Cloning the Promoter for Transforming Growth Factor-β Type III Receptor

Transforming growth factor-β binds to three high affinity cell surface molecules that directly or indirectly regulate its biological effects. The type III receptor (TRIII) is a proteoglycan that lacks significant intracellular signaling or enzymatic motifs but may facilitate transforming growth factor-β binding to other receptors, stabilize multimeric receptor complexes, or segregate growth factor from activating receptors. Because various agents or events that regulate osteoblast function rapidly modulate TRIII expression, we cloned the 5′ region of the rat TRIII gene to assess possible control elements. DNA fragments from this region directed high reporter gene expression in osteoblasts. Sequencing showed no consensus TATA or CCAAT boxes, whereas several nuclear factors binding sequences within the 3′ region of the promoter co-mapped with multiple transcription initiation sites, DNase I footprints, gel mobility shift analysis, or loss of activity by deletion or mutation. An upstream enhancer was evident 5′ proximal to nucleotide −979, and a silencer region occurred between nucleotides −2014 and −2194. Glucocorticoid sensitivity mapped between nucleotides −687 and −253, whereas bone morphogenetic protein 2 sensitivity co-mapped within the silencer region. Thus, the TRIII promoter contains cooperative basal elements and dispersed growth factor- and hormone-sensitive regulatory regions that can control TRIII expression by osteoblasts.

Transforming growth factor-␤ binds to three high affinity cell surface molecules that directly or indirectly regulate its biological effects. The type III receptor (TRIII) is a proteoglycan that lacks significant intracellular signaling or enzymatic motifs but may facilitate transforming growth factor-␤ binding to other receptors, stabilize multimeric receptor complexes, or segregate growth factor from activating receptors. Because various agents or events that regulate osteoblast function rapidly modulate TRIII expression, we cloned the 5 region of the rat TRIII gene to assess possible control elements. DNA fragments from this region directed high reporter gene expression in osteoblasts. Sequencing showed no consensus TATA or CCAAT boxes, whereas several nuclear factors binding sequences within the 3 region of the promoter co-mapped with multiple transcription initiation sites, DNase I footprints, gel mobility shift analysis, or loss of activity by deletion or mutation. An upstream enhancer was evident 5 proximal to nucleotide ؊979, and a silencer region occurred between nucleotides ؊2014 and ؊2194. Glucocorticoid sensitivity mapped between nucleotides ؊687 and ؊253, whereas bone morphogenetic protein 2 sensitivity comapped within the silencer region. Thus, the TRIII promoter contains cooperative basal elements and dispersed growth factor-and hormone-sensitive regulatory regions that can control TRIII expression by osteoblasts.
Several cell surface receptors for transforming growth factor-␤ (TGF-␤) 1 are now known. Type I and type II receptors (TRI and TRII) have intracellular kinase domains responsible for heterologous receptor activation or downstream signal transduction (1)(2)(3)(4). The type III receptor (TRIII), a membraneanchored proteoglycan also termed betaglycan, is thought to have a biological function distinct from TRI and TRII (5)(6)(7). The rat TRIII gene encodes a 91.6-kDa protein core that is modified by approximately 10 kDa of N-linked glycosyl residues and 150 -200 kDa of heparan and chondroitin sulfate side chains.
TRIII has a relatively short, 43-amino acid cytoplasmic domain that lacks commonly recognized protein docking or kinase like motifs but is enriched with serines and threonines to approximately 42% (5,6).
TRIII is prevalent on many fetal cells, where it can be the most abundant TGF-␤-binding site. All TGF-␤ isoforms bind to TRIII with comparably high affinity, although this is about 3-5-fold less than that for TRI and TRII (7). Certain cells lack TRIII but maintain TGF-␤ sensitivity. Nonetheless, TRIII may attract and enhance TGF-␤ binding to TRII and form a more stable ligand-receptor complex (6). On certain cell types this appears more pronounced or limited to the TGF-␤2 isoform (8,9). Moreover, disproportionately high levels of TRIII may sequester and possibly limit its binding to signaling receptor complexes (10 -12).
The relative amount of TRIII is thought to vary with development, with differentiation, or in a tissue-specific manner. For example, TRIII levels are prevalent on less differentiated bone, endothelial, adrenocortical, prostate, and muscle cells but are rapidly regulated by agents or events that control cell differentiation. On bone cells, TRIII levels decrease when differentiation is enhanced by bone morphogenetic protein (BMP)-2 but rise in response to glucocorticoid or agents that increase intracellular cAMP (10,12,13). TRIII levels also decrease with endothelial cell differentiation in three-dimensional culture and during the transition from myoblasts to myotubes. In the ventral prostate, TRIII levels rise after castration and are resuppressed by androgen administration (14 -16). Analogous to the effect of parathyroid hormone (PTH) on osteoblasts (10), corticotropin produces a cAMP-dependent increase in TRIII on adrenal cells (17).
Based on these findings, we predicted that complex changes in TRIII are regulated in part by constitutive, developmental, and hormone-dependent genomic elements and in this way control how the effects of TGF-␤ are perceived within various tissues. To define how these events might occur at the molecular level, we cloned the promoter for rat TRIII. Because we previously defined situations and agents that regulate TRIII levels on bone cells, we have also begun to assess regions within the promoter that may account for constitutive and hormone-dependent changes in its expression by osteoblasts.
DNA Sequencing-Phage inserts were cloned into pBluescript-KSII vector (Stratagene, La Jolla, CA) and mapped with an assortment of restriction endonucleases, and subclones were produced by restriction site cleavage (see Fig. 1 and Table I). Double-stranded plasmid DNA was denatured with 0.2 N sodium hydroxide and sequenced by the dideoxy chain termination method (19) with a T7 Sequenase sequencing kit (U. S. Biochemical Corp.) and specific synthesized oligonucleotide primers. The sequence was analyzed by data bank searches to assess restriction enzyme cleavage sites and nuclear factor-binding elements.
Construction of TR3 Promoter/Reporter Plasmids-Fragments from the 5-kb rat TRIII genomic DNA clone, flanked by XhoI and NotI restriction cleavage sites (pTR3Bs/5XN), were subcloned into the promoterless reporter vector pGL2-Basic (Promega Corp.) at convenient restriction sites or by polymerase chain reaction (PCR). To produce pTR3/3.7, a 3.7-kb StuI/SacI fragment was generated by cleavage with StuI and partial digestion with SacI and inserted at SmaI and SacI sites by blunt and cohesive end ligations. pTR3/1.9 (forward orientation) and pTR3/1.9R (backward orientation) were produced by inserting the 1.9-kb fragment generated by SacI digestion into the SacI site, and orientation was determined by restriction fragment analysis. pTR3/1.8 was produced by digestion with MscI and SacI and inserted at SmaI and SacI sites by blunt and cohesive end ligations. pTR3/1.9␦ was produced by deleting the 1.1-kb internal fragment flanked by NcoI sites from pTR3/1.9. pTR3/0.4K was produced by inserting the 0.4-kb KpnI to SacI fragment into homologous vector sites. pTR3/0.4S, pTR3/0.2B, pTR3/0.2E, and pTR3/0.1A were produced by deleting the KpnI/SmaI, the KpnI/BssHII, the KpnI/EagI or the KpnI/ApaI fragments from pTR3/0.4K, respectively. pTR3/0.3P was produced by PCR with GLprimer 2, complementary to DNA sequences in pGL2-Basic, and primer TR3F1 (5Ј-CGGGGTACCAGGAGGAGAGGAGGGGCAGGAGGAGGA-GTTTC-3Ј), defined by nucleotides Ϫ534 to Ϫ503 from the rat TRIII promoter and modified at the 5Ј end to include a KpnI cleavage site (underlined) to facilitate cloning the PCR product into pGL2-Basic. pTR3/0.3P, introducing a mutation within a putative GC box, was produced by substituting two nucleotides (bold) in forward primer TR3F1M (5Ј-CGGGGTACCAGGAGGAGAGGTTGGGCAGGAGGAGG-AGTTTC-3Ј). pTR3/0.2B, introducing a mutation within a putative Sp1-binding sequence, was produced by substituting three nucleotides (bold) in forward primer TR3F2M (5Ј-CGGGGTACCGCGCGCCCGAC-CCTTTCCGCGCGTGT-3Ј), defined by nucleotides Ϫ441 to Ϫ416 from the rat TRIII promoter, and retaining the BssHII restriction cleavage site at the 5Ј end (underlined) for subsequent plasmid cloning. Fragments were verified by sequence analysis, and plasmids were purified with commercial purification kits (Qiagen Corp.). Construct names and nucleotide numbering are collected in Table I.
Cell Cultures-Osteoblast-enriched cell cultures were prepared from parietal bones of 22-day-old Harlan Sprague-Dawley rat fetuses (Charles River Breeding Laboratories, Raleigh, NC) by methods approved by Yale Animal Care and Use Committee. Sutures were eliminated by dissection, and cells were released from parietal bones by five sequential collagenase digestion intervals, as described previously (10,12,13). Cells released during the last three digestions exhibit biochemical characteristics associated with differentiated osteoblasts, including high levels of PTH receptors and type I collagen synthesis, and a rise in osteocalcin expression in response to 1,25(OH) 2 D 3 (10,20). Histochemically, approximately 80% of the cells express alkaline phosphatase, 2 although this is not entirely specific for osteoblasts. By these combined criteria, differential sensitivity to TGF-␤, BMP-2, various prostaglandins, and the ability to express nuclear factor CBFa1 and to form mineralized nodules in vitro (12,(21)(22)(23)(24), osteoblast-enriched cultures are well distinguished from less differentiated periosteal cells from the same tissue source. Cells were plated at 4,000 cells/cm 2 in Dulbecco's modified Eagle's medium with 20 mM HEPES (pH 7.2), 100 g/ml ascorbic acid, penicillin, and streptomycin (Life Technologies, Inc.), and 10% fetal bovine serum. Fetal rat fibroblasts were isolated from skin flaps obtained during parietal bone dissection. The flaps were minced with sterile scissors, and the cells were released by 20 min of digestion with collagenase plated and in the same medium described above (26). To test effects of glucorticoid or BMP-2 on TRIII mRNA, TRIII gene promoter activity, or radiolabeled TGF-␤ binding, cells were serumdeprived and treated as indicated in the figure legends.
Transfections-Promoter/reporter plasmids were co-transfected with a vector carrying the ␤galactosidase gene under control of the SV40 promoter using LipofectAMINE (Life Technologies, Inc.). Briefly, cultures at 50 -75% confluent density were rinsed and exposed to plasmids in serum free medium, and the solutions were then replaced with medium supplemented with 5% fetal bovine serum. Cultures were expanded for 48 h, rinsed, and treated as indicated in the figure legends. After treatment, cultures were rinsed with phosphate-buffered saline and lysed in 100 l of a solution containing 25 mM Tris-phosphate (pH 7.8), 2 mM dithiothreitol, 2 mM EDTA, 10% glycerol, and 1% Triton X-100. Lysates were collected, nuclei were cleared by centrifugation at 12,000 ϫ g for 5 min, and supernatants were analyzed for reporter gene activity and corrected for protein content (24).
RNA Preparation and Northern and Ribonuclease Protection Assay-Total RNA was extracted from primary osteoblast-enriched cell cultures with acid guanidine-monothiocyanate, precipitated with isopropyl alcohol, and dissolved for assay. To assess TRIII mRNA, total RNA was fractionated on a 1.5% agarose and 2.2 M formaldehyde gel, blotted onto charged nylon and hybridized with a 32 P-labeled cDNA restriction fragment of plasmid bg7 encoding rat TRIII (6). To assess start sites of TRIII transcription, two antisense cRNA probes were synthesized. For probe 1, pTR3Bs/5XN was linearized with KpnI and transcribed with T3 RNA polymerase, generating a fragment corresponding to nucleotides Ϫ692 to Ϫ179. For probe 2, the TRIII genomic fragment corresponding to nucleotides Ϫ692 to Ϫ273 was cloned into pBluescript KS II by PCR. The plasmid was linearized with HindIII and transcribed with T7 RNA polymerase. Probes were labeled with [␣-32 P]UTP using the Maxiscript kit (Ambion Corp.). 10 g of total cell RNA and 1 ϫ 10 5 cpm of probe cRNA were combined in 30 l of hybridization buffer (80% formamide, 1 mM EDTA, 100 mM sodium citrate, 300 mM sodium acetate, pH 6.4) for 16 h at 45°C. The samples were then digested at 37°C for 30 min by adding 300 l of a solution containing 5 mM EDTA, 300 mM NaCl, 10 mM Tris-Cl (pH 7.5), 1 unit/ml RNase A, and 40 units/ml RNase T1. RNase was inactivated with 17 l of 10% SDS and 3 l of proteinase K at 20 mg/ml. Protected transcript fragments were precipitated with isopropanol and resolved on a denaturing 6% polyacrylamide gel alongside sequencing ladders. Bound or protected RNA probes were visualized by autoradiography. rRNA standards for Northern analysis were stained with ethidium (12,25).
DNase I Footprinting-The TRIII promoter DNA fragment corresponding to nucleotides Ϫ641 to Ϫ350 was generated by PCR from reporter construct pTR3/0.4S using primers GLprimer 1 complementary to sequences in pGL2-Basic, and primer TR3R2, defined by nucleotides Ϫ367 to Ϫ350 from the rat TRIII promoter (5Ј-CCGAGCTCG-GCGTCCCCGAAAGCCTGGATCA-3Ј) that was modifed at the 5Ј end to include a SacI restriction site (underlined). The fragment was cloned directionally into pGL2-Basic at SmaI and SacI sites. Plasmid was linearized by digestion with NheI and labeled with [␣-32 P]dCTP using the Klenow fragment of DNA polymerase I. Probe was released by digestion with SmaI, purified, and 2-10 ϫ 10 4 cpm were incubated with 50 g of bovine serum albumin or 40 -160 g of nuclear extract in gel mobility shift assay buffer (see below). After 15 min at 25°C, samples were supplemented with 5 mM MgCl 2 , 2.5 mM CaCl 2 and 20 -100 ng of DNase I and incubated for 2 min, and the reaction was stopped with an equal volume of 20 mM EDTA, 1% SDS, 0.2 M NaCl. After phenol/ chloroform extraction and ethanol precipitation, samples were fractionated by electrophoresis through a 7% denaturing polyacrylamide gel in 1ϫ Tris borate-EDTA. DNA fragments were visualized by autoradiography (27).
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assay experiments followed previously published methods (24,25). Briefly, commercially produced double-stranded probes (see Table II) were radiolabeled by annealing complementary oligonucleotides, followed by fill in of single-stranded overhangs with dCTP, dGTP, dTTP, 2 M. Centrella and T. L. McCarthy, unpublished results. and [␣-32 P]dATP with the Klenow fragment of DNA polymerase I. 5-10 g of nuclear extract protein was preincubated for 20 min on ice with 2 g of poly(dI⅐dC), without or with unlabeled specific or nonspecific competitor DNA, in 60 mM KCl, 25 mM HEPES (pH 7.6), 7.5% glycerol, 0.1 mM EDTA, 5 mM dithiothreitol, and 0.025% bovine serum albumin. After adding 0.1-0.2 ng of DNA probe (5 ϫ 10 4 cpm) for 30 min on ice, samples were fractionated by electrophoresis on a 5% nondenaturing polyacrylamide gel that was prerun for 30 min at 12.5 V/cm at 25°C in 45 mM Tris, 45 mM boric acid, 1 mM EDTA. To assess nuclear factors by antibody reactivity, nuclear extract was incubated with 0.2-1.0 l of antisera for 30 min at 4°C before adding 32 P-labeled probe. Electrophoresis was performed for 2.5 h under identical conditions. Radioactive DNA bound protein complexes were visualized by autoradiography.
Radioligand Binding-TGF-␤1 was radioiodinateded with chloramine T to specific activity of 4500 Ci/mmol and isolated by gel filtration in 0.1 N acetic acid with 4 mg/ml of bovine serum albumin. Cells were rinsed and incubated at 4°C with 200 pM of 125 I-labeled TGF-␤ diluted in cold serum-free medium supplemented with 4 mg/ml of bovine serum albumin. After 3 h, cultures were rinsed with chilled phosphate-buffered saline, cross-linked with 0.2 mM disuccinimidyl suberate (Pierce), and extracted, and equal amounts of cell protein were fractionated by polyacrylamide gel electrophoresis and examined by autoradiography and densitometry, as described previously (12).
Protein Synthesis-Cells were transfected to overexpress native rat TRIII using DNA subcloned from plasmid pBG7 (a gift from Dr. J. Massague, Memorial Sloan-Kettering Cancer Center, New York; Ref. 6) into plasmid pSV7d (9,28). Control cells were transfected with an equal amount of empty pSV7d expression vector. After 48 h of plasmid expression, cells were serum-deprived, treated for 24 h with 120 pM of TGF-␤1, and labeled with 5 Ci/ml [ 3 H-2,3]proline (2.5 Ci/mmol) for the last 2 h of culture. Cells were lysed by freeze-thaw and extracted in 0.5% Triton X-100, and samples were collect by precipitation in 10% trichloroacetic acid. Precipitates were acetone extracted, dissolved in 0.5 N acetic acid, and neutralized with NaOH. [ 3 H]Proline incorporation into collagen and noncollagen protein was measured by differential digestion with bacterial collagenase free of nonspecific protease activity, as described previously (12).
Reagents-Transfection vectors pGL2-Basic and pGL2-Control were obtained from Promega Corp. (Madison, WI). Hydrocortisone (cortisol) was obtained from Sigma. Recombinant human BMP-2 was generously provided by Genetics Institute, Inc. (Cambridge, MA). Antisera to Sp1 and Sp3 were obtained from SantaCruz Biotechnologies (Santa Cruz, CA). Antisera to AP-2␣, ␤, and ␥ and a recombinant expression construct encoding AP-2 were generously provided by Dr. Trevor Williams (Yale University). Antiserum to AP-4 was generously provided by Dr. Richard Gaynor (Southwestern Medical Center, Dallas, TX).
Statistical Analysis-Statistical differences were assessed by oneway analysis of variance and the Kruskal-Wallis or Bonferonni methods for post hoc analysis, with SigmaStat software (Jandel Scientific, San Rafael, CA).

Isolation of Rat TRIII Receptor Genomic
Clones-Using a 0.4-kb rat TRIII probe containing 334 bp of 5Ј-untranslated region and 61 bp of coding sequence, four positive clones were selected out of approximately 6 ϫ 10 5 recombinant phages from a rat genomic library. Of these, clone 22 contained a 15-kb insert that produced only two fragments of 5 and 10 kb by digestion with NotI. Both fragments were subcloned into pBluescript and analyzed by restriction enzyme cleavage and sequencing. The 5-kb DNA fragment, which we designate as pTR3Bs/5XN, was located directly 5Ј to coding sequence of the rat TRIII gene.
Sequence Analysis-All 5136 nucleotides of pTR3Bs/5XN were sequenced. The DNA sequence corresponding to nucleotides Ϫ2130 to Ϫ179 is shown in Fig. 1. Nucleotides Ϫ334 to Ϫ179 correspond to the 5Ј-untranslated region of a previously reported rat cDNA clone, which was originally numbered with reference to the first nucleotide of the initial ATG codon (5). The sequence was analyzed by the Wisconsin University Genetics Computer Group program and MatInspector data bank searches (25,29) to assess restriction enzyme cleavage sites and nuclear factor-binding elements. Analogous to the TRI and TRII gene promoters (25,30), this sequence lacks a TATA box.
Two nuclear factor Sp1-binding sequences (31), at nucleotides Ϫ31 to Ϫ424 and nucleotides Ϫ527 to Ϫ518 (a GC box), are located within 0.7 kb upstream of the initial ATG codon. Another GC box is located further upstream (Ϫ1880 to Ϫ1871). The 0.5-kb region between nucleotides Ϫ179 to Ϫ692 is highly GC enriched to 69%, consistent with CpG islands often associated with the initiation of gene transcription (32,33). Three sequences consistent with binding sites for nuclear factor CCAAT/enhancer-binding protein (C/EBP; 34) occur at nucleotides Ϫ2073 to Ϫ2065, Ϫ1523 to Ϫ1515, and Ϫ574 to Ϫ565. In addition, the data bank searches suggested a variety of other elements that might be responsible for conditional, hormonedependent, or tissue-specific TRIII gene expression. These sequences can be located and evaluated through GenBank TM accession number AF117811.
Transcription Initiation Sites-To identify the start site of TRIII transcription, total RNA from fetal rat osteoblasts was probed with two cRNA probes corresponding to nucleotides Ϫ692 to Ϫ179 and Ϫ692 to Ϫ273 of the rat TRIII gene and analyzed by ribonuclease protection assay. Transcription start sites were determined by aligning protected RNA bands with a DNA sequencing ladder. As shown in Fig. 2, several protected fragments were obtained with each cRNA probe, indicating multiple transcription initiation sites. Similar start sites were obtained with each of the two probes. By this analysis, all major FIG. 1. Nucleotide sequence of the 5 region of the rat TRIII gene. Fragments of rat TRIII genomic clones were subcloned into plasmids at convenient restriction sites, and overlapping fragments were sequenced and aligned. Of the 5-kb sequence of the rat TRIII genomic clone 22 (see Fig. 3) obtained, 2021 bp are shown here. Sequence analysis by the Genetics Computer Group identified several regions consistent with previously identified nuclear factor-binding sites, which are shown in bold type, underlined, and identified above. Restriction sites used for promoter fragment analysis are shown in italics and identified above. The 5-kb sequence of 22 has been deposited as GenBank TM accession number AF117811.
transcripts of the rat TRIII gene initiate within an approximate 200-nucleotide span downstream from the GC box and Sp1 elements noted above, consistent with the previously reported 5Ј end of rat TRIII receptor cDNA (5,6).
TRIII Gene Promoter Activity in Osteoblasts-To determine regions within the rat TRIII 5Ј genomic DNA that might specify functional promoter activity, a series of intact, deleted, truncated, and reversed orientation DNA fragments were subcloned into transfection vector pGL2-Basic (Table I) and transfected into fetal rat osteoblasts, and reporter gene expression was measured. As shown in Fig. 3, the nest of DNA fragments that initiated at the 5Ј end with nucleotides ranging from Ϫ4045 to Ϫ534 and all terminating with nucleotide Ϫ253 within the untranslated region of exon 1 (5) directed significant reporter gene expression. The highest degree of promoter activity occurred with pTR3/1.8 (nucleotides Ϫ2013 to Ϫ253). Only a very low level of reporter gene expression was evident with pTR3/ 0.2B, which spans nucleotides Ϫ440 to Ϫ253 and eliminates the GC box at nucleotides Ϫ527 to Ϫ518. Even further truncation from the 5Ј end in constructs pTR3/0.2E and pTR3/0.1A, which lack the Sp1-binding site at nucleotides Ϫ431 to Ϫ424, completely eliminated promoter activity. By comparison to pTR3/1.8, inclusion of the next 185 bp upstream sequence in pTR3/1.9 (nucleotides Ϫ2194 to Ϫ253) significantly suppressed promoter activity. This lower level of gene promoter activity was also evident with construct pTR3/3.7 (nucleotides Ϫ4045 to Ϫ253). Reporter gene expression was also reduced with construct pTR3/1.9␦, where an internal deletion in pTR3/1.9 removed nucleotides Ϫ2061 to Ϫ980. Finally, no significant amount of reporter gene expression was induced by construct pTR3/1.9R, containing nucleotides Ϫ253 to Ϫ2199 in the reversed orientation.
cis-acting Sp1 Elements in the Basal TRIII Promoter-The basal TRIII promoter region defined by nested DNA fragment analysis was then examined by DNase I footprinting with nuclear extract from osteoblast-enriched cell cultures. As shown in Fig. 4, two major protected regions termed FP1 and FP2 correspond to the GC box at nucleotides Ϫ527 to Ϫ518 and to the Sp1-binding site at nucleotides Ϫ431 to Ϫ424. Consistent with this, oligonucleotide probes spanning each of the FP1 and FP2 sites (Table II) formed nuclear protein-DNA complexes that were identical to those obtained with a consensus Sp1 oligonucleotide probe (Fig. 5). Oligonucleotides with mutations within either the GC box or the Sp1-binding sequences failed to compete for nuclear factor binding to radiolabeled consensus Sp1 probe or by themselves to form nuclear protein/DNA complexes. Furthermore, antibody preparations specific for Sp1 or Sp3 depleted or supershifted complex formation. The lower molecular mass Sp3-DNA complex is thought to represent a processed, less abundant form of Sp3 (35). However, mutations introduced at either Sp1-binding site within TRIII promoter/ reporter transfection constructs (Fig. 6) only partially decreased reporter gene expression. Importantly, mutation of the GC box in pTR3/0.3P did not reduce promoter activity to the lower level of the truncated promoter construct pTR3/0.2B, in which the 94 bp upstream of pTR3/0.3P, which contain the GC box, were removed.
cis-acting AP-2 Elements in the Basal TRIII Promoter-To analyze the basal promoter region further, gel shift analysis with three overlapping oligonucleotide probes (GS1, GS2, and GS3) that spanned the area between the GC box and the Sp1-binding site described above showed distinct nuclear factor-binding patterns. Nuclear factor binding was strongest with probes GS2 (nucleotides Ϫ485 to Ϫ455) and GS3 (nucleotides Ϫ464 to Ϫ434) (Fig. 7, left panel). The several nuclear fractor/DNA complexes formed by these probes may correspond to less distinct footprints observed between FP1 and FP2 (Fig.  4). The Genetics Computer Group and MatInspector sequence analyses of the region between FP1 and FP2 suggested possible binding sites for nuclear factors Sp1, AP-2, and several E-box and zinc finger DNA-binding proteins (Table II and Refs. 34 -37). No discernible Sp1-like complexes formed with GS1 (nucleotides Ϫ511 to Ϫ480), GS2, or GS3, by comparison with a consensus Sp1 probe (Fig. 7, left panel), by competition with radiolabeled consensus SP1 oligonucleotide, or by reactivity with anti-Sp1 or anti-Sp3 antibody (data not shown). In contrast, GS2 and GS3 each competed with a probe containing consensus AP-2-binding sequence (Fig. 7, middle panel). By sequence analysis, GS3 contains consensus nuclear factor AP-2 and AP-4-binding sites (36). Complex formation was resistant to antibody to AP-4 (data not shown), whereas complexes consistent with AP-2 were readily evident with GS-3. Studies with specific anti-AP-2 antisera (37) showed that AP-2␣ and AP-2␥, but not AP-2␤, were present in extract from osteoblast-en- riched cultures (Fig. 7, right panel). Even so, overexpression of AP-2 by transfection with an AP-2␣ expression construct (37) did not significantly increase TRIII promoter activity directed by pTR3/0.3P (data not shown).
Regulation of TRIII Promoter Activity in Osteoblasts-On fetal rat osteoblasts, TGF-␤ binding to TRIII is rapidly suppressed by treatment with BMP-2 (12) and increased by glucocorticoid (13), consistent with changes in TRIII mRNA (Refs. 37 and 38 and Fig. 8, left panel). To assess whether these differences occur at least in part through variations in TRIII gene promoter function, we examined effects by BMP-2 and cortisol on four transfection reporter constructs (pTR3/3.7, pTR3/1.9, pTR3/1.8, and pTR3/0.4K) that span active upstream and downstream regions of the TRIII promoter. Treatment with cortisol enhanced reporter gene expression by all four constructs, predicting a glucocorticoid-dependent regulatory element located at minimum between nucleotides Ϫ687 and Ϫ253. Also consistent with its effect on TRIII mRNA and protein in these cells, BMP-2 suppressed TRIII promoter activity. However, this effect was only evident with pTR3/1.9 and TR3/ 3.7, predicting that it augments the effect of the putative si-lencer region noted between nucleotide Ϫ2194 and Ϫ2014 (Fig.  8, right panel). In contrast, neither glucocorticoid nor BMP-2 significantly altered TRIII promoter activity in fetal rat fibroblasts transfected with TR3/3.7 or with TR3/1.8, which are enhanced by glucocorticoid and/or suppressed by BMP-2 in transfected osteoblasts. Although these findings reflect the similarly limited effects by glucocorticoid and BMP-2 on radiolabeled TGF-␤ binding to TRIII on fibroblasts (Fig. 9), chronic exposure to either of these factors may perhaps cause more obvious differences in TRIII expression by these cells.
Overexpression of TRIII Suppresses the Stimulatory Effect of TGF-␤ on Osteoblast Protein Synthesis-Our earlier studies suggest that hormone-and growth factor-dependent changes in TRIII occur in parallel with variations in the sensitivity of bone cells to TGF-␤. In particular, a relatively higher level of TRIII correlates well with a reduced response to treatment with   Table I. U, StuI; C, SacI; N, NcoI; M, MscI; K, KpnI; S, SmaI; E, EagI; B, BssHII; A, ApaI. Constructs were co-transfected with pSV-␤-galactosidase into primary osteoblast-enriched cell cultures with LipofectAMINE. Luciferase reporter gene activity was measured after 48 h and corrected for protein content and relative ␤-galactosidase expression. Bars represent means Ϯ S.E. of data from 3-4 independent overlapping studies and 9 -12 replicate samples per condition.
FIG. 4. DNase I footprinting assay of the rat TRIII gene promoter. Nuclear extract obtained from primary osteoblast-enriched cultures was hybridized with a 3Ј 32 P-end labeled DNA probe encompassing nucleotides Ϫ641 to Ϫ350 of the rat TRIII gene promoter. Undigested DNA fragments were analyzed by eletrophoresis on a sequencing gel and visualized by autoradiography. Lanes 1-4 show results with 40, 80, 160, and 0 g of nuclear extract, as indicated. The sequence of the two major protected regions, FP1 and FP2, are shown on the right, and the Sp1-binding sequences that they encompass are shown by the vertical bars.
TGF-␤1 (10,12,13,40). To address this in a way that would limit the contribution of variations in TRI, TRII, downstream signaling components, or nuclear effectors of osteoblast activity (13,39), osteoblasts were transiently transfected to overexpress rat TRIII. By comparison to vector transfected osteoblasts, radioligand binding to TRIII increased 2.8-fold in cells transfected with the TRIII expression construct. Consistent with situations where higher levels of TRIII are expressed, such as on less differentiated bone cells or on osteoblasts treated with glucocorticoid, PTH, or its related protein, PTHrelated protein (10,12,13,40,41), the stimulatory effect of TGF-␤ on collagen and noncollagen protein synthesis was significantly lower, reduced by 43 and 32% relative to control (Fig. 10). DISCUSSION On many cells, changes in the TR profile can significantly alter TGF-␤ activity (7). However, mechanisms that regulate TR expression or stability are still poorly understood. Our earlier studies in osteoblasts showed that TR levels are controlled by transcriptional and post-transcriptional events. In isolated bone cells, TR mRNAs and proteins exhibit relatively short half-lives, offering the opportunity for rapid changes in TGF-␤ sensitivity (40). Moreover, osteotropic factors like BMP-2, glucocorticoid, PTH, and PTH-related protein specifically alter the TR profile and modify the effects of TGF-␤ on osteoblasts (10,12,13,38,39,41). To understand these events at the molecular level, we first isolated and cloned the rat TRI gene promoter and characterized several cis-and trans-acting elements that control constitutive and hormone-dependent TRI expression (24,25,39,42). TRIII is often the most abundant TR and can help to define TGF-␤ isoform activity or its biological effects (6 -17). In the current study, we therefore cloned the rat FIG. 5. Gel mobility shift assays of the GC box and Sp1-binding sequences in the rat TRIII gene promoter. 32 P-Labeled oligonucleotide probes described in Table II were incubated with nuclear extract from primary osteoblast-enriched cultures without or with a 100fold molar excess of unlabeled oligonucleotides (left panel) or anti-Sp1 or anti-Sp3 antibody preparations (right panel) as indicated. Protein-DNA complexes were resolved on 5% nondenaturing polyacrylamide gel and visualized by autoradiography. No reaction occurred with normal rabbit serum (not shown).
FIG. 6. Effects of mutations in the GC box and Sp1-binding sequences on rat TRIII gene promoter activity. Plasmids pTR3/ 0.3P, pTR3/0.3P, pTR2/0.2B, and pTR3/0.2B containing the rat TRIII gene promoter fragment inserts with native and mutated GC box and Sp1-binding sequences described in Table I Table II were incubated with nuclear extract from primary osteoblast-enriched cultures without (left panel), with a 100-fold molar excess of unlabeled oligonucleotides (middle panel), or with anti-AP-2 isoform-specific antibody preparations (right panel) as indicated. Protein-DNA complexes were resolved on 5% nondenaturing polyacrylamide gel and visualized by autoradiography. No reaction occurred with normal rabbit serum (not shown).

TABLE II
Oligonucleotide probes used in electrophoretic mobility shift assay Oligonucleotides G, S, GS1, GS2, and GS3 were derived from the 5Ј region of the rat TRIII gene at the positions shown. Oligonucleotides G and S were designed to include disruptions (bold) in the possible GC box and Sp1 binding sites. Oligonucleotides SP1 and AP2 were designed to include consensus Sp1-or AP-2-binding sequences. TRIII gene promoter and have begun to define elements that can account for basal and conditional TRIII expression by bone cells.
Sequence analysis showed that DNA within 5.0 kb upstream of the coding region of the rat TRIII gene lacks TATA and CCAAT boxes. However, it contains two nuclear factor Sp1binding sites in the highly GC-enriched 3Ј basal promoter region, comprising a so-called CpG island. Multiple transcription initiation sites occur within the basal TRIII promoter, a situation often associated with genes lacking TATA and CCAAT box elements (32,33). In general, Sp1 can activate gene expression, whereas Sp3 can be stimulatory or suppressive (33). Proximal Sp1-binding sites are thought to function in a cooperative way, forming complexes that initiate transcription from multiple sites. Furthermore, several transcription factors associate with Sp1, and in this way enhance or reduce the activation of specific gene promoters (43). Mutation of either of the two Sp1binding sequences in this region of the TRIII gene promoter reduced its activity by approximately one-third. However, a truncation removing 94 nucleotides that included the more upstream Sp1-binding site caused an 85% decrease in promoter function. The sequence between these sites is GC enriched to 75% and contains several possible nuclear factor-binding sites, including Sp1, AP-2, and Ap-4. By gel shift analysis and antinuclear factor antibody reactivity, little or no binding by Sp1 or AP-4 occurred in this region, whereas complexes consistent with AP-2 were evident. Nonetheless, overexpression of AP-2 by transfection did not further stimulate this region of the TRIII promoter. This suggests that the amount of endogenous AP-2 may be sufficient for TRIII expression or that complex interactions between Sp1 and AP-2 may govern overall TRIII expression (44). Other oligonucleotide probes from this region also formed several gel shift complexes, but their identity is not yet known. Thus, our current findings suggest effects by Sp1 and AP-2, perhaps among other nuclear factors within this region, although none of these sites by themselves appears to have a dominant influence.
Organization of the basal region of the TRIII gene promoter is similar to that for other growth factor receptor genes (25) and does not itself seem to account for differences in TRIII expression among various tissues, during development, or in response to regulatory factors or events. In osteoblasts, we found that TRIII promoter activity was induced by glucocorticoid and suppressed by BMP-2. Although promoter-dependent reporter gene expression only represents an indication of relative changes in authentic TRIII gene expression, the magnitude of the effects that we observed were consistent with our earlier evidence for changes in TGF-␤ binding to TRIII and on TRIII mRNA (12,13,38,39). Importantly, analogous effects are either not seen or less evident in undifferentiated periosteal cells or fibroblasts (Refs. 12, 40, and 48 and our current studies), suggesting phenotype-related differences. Sensitivity to glu- FIG. 8. Regulation of rat TRIII gene promoter activity by glucocorticoid and BMP-2. Left panel, osteoblast-enriched cultures were treated for 24 h with 10 nM cortisol or 1 nM BMP-2 and extracted, and total RNA was assessed by Northern blot analysis with a rat-specific cDNA probe (4). Numbers to the left refer to an ethidium-stained sizing ladder from a parallel gel lane. Ethidium stained rRNA profiles are shown below, with 28 and 18 S rRNA bands indicated. Right panel, the relative sizes of plasmids pTR3/3.7, pTR3/1.9, pTR3/1.8, and pTR3/0.4K with the rat TRIII gene promoter fragment inserts described in Table I are shown to the left. Constructs were transfected for 24 h into osteoblast-enriched cultures. Cells were then treated for 24 h with vehicle, 10 nM cortisol, or 1 nM BMP-2 in serum-free medium, and reporter gene expression was assessed as described in the legend to Fig. 3. Promoter regions with potential cortisol or BMP-2-responsive sequences are shown as dark segments overlaying the rectangles representing promoter fragments. Luciferase activity was corrected for protein content and relative ␤-galactosidase expression. Data bars represent means Ϯ S.E. from 4 -6 independent overlapping studies and 12-30 replicate samples per condition. The numbers in parentheses are results expressed as a percentage of control, set as 1 in untreated cells transfected with each promoter/reporter construct. Their individual control activities differed from each other precisely as as shown in Fig. 3.   FIG. 9. Effects by glucocorticoid and BMP-2 on TRIII expression by fetal rat fibroblasts. Left panel, fetal rat fibroblasts were transfected for 24 h with plasmid constructs pGL2-Basic, pGL2-Control, pTR3/1.8, or pTR3/3.7, treated for 24 h with vehicle, 10 nM cortisol, or 1 nM BMP-2 in serum-free medium, and reporter gene expression was assessed as described in the legend to Fig. 8. Data bars represent means Ϯ S.E. from 2 independent studies and 6 replicate samples per condition. pGL2-Control enhanced reporter gene activity by 479 Ϯ 48-fold, pTR3/1.8 enhanced reporter gene expression by 120 Ϯ 7-fold, and pTR3/3.7 enhanced reporter gene expression by 68 Ϯ 6-fold, relative to pGL2-Basic. By analysis of variance, no significant effects were induced by cortisol or BMP-2 on TRIII reporter gene expression. Right panel, fetal rat fibroblasts were treated for 24 h with vehicle, 10 nM cortisol, or 1 nM BMP-2 in serum-free medium. Cultures were labeled with 125 I-labeled TGF-␤1 and extracted, and TR profiles were assessed by polyacrylamide gel electrophoresis and autoradiography, as described (12,40). Collagen and noncollagen protein synthesis were determined by differential sensitivity to purified bacterial collagenase. Data bars represent the means Ϯ S.E. from 4 independent studies and 13 replicate samples per condition. In vector transfected cells, TGF-␤ enhanced collagen synthesis by 7.6 Ϯ 0.9-fold and noncollagen protein synthesis by 5.2 Ϯ 0.3-fold. By analysis of variance, no significant effects were induced by TRIII overexpression in untreated cultures, but the stimulatory effects of TGF-␤ on collagen and noncollagen protein synthesis were significantly reduced (p Ͻ 0.05) by 43 Ϯ 4 and by 32 Ϯ 3%, respectively. Right panel, osteoblasts transfected with empty vector or the rat TRIII expression construct were labeled with 125 I-labeled TGF-␤1 and extracted, and TR profiles were assessed by polyacrylamide gel electrophoresis and autoradiography, as in Fig. 9. cocorticoid occurs near the 3Ј end of the TRIII promoter. Initial studies to locate possible cis-acting elements that allow this effect suggest at least two response regions between nucleotides Ϫ687 and Ϫ495 and between nucleotides Ϫ440 and Ϫ386. 3 The more upstream region contains a consensus C/EBP-binding site, consistent with the stimulatory effect of glucocorticoid on C/EBP expression in adipocytes (45,46) and bone cells. 4 However, the more downstream response region contains no identifiable glucocorticoid response element or C/EBP-binding site, suggesting interactions with other transacting factors. The inhibitory effect of BMP-2 is only evident with TRIII promoter fragments above nucleotide Ϫ2013, consistent with its ability to enforce the effect of an endogenous silencer region initially apparent by nested fragment deletion analysis. Sequence analysis shows a variety of possible binding sites in this region. Notably, it contains two binding domains for Myc/Max nuclear factors whose activity may be suppressed when Mad subunit expression increases during tissue and organ development (47). Further studies to define this element may therefore help to explain the significant decrease in TRIII expression that occurs with native or BMP-2 induced differentiation of osteoblasts (10,12).
In summary, we have cloned the rat TRIII gene promoter and have begun to indentify regulatory regions that control basal, hormone, and growth factor induced changes in TRIII expression previously observed on rat osteoblasts. Indeed, forced overexpression of TRIII reduced the effectiveness of TGF-␤1 treatment, consistent with the relatively lower activity of TGF-␤ in less differentiated bone cell cultures where proportionately more TRIII is endogenously expressed (12,40,48). Future studies to define in more detail the conditional elements that alter TRIII expression may help to decipher the complex events that control the changes in TGF-␤ sensitivity and its biological effects in bone and in other tissues. A better understanding of the nuclear factors that regulate the loss of TRIII expression with differentiation may further increase our understanding of the gene repression that must occur to limit tissue growth during development and to avoid hyperplastic disease.