding sites for the ubiquitous transcription factors Sp1 and AP2, both of which can function as enhancer-binding proteins and can stimulate basal transcription through direct interaction with components of the basal transcription machinery. Both proteins are also targeted by certain signaling pathways through protein-protein interactions or post-translational modifications. For instance, AP2 activity is stimulated by phorbol 12-myristate 13-acetate (PMA) (21), and both AP2 and Sp1 have been shown to increase transcription in response to the second messenger cyclic AMP (cAMP) (21,22).
In the course of studies that investigated the role of cAMP in the regulation of the prolactin (PRL) gene in myometrial smooth muscle cells (mSMC) (23), it was discovered that BGN mRNA expression was strongly induced in long term cultures of myometrial tissue explants; since after several days in culture mSMC spontaneously increase the formation of cAMP (23), it was conceivable that up-regulation of the BGN gene occurred in response to cAMP in one or more myometrial cell types. Prompted by this observation in conjunction with recent results from our laboratory showing that the human BGN gene is transcriptionally regulated by cAMP in osteosarcoma cells (24), we decided to investigate in greater detail the molecular basis of basal and cAMP-regulated transcription in myometrial cells. By using the leiomyosarcoma cell line SK-UT-1, we demonstrate that 1) BGN mRNA expression is strongly up-regulated by forskolin and that this up-regulation is, at least in part, transcriptional, 2) induction of BGN gene expression by cAMP involves activation of the cAMP-dependent protein kinase A (PKA), and 3) basal and cAMP-induced transcription from the BGN gene is partially mediated through Sp1-like/Sp3 proteins binding to a pyrimidine-rich cis-element in the proximal promoter region. This study is thus the first one to provide an identification of a trans-acting factor(s) involved in BGN gene expression in vivo.

EXPERIMENTAL PROCEDURES
Tissues, Cell Lines, and Cell Culture-Human uterine tissues were obtained from either cycling or menopausal nonpregnant women at routine hysterectomy. Tissues were dissected post-surgically into endometrium and myometrium and rapidly frozen in liquid nitrogen (for RNA analysis) or cultured for 3 weeks. The isolation of primary mSMC and their cultivation as monolayers has been described earlier (23). The human leiomyosarcoma cell line SK-UT-1 (HTB-114, American Type Culture Collection, Rockville, MD) was obtained from the European Collection of Animal Cell Cultures (Salisbury, UK) and maintained in Dulbecco's modified Eagle's medium/Ham's F-12 (1:1) supplemented with 10% fetal calf serum, 2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin (all chemicals from Life Technologies, Inc., Eggenstein, Germany). To increase intracellular levels of cAMP, cells were treated with the indicated concentrations of either 8-Br-cAMP or forskolin (both from Calbiochem, Bad Soden, Germany).
RNA Isolation, Ribonuclease Protection Assays, and Northern Blot Hybridization-RNA was isolated using RNA Clean (AGS, Heidelberg, Germany). Prior to dissolution in RNA Clean, frozen pieces of human myometrial tissue were finely minced, whereas cultured cells were lysed directly in the wells. Poly(A) ϩ -enriched RNA was prepared by chromatography on oligo(dT)-cellulose (25). Ribonuclease protection assays were carried out with the RPA II kit (Ambion, Austin, TX) as described elsewhere (19). The probe used was a 181-base cRNA containing 158 bases of BGN sequence that had been transcribed in vitro with T7 RNA polymerase from nt ϩ1267 to the NcoI site at ϩ1110 of the human BGN cDNA (15). Northern blotting and hybridization were carried out as described in detail earlier (18). The Northern blots were probed with a 1144-bp human BGN cDNA fragment (nt ϩ124 to ϩ1267) generated by PCR as described previously (18). The probe DNAs were labeled with [␣-32 P]dCTP by the random priming procedure to a specific activity of 1 ϫ 10 9 cpm/g (26).
Reverse Transcriptase-PCR Analysis of BGN mRNA in SK-UT-1 Cells and Southern Blotting-Total RNA (5 g) was reverse-transcribed using oligo(dT) 16 and Superscript II reverse transcriptase (Life Technologies, Inc.) according to the manufacturer's instructions. After termination of the reaction, the reaction volume was diluted 5-fold with water, and 1/50th of first strand cDNA was used for PCR amplification. For BGN, a stringent hot-start-touch-down program was designed that comprised 35 cycles using oligonucleotide primers P ϩ 124 and P ϩ 1267 (18). In a parallel reaction identical amounts of cDNA were used for amplification of GAPDH cDNA for 16 cycles with primers GAPDH 5Ј (5Ј-GGCGTCTTCACCACCATGGAG-3Ј) and GAPDH 3Ј (5Ј-AAGTTGT-CATGGATGACCTTGGC-3Ј) resulting in a 206-bp fragment (nt ϩ358 to ϩ563). The BGN amplification products were blotted onto Hybond N membrane (Amersham-Buchler, Braunschweig, Germany) and hybridized with an internal 32 P-labeled cDNA fragment (nt ϩ150 to ϩ538 of the human BGN cDNA).
Plasmid Construction-A detailed description of the generation of the BGN promoter 5Ј deletion luciferase fusion genes designated BGNSac-Luc, BGNBam-Luc, BGNBglI-Luc, BGN-78-Luc, BGN-46-Luc, and BGN-1-Luc has been given previously (19). The plasmid Ϫ36rPRL-Luc containing the minimal promoter from the rat prolactin (rPRL) gene (nt Ϫ36 to ϩ34) was a generous gift from Dr. M. G. Rosenfeld, Howard Hughes Medical Institute, San Diego, CA. Inserting the Ϫ78 to Ϫ35 and Ϫ46 to Ϫ1 regions of the BGN gene in sense orientation into the SmaI site in the polylinker of Ϫ36rPRL-Luc gave rise to the plasmids Ϫ78hBGN-36rPRL-Luc and Ϫ46hBGN-36rPRL-Luc, respectively. These fragments had previously been amplified by PCR using Pfu polymerase (Stratagene) and the following oligonucleotide primers: 5Ј-CCGCCCTCTCCCCGCTGTCC-3Ј (nt Ϫ78 to Ϫ59) and 5Ј-TGGGCAG-GCGGGCCGACGGGGAGGGGACAG-3Ј (antisense to nt Ϫ64 to Ϫ35), or 5Ј-CCCGCCTGCCCAGCCTTTAGC-3Ј (nt Ϫ46 to Ϫ26) and 5Ј-GG-GAGACAGAGGCGGCGGG-3Ј (antisense to nt Ϫ19 to Ϫ1), respectively. The plasmid Ϫ78hBGNrev-36rPRL-Luc contains the Ϫ78 to Ϫ35 fragment from the human BGN gene promoter in reverse orientation in the SmaI site of Ϫ36rPRL-Luc. The plasmid CRE-36rPRL-Luc3 contains two composite cAMP-responsive elements (CREs, bold) (5Ј-TTG-GCTGACGTCAGAGAGAGGCCGGCCCCTTACGTCAGAGGCGAG-3Ј) linked to the rPRL promoter and the luciferase gene in the vector pGL3-Basic (Promega). Its construction is described elsewhere (27). A derivative of the plasmid BGN-78-Luc containing point mutations (BGN-78mut-Luc) was synthesized as follows: an upstream oligonucleotide primer (Ϫ78mut-l: 5Ј-CCGCCCTCTCCCCGCTGgCCggTggCCGT-CGGCCCGCCTGCC-3Ј) with the desired nucleotide changes indicated in lowercase letters was used in a PCR reaction together with downstream primer 5Ј-GACCGAAAGCAGCTACTCACTCCTGG-3Ј (antisense to nt ϩ42 to ϩ17) and Pfu polymerase to amplify a fragment of 120 bp (nt Ϫ78 to ϩ42). This was ligated directly in sense orientation into the SmaI site of pGL2-Basic to give BGN-78mut-Luc. For all plasmids the correct orientation of the inserts as well as the presence of the desired mutation was verified by sequencing (28). The PKI expression vectors for the heat-stable inhibitor of PKA, pRSV-PKI ver2, and an inactive mutant thereof, pRSV-PKI mut ver2, as well as the expression vectors for the PKA catalytic subunits, pRSV-C␤ and pRSV-C␤ mut , were kindly donated by Dr. R. A. Maurer (Oregon Health Sciences University, Portland, OR). The construction of the pRSV-PKI ver2 and pRSV-PKI mut ver2 vectors is described in Day et al. (29) and that of the expression vectors for the catalytic subunits of PKA in Maurer et al. (30) and Howard et al. (31).
Transient Transfection and Reporter Gene Assays-SK-UT-1 cells (2 ϫ 10 5 ) were seeded in 12-well plates, allowed to adhere overnight, and transfected at 80 -90% confluence. The transfection solution containing LipofectAMINE (Life Technologies, Inc.) and the plasmid DNA mixtures specified in the legends were removed from the cells after 2-3 h and replaced by normal medium which in some experiments contained 0.5 mM 8-Br-cAMP or 10 M forskolin. After an additional 18 h cells were lysed and luciferase activities determined luminometrically. To control for differences in transfection efficiencies among wells, the ␤-galactosidase-encoding plasmid pCH110 (Pharmacia, Freiburg, Germany) was routinely included in every transfection experiment. ␤-Galactosidase activities were determined with the Galacto-Light kit (Tropix/Perkin-Elmer, Weiterstadt, Germany) and used to normalize luciferase activities.
Electrophoretic Mobility Shift (EMSA) and Supershift Assays-Crude nuclear extracts from SK-UT-1 cells treated for 8 h with 8-Br-cAMP (0.5 mM), forskolin (50 M), or 2.5 h with PMA (0.1 M), respec-tively, were prepared according to a published method (32). Briefly, growth medium was removed by washing the cells thoroughly in phosphate-buffered saline. Scraped cells (1 ϫ 10 7 -1 ϫ 10 8 ) were transferred to a microcentrifuge tube, pelleted by centrifugation, and lysed in 500 l of cold lysis buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.2 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin) for 5 min on ice. Following centrifugation in a microcentrifuge (14,000 rpm, 30 s, 4°C) and removal of the supernatant, the pellet was resuspended in 50 -100 l of extract buffer (10 mM Hepes, pH 7.9, 0.4 M NaCl, 0.2 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin), rocked on ice for 15 min, and centrifuged at 14,000 rpm for 10 min at 4°C. Supernatants were aliquoted and stored at Ϫ70°C. To analyze nuclear protein binding to the BGN gene promoter, we employed the "Gel Shift Assay Core System" (Promega) according to the manufacturer's instructions. Prior to electrophoresis on 4% PAGE under non-denaturing conditions, 5-10 g of nuclear extract were incubated for 20 min with the 32 P-end-labeled DNA probe (see Fig. 6A for the DNA fragments used as probes and competitors). Competition studies used an excess of homologous or heterologous unlabeled sequences from the human BGN or an oligonucleotide containing the consensus sequence for Sp1 (5Ј-ATTC-GATCGGGGCGGGGCGAGC-3Ј) (Promega). Supershift reactions were identical to those described above, except that 1-2 l of appropriate antibody was added to the binding reaction and incubated for 1-2 h at 4°C prior to addition of the labeled probes and electrophoresis. Antibodies specific to Sp1 (PEP 2), Sp2 (K-20), Sp3 (D-20), Sp4 (V-20), and AP2 (C-18) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Western Blot Analysis-The preparation of Western blots and antigen detection was essentially performed as described (33). In brief, 10 -20 g of crude nuclear extracts from SK-UT-1 cells were loaded on a standard 10% polyacrylamide gel. Fractionated proteins were blotted onto polyvinylidene difluoride membrane (Immobilon-P, Millipore, Eschborn, Germany), and Sp1 and Sp3-related antigens were detected using anti-Sp1 (PEP-2) and anti-Sp3 (D-20) antisera and a chemiluminescent detection system (ECL kit, Amersham-Buchler). Specificity of the signals obtained was verified in control blots in which the antibodies had been preincubated for 1 h with a 10-fold molar excess of the respective blocking peptides (Santa Cruz).
Densitometric Analysis-Autoradiographs and ribosomal RNA bands in ethidium bromide-stained gels were scanned and analyzed with the program NIH Image 1.61. The exposure times were chosen as to stay within the linear range of the x-ray film used.

Expression of the BGN Gene in Normal Human Myometrium and Its Regulation by Forskolin in the Leiomyosarcoma Cell
Line SK-UT-1-To investigate if the BGN gene is expressed in human myometrium, RNA was extracted from non-pregnant human myometrium and hybridized with a BGN cRNA probe in a ribonuclease protection assay. Significant expression was noted in myometrial tissue from all three patients analyzed (Fig. 1A). BGN expression in the myometrium was confirmed by Northern blot hybridization of total RNA from monolayer cultures of mSMC ( Fig. 1B) and in poly(A) ϩ -enriched RNA from cultured tissue explants (Fig. 1C). Interestingly, BGN mRNA was more abundant in myometrial tissue from long term culture as compared with uncultured tissue (Fig. 1C); it had been shown previously that cultured mSMC cells spontaneously increase the formation of cAMP, eventually leading to changes in the expression of cAMP-responsive genes (23). We therefore sought to determine whether treatment of myometrial cells with agents known to increase intracellular cAMP, such as forskolin and stable cAMP analogs, would affect BGN gene expression. Since primary isolates of mSMC were not available for subsequent experimentation, we employed the leiomyosarcoma cell line SK-UT-1 which was previously shown to express both the endogenous BGN gene and transfected BGN promoter-reporter fusion genes (19). As depicted in Fig. 1D, SK-UT-1 cells responded to a 24-h treatment with forskolin (50 M) or 8-Br-cAMP (1 mM, data not shown) with a strong up-regulation of BGN mRNA.
The Transfected Human BGN Promoter Is Responsive to 8-Br-cAMP in SK-UT-1 Cells-To investigate whether the molecular basis for the up-regulation of BGN mRNA levels in SK-UT-1 cells is an enhanced transcription, we performed transient transfection assays with various BGN promoter-luciferase reporter constructs (Fig. 2). Transfection of the fulllength promoter construct BGNSac-Luc (Ϫ1218) into SK-UT-1 cells and treatment with 0.5 mM 8-Br-cAMP resulted in a strong induction of promoter activity. Upon removal of 5Ј- A, ribonuclease protection assay of total RNA extracted from the myometrium of three different patients (M1-M3). Ten g of RNA were hybridized to a BGN cRNA probe followed by RNase digestion and electrophoresis through 5% PAGE. Yeast RNA (Y) served as a negative control. P, undigested probe. B, Northern blot of total RNA (20 g) isolated from monolayer cultures of mSMC from two different patients (S1-S2). Hybridization was carried out using a 1144-bp BGN cDNA as the probe. The positions of the 28 S and 18 S ribosomal RNA bands are indicated on the right by an upper and a lower bar, respectively. The blot was exposed for 8 days. C, Northern blot of poly(A) ϩ -enriched RNA (10 g) from myometrial tissue explants that had been cultured for 3 weeks. The abundance of BGN mRNA in this culture (ϩ) was assessed relative to that in uncultured myometrium (Ϫ) from the same patient using the same probe as in B. The exposure time was 2 days. D, reverse transcriptase-PCR detection of BGN transcripts in SK-UT-1 cells. Duplicate wells of cells were left untreated (Ϫ) or treated (ϩ) for 24 h with 50 M forskolin (FSK). RNA was reverse-transcribed, and equal amounts of first strand cDNA subjected to simultaneous PCR amplification with BGN and GAPDH (GAP)-specific primers. BGN PCR products were visualized by hybridization with an internal cDNA probe. flanking sequences the cAMP inducibility increased and even the shortest construct with appreciable basal promoter activity (BGN-78-Luc) (19) maintained high responsiveness. Under the same transfection conditions the plasmid Ϫ36rPRL-Luc harboring the minimal rat prolactin promoter in the same plasmid backbone was only weakly inducible. Very similar results were obtained with 10 M forskolin (data not shown). These results show that 1) cAMP inducibility was conferred to the plasmids by the BGN promoter sequences and 2) the sequence(s) mediating the cAMP response of the human BGN gene reside in the proximal promoter region and/or the first 42 bp of exon 1.
To address the question whether regulation by cAMP is direct or indirect (e.g. that a nuclear protein is initially (de novo) synthesized which then directs efficient BGN gene expression), we followed the time course of BGN promoter activation. As shown in Fig. 2B, BGN promoter activation was rapid with a 3-fold induction over unstimulated controls after 3 h and was essentially complete by 6 h. Although being of lower magnitude, the overall activation kinetics of the BGN promoter were similar to those of the CRE-containing control construct known to be induced transcriptionally via phosphorylated cAMP response element-binding protein.
cAMP Responsiveness of the BGN Promoter in SK-UT-1 Cells Is Mediated by PKA-Cyclic AMP almost exclusively exerts its effects via activation of PKA and subsequent phosphorylation of nuclear proteins (34). To test the role of PKA in mediating the transcriptional activation of BGN by cAMP, the high affinity peptide inhibitor of the catalytic subunit(s) of PKA, PKI, was used to block the actions of PKA. Expression vectors encoding either PKI or a mutant, biologically inactive form thereof, PKI mut , were cotransfected with BGN-78-Luc into SK-UT-1 cells (Fig. 3A). Cotransfection of PKI mut had no effect on the stimulation of BGN promoter activity by 8-Br-cAMP (5.5fold stimulation). In contrast, transfection of PKI diminished the basal level of BGN-78-Luc promoter activity by 38% and inhibited 8-Br-cAMP-induction by 73.1%. To confirm that the effect of 8-Br-cAMP was mediated through activation of PKA, cotransfection experiments were carried out with an expression vector containing the coding sequence for the catalytic subunit of PKA, pRSV-C␤. In these experiments, expression of the PKA catalytic subunit alone (without the addition of exogenous cAMP) increased BGN promoter activity at least to a similar level as the 8-Br-cAMP stimulation and could not be increased further by addition of this cAMP analog (Fig. 3B).
Mapping and Functional Analysis of the cAMP-responsive Regions within the Proximal Promoter Region of the BGN Gene-Results from previous transfection experiments suggested the presence of (at least) one element between positions Ϫ78 and Ϫ46 of the BGN gene that dramatically increased basal transcription in a cell type-independent manner (19). As shown in Fig. 4, removal of this region also partially reduced cAMP responsiveness, since BGN-46-Luc was only inducible 3.7-fold (versus 5.5-fold in BGN-78-Luc). Further truncation (BGN-1-Luc) largely abolished promoter activity.
In an attempt to dissect further the relative importance of the regions between Ϫ78 and Ϫ46 and Ϫ46 and Ϫ1, respectively, for BGN transcription, we tested their ability to confer basal activation and cAMP responsiveness on a heterologous promoter. SK-UT-1 cells were cotransfected with pRSV-C␤ or pRSV-C␤ mut and fusion plasmids in which two overlapping fragments comprising nt Ϫ78 to Ϫ35 and Ϫ46 to Ϫ1 of the BGN promoter, respectively, were cloned upstream of the minimal rPRL promoter in the plasmid Ϫ36rPRL-Luc (Fig. 5A). As shown in Fig. 5B, the plasmid Ϫ78hBGN-36rPRL-Luc displayed a basal luciferase activity that was only 1.8-fold higher than that for the control vector Ϫ36rPRL-Luc, whereas cAMP inducibility was increased dramatically (31.6-fold over basal levels). When the Ϫ78 to Ϫ35 region was inserted in the reverse orientation (clone Ϫ78hBGNrev-36rPRL-Luc), cAMP responsiveness was still increased over the parent vector albeit to a lower extent (11.5-versus 5.5-fold over controls), thus indicating orientation independence. In contrast, the clone Ϫ46hBGN-36rPRL-Luc was only slightly more cAMP-inducible than the minimal rPRL promoter alone (6.3-versus 5.5-fold over controls). These experiments clearly show that the sequence from Ϫ78 to Ϫ35, independently from downstream sequences, can function as a cAMP-inducible module.
Specific Protein-DNA Complexes Are Formed between Positions Ϫ82 and Ϫ35 of the BGN Promoter-Deletion of the region from Ϫ78 to Ϫ46 of the BGN promoter resulted in a substantial loss of basal promoter activity and cAMP responsiveness. In turn, this sequence could increase cAMP responsiveness when cloned upstream of a heterologous promoter. Therefore, we sought to define the trans-acting factors that interact with this sequence. A PCR-generated fragment encompassing the region from Ϫ78 to Ϫ35 (Fig. 6A) was used as a probe in EMSAs with nuclear extracts from 8-Br-cAMP and forskolin-treated, PMA-treated, and untreated SK-UT-1 cells. As shown in Fig. 6B, multiple shifted bands were obtained, but there was no apparent effect of any of the treatments on the relative proportions of the factors bound to the Ϫ78/Ϫ35 probe, except for a band of fast electrophoretic mobility that likely represented a degradation product. We (19) and others (15) have previously identified a consensus binding site in the BGN promoter for the ubiquitous transcription factor Sp1 being present in reverse complement form at Ϫ78 (5Ј-CCGCCC-3Ј) and a sequence 5Ј-CCCTCCCC-3Ј at Ϫ59 that is identical to a previously described functional Sp1-binding site in the rat ornithine decarboxylase promoter (35). Of note, both sequences are conserved between the human (19) and the mouse (36,20) BGN gene (Fig. 6A). The presence of Sp1-like binding activity in SK-UT-1 nuclear extracts was confirmed using a specific Sp1 antibody in supershift reactions (Fig. 6B, lanes 3, 5, 7, and 9). Again, no major qualitative differences were seen among the various extracts in the supershift patterns.
To more precisely map the nuclear protein-binding sites, two overlapping ds oligonucleotides were designed (Ϫ82/Ϫ57 and Ϫ64/Ϫ39, Fig. 6A) and analyzed in EMSAs. As shown in Fig.  6C, lane 1, weak binding of nuclear proteins from 8-Br-cAMPtreated SK-UT-1 cells could be detected when the sequence extending from Ϫ82 to Ϫ57 was used as a probe. The DNAprotein complex observed most likely involved the Sp1 site at Ϫ78 as its formation was effectively competed for by a 50-fold molar excess of unlabeled Sp1 consensus oligonucleotide (Fig.  6C, lane 3). The presence of Sp1 or an Sp1-like protein in this complex was confirmed with an Sp1-specific antibody, the addition of which resulted in a supershift of this complex (Fig. 6C,  lane 2). Much stronger sequence-specific gel shifts, however, were noted with the Ϫ64/Ϫ39 probe (Fig. 6C, lane 4). Three complexes were visible of which complexes I and II were specific since their formation could be strongly decreased by the The results from the initial supershift experiments shown in Fig. 6B together with data from the competition experiments with oligonucleotides containing known Sp1-binding sites strongly suggest that complexes I and II contain Sp1-related proteins. In order to identify which Sp1 family members are involved, we performed supershift assays with antibodies to Sp1, Sp2, Sp3, and Sp4. Whereas antisera to Sp2 (Fig. 6C, lane  14), Sp4 (lane 16), and AP2 (lane 17) failed to shift the electrophoretic mobility of any band, formation of the specific complex of slowest mobility (I) was completely supershifted by antiserum to Sp1 (lane 13). Likewise, incubation with an antiserum to Sp3 resulted in the disappearance of complex II and the occurrence of a weak supershift (lane 15).
Another set of EMSA competition experiments was performed to identify the coordinates and the critical nucleotides of the putative Sp1/Sp3-binding site (Fig. 6D). Changing the C residues at Ϫ58/Ϫ57 and/or Ϫ55/Ϫ54 to a purine abolished the ability of these oligonucleotides to compete for binding with the Ϫ64/Ϫ39 probe in complexes I and II (lanes 3, 4, 6, and 7). However, an oligonucleotide containing point mutations of nts  nuclear extracts (nucl. extr., 8 g) from SK-UT-1 cells (untreated control, or treated for 2.5 h with PMA, or 6.5 h with 8-Br-cAMP or forskolin) were incubated with the labeled probe for 20 min at room temperature followed by 4% PAGE Ϫ62 through Ϫ60 partially retained its ability to compete (lane 2) indicating that these nts are less critical for the interaction of these factors with the BGN promoter thereby mapping the 5Ј end of the binding site to Ϫ59. Its 3Ј end was localized to somewhere between nt Ϫ52 and Ϫ49 as the Ϫ74/Ϫ49 oligonucleotide (lane 5) unlike the Ϫ64/Ϫ52 oligonucleotide (Fig. 6C, lane 10) competed with the Ϫ64/Ϫ39 probe for binding. As suggested from the corresponding competition experiment (compare Fig. 6C, lane 8) complexes I and II (but not the nonspecific complex) were also formed with a shorter version of the probe sequence (Ϫ64/Ϫ47, Fig. 6D, lane 12) but were absent from mutant versions of this oligonucleotide containing C to G (mut1) or C to A (mut2) transversions in positions Ϫ58/Ϫ57 and Ϫ55/Ϫ54 (lanes 8 -11).
A Mutation in the Pyrimidine-rich Sequence between Ϫ59 and Ϫ49 in the BGN Promoter Reduces Both Basal and cAMPinduced Promoter Activity-The mutant oligonucleotide Ϫ64/ Ϫ47mut1 which was shown above to be unable to bind nuclear proteins from SK-UT-1 cells was used to replace the corresponding wt sequence in the construct BGN-78-Luc. BGN-78mut-Luc was then functionally compared with the wt construct for its ability to confer basal and cAMP-dependent transcriptional induction in the natural context of the BGN promoter (Fig. 7). Basal promoter activity of BGN-78mut-Luc was only 31% that of the wt construct. Moreover, cAMP-dependent transcription was reduced to 20.8%. These data further attest to the dual role of the CT-rich sequence between nt Ϫ59 and Ϫ49 in the promoter in modulating BGN gene expression.
Further Characterization of the Nuclear Protein from Complex I-The competition and supershift data presented in Fig.  6, B-D, suggested that complex I contained an Sp1-like protein. We then addressed the question whether this protein was indeed Sp1 by investigating if recombinant Sp1 was capable of binding in vitro to the Ϫ64 to Ϫ39 region in the BGN promoter (Fig. 8). Surprisingly, no binding could be detected (lanes 1 and  2). This was not due to degradation of the Sp1 protein since it was able, like the endogenous Sp1 from SK-UT-1 cells, to produce a strong gel shift with the Sp1 consensus oligonucleotide (compare lanes 6 and 8 -10) which could be supershifted using the same Sp1 antibody as above (lanes 5 and 9, respectively). The presence in SK-UT-1 cells of Sp1 as well as proteins antigenically related to Sp1 was further suggested by results from immunoblot analysis (Fig. 9). The Sp1 antibody specifically recognized several proteins in the size range of 60 -120 kDa in addition to the p106/p95 doublet characteristic of Sp1 in many cell types (Fig. 9, lanes 1-2). In contrast, the Sp3 antibody detected two doublets of Ϸ120 and Ϸ80 kDa (Fig. 9, lanes 4 -5) which have been reported to represent the full-length Sp3 protein and Sp3 isoforms that arise from internal translational initiation, respectively (37). Immunoreactivity of the Sp1 and Sp3 antibodies was fully abolished by preincubation with the respective blocking peptides (lanes 3 and 6, respectively). No qualitative or quantitative differences in Sp1(-like) and Sp3 protein expression were evident between 8-Br-cAMP-treated and control cells (compare lanes 1 and 2 for Sp1 and lanes 4 and 5 for Sp3, respectively). Taken together, we conclude that Sp3 and an Sp1-like protein other than Sp1 bind to the sequence 5Ј-CCCTCCCCGTC-3Ј in the proximal BGN promoter and that this interaction can enhance both basal transcriptional activity and cAMP responsiveness.

DISCUSSION
In this study we have analyzed the molecular components involved in basal and cAMP-dependent transcription of the BGN gene in the human leiomyosarcoma cell line SK-UT-1. The increase in BGN mRNA levels observed upon treatment of these cells with 8-Br-cAMP or forskolin was primarily transcriptional as judged from up-regulation of transiently transfected BGN promoter-luciferase reporter constructs and was mediated by PKA. Partial mapping of the cAMP-responsive regions in a homologous and a heterologous promoter context in combination with DNA-nuclear protein binding, competition, and supershift studies led to the identification of a short pyrimidine-rich sequence in the proximal promoter (nts Ϫ59 to Ϫ49) that served as a binding site for Sp1-like and Sp3 transcription factors. A mutation of this element that disrupted binding of the Sp1-like/Sp3 proteins dramatically decreased both basal and cAMP-stimulated transcription from transfected human BGN promoter-reporter constructs.
Transcriptional regulation by cAMP typically involves nuclear translocation of the free catalytic subunits of PKA which then phosphorylate and thereby increase the transactivation potential of CRE-binding proteins (38). However, since no CRE- under non-denaturing conditions (lanes 2, 4, 6, and 8). In a parallel set, supershift reactions were performed with an Sp1-specific antibody that had been incubated with the nuclear extracts for 2 h prior to addition of the probe (lanes 3, 5, 7, and 9). SS, supershift. Note 1) the qualitatively similar pattern in protein binding to the probe between control, PMA, 8-Br-cAMP, and forskolin-treated SK-UT-1 nuclear extracts, respectively, and 2) that multiple bands were shifted by the Sp1 antibody. The lowest band represents a nonspecific complex. C, analysis of nuclear protein binding to the Ϫ82 to Ϫ39 region of the human BGN promoter by EMSA competition and antibody supershift assays. SK-UT-1 nuclear extracts from 8-Br-cAMP-treated cells were mixed with the labeled probes as described under B. Competitions were performed by mixing the indicated molar excess of unlabeled oligonucleotides with the nuclear proteins and incubating for 10 min prior to addition of the labeled probe. The labels I, II, and NS (nonspecific binding) refer to lanes 4-17 only. Note the absence of complex II and the appearance of a weak supershift in lane 15. To better visualize this supershift the autoradiograph has been slightly overexposed. SS, supershift; cs, consensus sequence. D, EMSA with nuclear proteins (lanes 1-11: 8 g, lanes 12-13: 4 g)  like sequences are present in the proximal BGN promoter, a participation of this class of transcription factors seems unlikely. Nevertheless, the rapid rise in cAMP-induced promoter activity and the fact that we could not observe any qualitative differences in specific protein binding to the Ϫ78 to Ϫ35 region between 8-Br-cAMP/forskolin-treated and untreated cells favors a direct mode of regulation (see below).
Interestingly, we found that PKI decreased BGN transcriptional activity in the absence of activators of PKA suggesting that SK-UT-1 cells contain a significant level of free PKA catalytic subunits that contribute to a basal level of transcription. The close functional connection between basal and cAMPmediated BGN transcription in SK-UT-1 cells is also reflected at the sequence level; 5Ј-deletion analysis and transfection studies with wt and mutant sequences showed that all or part of the sequence centered around Ϫ56 in the BGN promoter accounted not only for a significant portion of basal activity but also for cAMP responsiveness. Of note, this sequence retained some of these functions when present in the reverse orientation and in a heterologous promoter context and may thus constitute an enhancer. An examination of the promoter sequence revealed an Sp1-like site (CCCTCCCC) at Ϫ59/Ϫ52, and subsequent supershift and competition analysis demonstrated binding of both an Sp1-like and an Sp3-like protein at this element. More rigorous mutagenesis then showed that additional nucleotides at the 3Ј end were required for effective binding of Sp1-like/Sp3 proteins and that the C residues at positions Ϫ58/Ϫ57 and Ϫ55/Ϫ54 were critical for this interaction. Identical or very similar pyrimidine-rich sequences (also referred to as CT elements) have been found to bind Sp1related proteins in the promoters of several other genes, e.g. the rat ornithine decarboxylase gene (35), the c-myc gene (39), the hepatocyte growth factor gene (40), and the human ␣2(I) collagen gene (41).
Interestingly, the protein from complex I that reacted with the Sp1 antibody was likely not to be Sp1 as recombinant Sp1 failed to bind to the CT element in the proximal BGN promoter. Since this protein was not recognized by antibodies to Sp2, Sp3, and Sp4 in supershift experiments, it could represent a novel member of this class of transcriptional regulators. This was given support by results from immunoblots with SK-UT-1 nuclear proteins showing that the same Sp1 antibody detected several specific bands of which only two (106 and 95 kDa) are known to represent Sp1 itself. Studies are now underway to purify, characterize, and clone this Sp1-related factor. The protein from complex II was identified as Sp3 (42,43) by antibody supershift and subsequent Western blot analysis. This was not surprising as Sp1 and Sp3 bind to the same or very similar DNA-binding sites in gene promoters (44) eventually resulting in a gel shift pattern of striking similarity to that in the BGN promoter (45). Consistent with this notion, in vitro binding studies with various mutant versions of the Ϫ59/Ϫ49 sequence resulted in a simultaneous disappearance of both proteins suggesting a competition between them for binding to the CT element.
The precise functional role of each of these factors upon interaction with the proximal BGN promoter remains speculative at this stage. Whereas Sp1 in most cellular systems stimulates transcription, Sp3 has been shown to be a bifunctional regulator that can act either as a transcriptional repressor or activator of Sp1-mediated transactivation depending on cell type and promoter context (reviewed in Ref. 44). Repression may occur in an active fashion due to the presence of a putative repressor domain in the Sp3 protein (44) or may result from competition with Sp1 in conjunction with a lack of transactivation potential in Sp3. Recently, Kennett and co-workers (37) The membranes were incubated with primary antibody against Sp1 (anti-Sp1) or Sp3 (anti-Sp3) and subjected to ECL detection. Specificity controls were performed by incubating the primary antibodies with an excess of the respective blocking peptide prior to antigen detection. Molecular size markers are indicated on the right. The large arrowheads in the anti-Sp1-treated blot point to specific bands that likely represent Sp1-related nuclear proteins.
found that the 78 -80-kDa isoforms of Sp3 that arise via internal translational initiation function as potent inhibitors of Sp1/ Sp3-mediated transcription. Strikingly, we detected this doublet in immunoblots of SK-UT-1 nuclear proteins at an equal intensity as the full-length protein. However, it remains to be tested which of these different Sp3 isoforms bind to the CT element in the BGN promoter in vivo.
The mutation of the Sp1/Sp3-binding site in BGN-78-Luc reduced cAMP inducibility to the level seen with BGN-46-Luc, whereas basal promoter activity, although strongly diminished, was still significantly higher than in BGN-46-Luc. This indicates that additional cis-acting sequences within the Ϫ78 to Ϫ46 region contribute to basal transcription. A likely candidate is the Sp1 consensus sequence at Ϫ78 which was shown to bind an Sp1(-like) factor, most likely Sp1 itself. In fibroblasts, Sp1 binding has recently also been detected at an Sp1 consensus site located further upstream (Ϫ216) (20).
Members of the Sp1 family of transcription factors (others than Sp1) play a role in the transcriptional regulation of a large and diverse array of genes, among which are other extracellular matrix genes. In the mouse syndecan-1 gene whose promoter shares several features with the BGN gene (TATA-less, GC-rich 5Ј-flanking region with several putative Sp1 sites), Sp1-like factors were shown to be responsible for the constitutive expression in epithelial cells (46). In the human ␣2(I) collagen gene an Sp1/Sp3-binding site (TCCTCC) mediates both basal transcriptional activity and oncostatin M stimulation (41). Whereas the involvement of Sp1-like proteins in conferring high levels of constitutive and cytokine/hormonestimulated transcriptional activity to many genes is well established, their participation in cAMP-regulation is less well documented and has so far only been clearly demonstrated for the bovine CYP11A gene (22) and the rhesus monkey growth hormone-variant gene (45). If one postulates a direct mode of regulation for the cAMP-PKA signal transduction pathway in inducing BGN gene expression (see above), then the question arises how it is mechanistically linked to the Sp1-like/Sp3 proteins. Rather than recruitment of new proteins to DNAbinding sites, the effects of cAMP may involve modulation of protein-protein interactions, transcriptional complex assembly, or post-translational modifications of a trans-acting protein.
Post-translational modifications by the protein kinase A pathway have been described (47)(48)(49)(50)(51)(52). Specifically, Sp1 can be phosphorylated (53), but it does not have a PKA phosphorylation site. However, this may not be true for the Sp1-like protein binding to the BGN promoter, and it is thus conceivable that its transactivation function is increased upon phosphorylation by PKA. Alternatively, the Sp1-like/Sp3 factors bound to the BGN promoter in SK-UT-1 cells may interact with a cell-specific coactivator that itself is not a DNA-binding protein but can serve as a target for PKA and subsequently activate the Sp1like/Sp3 proteins.
Preliminary data from our laboratory suggest that cAMP regulation of BGN gene expression is also operating in mSMC, the predominant cell type in myometrial tissue. What could be the physiological significance of cAMP-mediated up-regulation of BGN expression in myometrial cells? Adenylate cyclase activity has been reported to increase in the human myometrium during pregnancy (54), reflecting stimulation by hormones like gonadotropins, PGE 2 , relaxin, and/or corticotropin-releasing factor that act through elevation of intracellular cAMP. This could result in chronic activation of BGN synthesis and may indicate a greater need for BGN for uterine growth during pregnancy and/or the preparation process of the uterus for parturition. A relevant reproduction-associated participation is well documented in the uterine cervix, a tissue that undergoes a profound connective tissue turnover in preparation for parturition. Interestingly, PGE 2 has been reported to increase BGN expression markedly in final cervical ripening during labor (55,56).
We have demonstrated previously that the human BGN gene is transcriptionally induced by cAMP in bone cells (24), and as shown in this study, this also applies to cells of myometrial origin. This would suggest that the cAMP/PKA signal transduction pathway is of more general importance for BGN regulation. This assumption may facilitate the identification of physiological modulators of BGN gene expression and may ultimately help to unravel the biological functions of this PG.