Identification of the cAMP Response Element That Controls Transcriptional Activation of the Insulin-like Growth Factor-I Gene by Prostaglandin E 2 in Osteoblasts*

,

The anabolic role of insulin-like growth factor-I (IGF-I) 1 in skeletal growth is well documented (1)(2)(3), and until recent years it was believed that much if not all of the IGF-I in the skeleton arrived through the circulation following production by the liver. However, we now know that osteoblasts synthesize IGF-I and that its expression is stimulated by parathyroid hormone (PTH) and by locally produced prostaglandin E 2 (PGE 2 ) (4,5). Both of these agents influence bone remodeling, both elevate intracellular levels of cAMP in osteoblasts, and each either directly or indirectly influences osteoblast and osteoclast activity, possibly in part through stimulatory effects on IGF-I expression (6 -13). Consequently, locally produced IGF-I is likely to be important in regulating the activity of skeletal cells. Earlier we observed that IGF-I functions as a coupling factor through its ability to stimulate collagen synthesis following intermittent exposure to PTH (10). Therefore, IGF-I is now believed to serve as a key factor in skeletal growth, integrity, and bone remodeling. This central role of IGF-I in skeletal health has prompted us to examine the molecular events involved in cAMP-dependent regulation of IGF-I expression in bone cells. PGE 2 was chosen as the prototypical cAMP-inducing agent because 1) it is a natural product of arachidonic acid metabolism; 2) PGE 2 is made by osteoblasts in response to PTH, various growth factors, or lymphokines; and 3) it may function as a signal-transducing agent in mechanically strained skeletal tissue (14 -19).
Although the rat IGF-I gene contains two promoters (promoter 1, P1, is 5Ј to exon 1, and P2 is 5Ј to exon 2 (20)), P1 functions selectively in cultures of primary fetal rat osteoblasts to control basal and PGE 2 -stimulated IGF-I expression (21). Agents that elevate intracellular cAMP concentrations increase transcription of the IGF-I gene in these cells (21,22). While the response to PTH and to PGE 2 in bone cells consists of stimulation of protein kinases A (PKA) and C and the mobilization of calcium (23)(24)(25), three pieces of evidence demonstrate the involvement of PKA in IGF-I gene activation. First, all hormones and chemical agents tested to date that significantly elevate cAMP levels in osteoblasts also increase IGF-I gene expression (26). Second, cotransfection of osteoblasts with an IGF-I P1-luciferase reporter gene and an expression plasmid for the catalytic subunit of PKA produces a constitutively high level of promoter activity that is not stimulated further by PGE 2 treatment (27). Finally, cotransfection of an IGF-I P1luciferase reporter gene with an expression plasmid for a mu-tant regulatory subunit of PKA that is unresponsive to cAMP completely blocks the induction of IGF-I promoter function by PGE 2 (27).
In previous studies, we mapped a functional cAMP response element (CRE) to the 5Ј-untranslated region (UTR) of exon 1 within a segment (ϩ196 to ϩ328) that lacks consensus CRE or AP-2 binding sites, features often associated with conventional cAMP-responsive promoters (28 -31). We now have identified a short region within this part of the 5Ј-UTR that is essential for full hormonal responsiveness and have characterized inducible DNA-protein interactions involving this site. Transient transfections of IGF-I P1-reporter genes showed that an intact previously footprinted site, HS3D (base pairs ϩ193 to ϩ215), was required for a full 5-fold response to PGE 2 . PGE 2 also stimulated nuclear protein binding to HS3D. Binding was not seen with protein extracts from untreated osteoblast cultures, was detected within 2 h of PGE 2 treatment, and was maximal by 4 h. This inducible DNA-protein interaction did not require ongoing protein synthesis, since it was not blocked by cycloheximide (Chx), and Chx did not impair stimulation of IGF-I mRNA levels by PGE 2 . Competition gel shifts with mutated DNA probes identified CGCAATCG as the minimal sequence needed for inducible binding. Although binding of nuclear proteins to this site was inhibited by an oligonucleotide containing a CRE consensus sequence (TGACGTCA), DNA-protein interactions involving the consensus CRE were not stimulated by PGE 2 . Taken together with previous studies, our observations define a potentially novel mechanism of gene regulation by cAMP that involves inducible binding to a hormone response element that is distinct from a canonical CRE.

EXPERIMENTAL PROCEDURES
Cell Cultures-Primary osteoblast-enriched cell cultures were prepared from the parietal bones of 22-day-old Sprague-Dawley rat fetuses (Charles River Laboratories, Raleigh, NC). Animals were housed and euthanized by methods approved by the Yale University Animal Care and Use Committee. Cranial sutures were eliminated during dissection, and the bones were digested with collagenase for five sequential 20-min intervals. The cell population released during the last three digestions exhibits biochemical characteristics associated with differentiated osteoblasts, including PTH receptors, type I collagen synthesis, and a rise in osteocalcin expression in response to dihydroxyvitamin D 3 (32,33). Histochemical staining demonstrates that approximately 80% of the cells express alkaline phosphatase, 2 although this itself is not entirely specific for osteoblasts. However, using these criteria, as well as differential sensitivity to transforming growth factor-␤, bone morphogenetic protein-2, various prostaglandins, and the ability to form mineralized nodules in vitro (34,35), these cells are well distinguished from the less differentiated cells released during earlier collagenase digestions. Cells from the last three digestions were pooled and then plated at 4800/cm 2 in Dulbecco's modified Eagle's medium containing 20 mM HEPES (pH 7.2), 0.1 mg/ml ascorbic acid, penicillin, and streptomycin (all from Life Technologies, Inc.), and 10% fetal bovine serum (Sigma). Cycloheximide (Sigma) was used at a final concentration of 2 M, and its use preceded other treatments by 15 min to ensure its effectiveness prior to incubation with vehicle or PGE 2 .
Plasmids-Rat IGF-I cDNA was kindly provided by Dr. Liam Murphy. Rat IGF-I promoter 1 constructs have been described previously (20). All plasmids were propagated in Escherichia coli strain DH5␣ with ampicillin selection and were prepared using a Qiagen ® Plasmid Kit (Qiagen Inc., Chatsworth, CA) and the manufacturer's recommended protocol. Mutant constructs were produced by PCR, and the mutations were verified by DNA sequence analysis.
RNA Isolation and Analysis-Cultures of 9.6 cm 2 were solubilized in buffer consisting of 5 M guanidine monothiocyanate, 25 mM trisodium citrate, 0.5% sarkosyl, and 0.1 M 2-mercaptoethanol, followed by extraction with phenol/chloroform/isoamyl alcohol (75:25:1) in the presence of 0.2 M sodium acetate (36). Total RNA was precipitated, ethanol-washed, dried, and resuspended in diethylpyrocarbonate treated water, and concentration and purity were determined by absorbance at 260 and 280 nm. Ten micrograms of RNA was denatured with 2.2 M formaldehyde, 12.5 M formamide at 65°C for 15 min and fractionated on a 1.5% agarose, 2.2 M formaldehyde gel. Co-electrophoresed RNA standards were excised and ethidium bromide-stained, and the remaining gel was blotted onto charged modified nylon (GeneScreen Plus TM , DuPont, NEN). A restriction fragment containing the rat IGF-I cDNA clone was purified from an agarose gel and labeled with [␣-32 P]deoxycytidine triphosphate and [␣-32 P]thymidine triphosphate by random hexanucleotide-primed second strand synthesis (37). Northern blots were hybridized with [ 32 P]IGF-I cDNA, and the filters were washed under conditions of progressively increasing stringency. Final washes were with 0.2 ϫ SSC (20 ϫ SSC contains 3 M NaCl, 0.3 M trisodium citrate, pH 7.0) and 0.1% sodium dodecyl sulfate for 1 h at 55°C. The bound radioactive material was visualized by autoradiography using Amersham Hyperfilm ® and a DuPont Cronex intensifying screen. Filters were eluted of specifically bound 32 P-labeled cDNA by washing in deionized water for 5 min at 100°C before probing with 18 S antisense ribosomal RNA (Ambion, Austin, TX).
Transfection Studies-IGF-I promoter 1-luciferase reporter plasmids (1.5 g/9.6-cm 2 culture well) were co-transfected with a vector carrying the ␤-galactosidase gene under SV40 promoter control (1.0 g/culture well; pSV-␤-Galactosidase Control Vector, Promega Corp.) to normalize for transfection efficiency. Cultures at 50% confluent density were rinsed in serum-free medium and exposed to plasmids in the presence of Lipofectin TM (Life Technologies, Inc.) for 3 h. The solution was then replaced with growth medium containing 5% fetal bovine serum, and the cultures were grown to confluence (48 h). Confluent cultures were rinsed with serum-free medium and treated for 6 h with vehicle (ethanol diluted 1:1000 or greater), or PGE 2 (Sigma). At the end of the treatment interval, the medium was aspirated, and cultures were rinsed with phosphate-buffered saline and then lysed in 100 l of 25 mM Tris-phosphate (pH 7.8), 2 mM dithiothreitol, 2 mM 1,2-diaminocyclohexane-N,N,NЈ,NЈ-tetraacetic acid, 10% glycerol, 1% Triton X-100 (cell lysis buffer, Promega Corp.). Nuclei were pelleted at 12,000 ϫ g for 5 min, and the supernatants were stored at Ϫ75°C until assay. Commercial kits were used to measure luciferase (Promega Corp.) and ␤-galactosidase (Tropix, Bedford, MA). Protein was determined by the Bradford assay (38).
Nuclear Protein Extracts-Confluent osteoblast cultures were serum-deprived for 20 h. The cultures then were rinsed with serum-free medium and exposed to vehicle or 1 M PGE 2 for 1-4 h. Medium was aspirated, and cultures were rinsed twice with phosphate-buffered saline at 4°C; all subsequent steps were performed on ice or at 4°C. Cells were harvested with a cell scraper and gently pelleted, and the pellets were washed with PBS. Nuclear extracts were prepared by the method of Lee et al. with minor modifications (27,39). Cells were lysed in hypotonic buffer (10 mM HEPES (pH 7.4), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol) with phosphatase inhibitors (1 mM sodium orthovanadate, 10 mM sodium fluoride, 0.4 M microcystin CL), protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 2 g/ml leupeptin, 2 g/ml aprotinin (Sigma), and 1% Triton X-100. Nuclei were pelleted, and the supernatant (cytoplasmic extract) was collected and dialyzed as described below for nuclear extracts. Nuclei were resuspended in hypertonic buffer containing 0.42 M NaCl, 0.2 mM Na 2 EDTA, 25% glycerol, and the phosphatase and protease inhibitors indicated above. Soluble proteins released by a 30-min incubation at 4°C were collected by centrifugation at 12,000 ϫ g for 5 min, and the supernatant was dialyzed for 2 h against 2000 volumes of buffer (20 mM HEPES, pH 7.4, 0.1 M KCl, 0.1 mM Na 2 EDTA, 0.5 mM dithiothreitol, 1 mM sodium orthovanadate, 20% glycerol), containing the protease inhibitors indicated above.
Gel Mobility Shift Assay-Gel mobility shift experiments followed previously published methods (40 -42). Briefly, radiolabeled doublestranded probes were prepared by annealing complementary oligonucleotides, followed by fill-in of single-stranded overhangs with dCTP, dGTP, dTTP, and [␣-32 P]dATP, using the Klenow fragment of DNA polymerase I. Twenty micrograms of nuclear extract or 50 g of cytoplasmic proteins were preincubated for 20 min on ice with 2 g of poly(dI-dC) with or without unlabeled specific or nonspecific competitor DNAs 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 the addition of 5 ϫ 10 4 cpm of DNA probe (0.1-0.2 ng) for 30 min on ice, samples were applied to a 4 -20% nondenaturing polyacrylamide gradient gel (Novex, San Diego, CA) that had been preelectrophoresed for 30 min at 12.5 V/cm at 25°C in 45 mM Tris, 45 mM boric acid, 1 mM EDTA. Electrophoresis proceeded for 2.5 h under identical conditions. The dried gels were exposed to x-ray film at Ϫ80°C with intensifying screens. 2  The nucleotide sequences of the sense strand of oligonucleotides used in the gel mobility shift assays were as follows: HS3D, 5Ј-TTCAGAG-CAGATAGAGCCTGCGCAATCGAAA-3Ј; Oct 1, 5Ј-TTTTAGAGGATC-CATGCAAATGGACGTACGAAA-3Ј; rat somatostatin CRE, 5Ј-TCCT-TGGCTGACGTCAGAGAGAGAGTTTAAA-3Ј.
Statistical Analysis-When statistical analysis was conducted, data were assessed by one way analysis of variance, using Kruskal-Wallis or Bonferonni methods for post hoc analysis.

RESULTS
Our previous studies showed that transcription of the major promoter of the rat IGF-I gene, promoter 1 (20,(43)(44)(45), was activated by treatment of osteoblast cultures with PGE 2 (21), and transient transfection experiments indicated that a portion of the 5Ј-UTR of exon 1 was required for full hormonal responsiveness (27). Additional studies demonstrated that PGE 2 stimulated IGF-I gene expression by a classical cAMP-mediated pathway that required activation of PKA, since IGF-I promoter function could be induced by cotransfection with the catalytic subunit of PKA, and could be blocked by a dominant negative mutant regulatory subunit (27). A single DNase I footprint was identified within this hormonally responsive segment of exon 1 at a site termed HS3D (27). The current experiments were designed to test the hypothesis that HS3D mediated PGE 2activated IGF-I gene transcription.
As seen in Fig. 1, and in agreement with previous results (27), a 6-h incubation of osteoblast cultures with 1 M PGE 2 stimulated a 5.3-fold increase in luciferase activity regulated by a transiently transfected recombinant plasmid containing a 2039-nucleotide IGF-I promoter 1 fragment that included the entire 328-nucleotide 5Ј-UTR of exon 1 (IGF1711b/Luc). As previously reported (27), deletion of exon 1 to position ϩ195 (IGF1711c/Luc) caused a substantial drop in hormonal activation of luciferase. We then examined specific substitution mutations within HS3D (HS3Dmut(AAA), HS3D LSM1, and HS3D LSM2; the nucleotide substitutions in these mutant constructs indicated in Fig. 1 are underlined and in boldface type). Each mutant construct exhibited a diminished response to PGE 2 treatment, leading to a decline in inducible luciferase activity that was equivalent to the level observed with plasmid IGF1711c/Luc (Fig. 1). By contrast, nucleotide substitutions in another DNase I footprint site, HS3C (ϩ147 to ϩ169), which was detected with rat liver nuclear extracts but not osteoblast extracts (27,46), had little effect on regulation of IGF-I promoter function by PGE 2 . Taken together, these results indicate that HS3D is involved in PGE 2 -stimulated IGF-I gene expression in bone cells.
We next employed gel mobility shift studies to examine the interaction of a 32 P-labeled double-stranded HS3D DNA probe with proteins from osteoblast cultures. Figs. 2 and 3 show time course experiments. PGE 2 treatment led to inducible binding of nuclear proteins to the HS3D probe. No binding was seen in extracts from vehicle-treated cells. A strong gel shift that migrated as a closely spaced doublet on a 4 -20% gradient gel was observed after 2-4 h of incubation with PGE 2 and occasionally could be detected within 1 h of PGE 2 treatment. The DNAprotein interaction stimulated by PGE 2 was confined to nuclear extracts and was not observed with cytoplasmic proteins (Fig.  3), indicating that a PGE 2 -activated signaling pathway did not stimulate nuclear translocation of a previously activated cyto- ). An additional mutant IGF-I P1 construct with nucleotide substitutions in another DNase I footprint site present in liver, but not in osteoblast nuclear extracts, HS3C (designated HS3C LSM1) also was tested in transient transfections. All constructs with the exception of IGF1711c/Luc (Ϫ1711 to ϩ195) were based on the wild-type IGF1711b/Luc (promoter 1 construct spanning Ϫ1711 to ϩ328 base pairs) and have nucleotide substitutions only where indicated. Altered bases are underlined and in boldface. Right panel, IGF-I promoter-luciferase reporter plasmids were cotransfected with a pSV-␤-galactosidase control vector into osteoblast enriched cultures (9.6 cm 2 ) using Lipofectin TM . Cultures were grown to confluence (48 h), the growth medium was aspirated, and the cultures were rinsed with serum-free Dulbecco's modified Eagle's medium. Cultures were exposed to control medium (containing vehicle), or PGE 2 (1 M) for 6 h. Cytoplasmic extracts were prepared, and luciferase activity was determined. Data are corrected for transfection efficiency (␤-galactosidase expression) and for protein content of cytoplasmic extracts. Transfections were performed in triplicate, and results are pooled data for three or more separate experiments and for a total of 9 or more replicate cultures. The mean Ϯ S.E. for luciferase expression compared with vehicle-treated cultures (-fold stimulation) for pooled experiments is shown. Deletion of ϩ196 to ϩ328 base pairs (IGF1711c/Luc), and base substitutions that span the HS3D binding site (HS3D(AAA), HS3D LSM1, HS3D LSM2) produced a comparable and statistically significant decrease in PGE 2 -stimulated luciferase expression (p Ͻ 0.05 versus IGF1711b/Luc), indicated by the asterisk.
plasmic DNA binding protein.
In contrast to the inducible binding of the HS3D probe to nuclear proteins from osteoblast cultures, a constitutive gel shift was seen with a 32 P-labeled double-stranded oligonucleotide containing the CRE from the rat somatostatin gene (Fig.  2). Despite the kinetic differences in nuclear protein binding between the HS3D and somatostatin CRE probes, in competition studies an unlabeled CRE oligonucleotide inhibited the interaction of nuclear protein extracts from osteoblast cultures with the 32 P-labeled HS3D probe with the same effectiveness as unlabeled HS3D (Fig. 4). A mutant CRE did not interfere with the HS3D gel shift (data not shown). By contrast, excess unlabeled HS3D only weakly diminished binding of the 32 Plabeled CRE probe to nuclear extracts from PGE 2 -treated osteoblast cultures (Fig. 4). These results suggest that similar proteins may interact with these DNA probes but with different affinities.
Additional competition gel shift experiments were performed to localize the nucleotide sequence within HS3D required for PGE 2 -inducible nuclear protein binding. A series of mutant oligonucleotides was synthesized with substitutions of purines for purines and pyrimidines for pyrimidines in consecutive blocks of four bases (Fig. 5A). Results using these competitor DNAs are presented in Fig. 5 and show that all but two mutants (M6 and M7) inhibited binding of the [ 32 P]HS3D probe when tested at a 200-fold molar excess. In addition, a doublestranded oligonucleotide containing the AAA mutation that lacked inducible promoter function (Fig. 1) also did not compete for binding to the [ 32 P]HS3D probe, and neither a labeled AAA nor M6 oligomer was able to bind osteoblast nuclear proteins (data not shown). Taken together, these data define the minimal binding sequence as 5Ј-CGCAATCG-3Ј, located at nucleotides ϩ202 to ϩ209 of exon 1 (20,27), although additional flanking nucleotides could potentially modulate DNA-protein interactions at this site. Since all IGF-I P1-luciferase reporter genes with mutations within this eight-nucleotide DNA binding core sequence showed a diminished response to PGE 2 ( Fig.  1), these results additionally demonstrate the functional importance of this inducible DNA-protein interaction.
Our previous studies showed that induction of IGF-I gene expression by PTH in osteoblast cultures was not blocked by the protein synthesis inhibitor, cycloheximide (Chx) (4). As shown in Fig. 6A, PGE 2 treatment also stimulated accumulation of IGF-I mRNA, even in the presence of 2 M Chx, a concentration that inhibited Ͼ90% of ongoing protein synthesis, as measured by incorporation of [ 3 H]proline (ϪChx, 7.2 Ϯ 0.1 ϫ 10 Ϫ3 cpm versus ϩChx, 0.71 Ϯ 0.05 ϫ 10 Ϫ3 cpm). However, at 24 h Chx caused a small increase in basal steady-state IGF-I transcripts and modestly reduced the stimulatory effect of PGE 2 . Pretreatment of osteoblast cultures with Chx also did not prevent the PGE 2 -induced gel shift of the 32 P-labeled HS3D probe (Fig. 6B). These results indicate that PGE 2 -regulated DNA-protein interactions at HS3D and PGE 2 -stimulated IGF-I gene transcription represent primary hormonal responses that do not require ongoing protein synthesis and suggest that PGE 2 treatment modifies a preexisting DNA-binding protein(s), leading to an increase in its affinity for HS3D. Lane 2 shows a gel shift of 32 P-labeled HS3D oligonucleotide in the absence of unlabeled competitor, while lanes 3-5 show results for competition with unlabeled wild-type HS3D, Oct-1, and consensus CRE oligonucleotides, respectively. Lanes 7-10 show nuclear protein binding to a 32 P-labeled rat somatostatin CRE probe (examined with extracts from PGE 2 -treated cultures). Lanes 7-10 show the gel shift pattern in the absence of unlabeled competitor, unlabeled wild-type HS3D, Oct-1, and consensus CRE oligonucleotides, respectively. Arrows point to the DNA-protein complexes. Lanes 1 and 6 indicate migration of the labeled probes in the absence of incubation with nuclear protein. Autoradiographic exposure was for 20 h at Ϫ80°C with intensifying screens. The experiments were repeated three times with identical results. The nucleotide sequences of oligonucleotides used in competition gel mobility shift assay are given under "Experimental Procedures." DISCUSSION Our discovery that the calciotropic hormone PTH stimulates IGF-I gene expression in primary fetal rat osteoblasts and that IGF-I serves as a coupling factor for bone remodeling (4, 10) has motivated an investigation into the mechanism(s) regulating IGF-I gene activation in these cells. Initially, a PKA-dependent pathway was implicated, since all agents tested that significantly elevated cAMP levels in osteoblasts, including PGE 2 , also enhanced IGF-I mRNA and protein production (26). Following these initial observations, we could only speculate about the existence of CREs in the IGF-I gene, since its two promoters were undefined at the time (26). Once promoters 1 and 2 were identified and characterized, we discovered that cAMP stimulated IGF-I gene expression in osteoblasts through a transcriptional mechanism that was exclusively dependent on promoter 1 (21). We further demonstrated a direct role for PKA in IGF-I gene activation in osteoblasts by showing that cotransfection of the catalytic subunit of PKA with an IGF-I P1luciferase fusion gene enhanced reporter gene expression without additional hormonal stimulation, while cotransfection of a mutant regulatory subunit that was unable to bind cAMP blocked PGE 2 -induced luciferase activity (27). Although IGF-I P1 contains a single, near consensus CRE and several potential binding sites for transcription factor AP-2, which also can stimulate cAMP-responsive genes (31), deletion of these regions did not block cAMP-activated reporter gene expression (27). Additional transient transfection experiments indicated that a segment within the 5Ј-UTR of exon 1 from ϩ196 to ϩ328 was essential for the PKA-dependent stimulation of IGF-I promoter function, and DNase I footprinting identified one binding site for osteoblast nuclear proteins within this functionally defined hormone response region (27). Since exon 1 lacked any nucleotide sequence similarity with known CREs, we concluded that the IGF-I gene contained a novel CRE within the 5Ј-UTR of exon 1 (27).
We now have identified the DNA sequence required for stimulation of IGF-I promoter function by cAMP (PGE 2 ). Point mutations in the previously footprinted HS3D site in the context of full-length promoter 1 diminished PGE 2 -induced luciferase expression to a level identical to that found with deletion of nucleotides ϩ196 to ϩ328 of the 5Ј-UTR of exon 1. Moreover, mutation of another site, HS3C (ϩ147 to ϩ169), which is not footprinted by osteoblast nuclear proteins (27), had no effect on PGE 2 -regulated promoter function. PGE 2 treatment also induced the binding of osteoblast nuclear proteins to HS3D, and neither this hormonal response nor activation of IGF-I gene expression by PGE 2 was blocked by the protein synthesis inhibitor, cycloheximide. Taken together, these results indicate that cAMP-activated binding of osteoblast nuclear proteins to HS3D initiates a primary signaling pathway that stimulates IGF-I gene transcription through promoter 1.
A series of mutagenesis experiments defined the minimal DNA binding segment of HS3D to be 5Ј-CGCAATCG-3Ј, located at base pairs ϩ202 to ϩ209 in exon 1, although additional studies will be required to assess the importance of adjacent nucleotides in modifying DNA-protein binding. As shown by functional studies, substitutions within these eight nucleotides also diminished responsiveness to cAMP. This sequence and its Total RNA (10 g) was run on a 1.5% agarose, 2.2 M formaldehyde gel, blotted onto charge-modified nylon, and probed with 32 P-rat IGF-I cDNA, or 32 P-18 S antisense rRNA (as described under "Experimental Procedures"). Panel B, gel mobility shift experiments were performed as described under "Experimental Procedures" with nuclear extract from osteoblast-enriched cultures treated for 4 h with 1 M PGE 2 in the absence or presence of 2 M Chx. Nuclear protein binding to 32 P-labeled HS3D probe is shown in lanes 1-4. Nuclear protein binding to 32 Plabeled rat somatostatin CRE probe is shown in lanes [5][6][7][8] flanking DNA are highly conserved in IGF-I genes from seven different vertebrate species (including human, pig, chicken, Xenopus, coho salmon, and rainbow trout) (47)(48)(49)(50)(51)(52), further indicating the possible regulatory importance of the HS3D region of exon 1. While primary human osteoblasts also respond to cAMP with enhanced IGF-I gene expression (53), it remains to be shown if cAMP stimulates IGF-I expression or whether hormone treatment induces nuclear protein binding to HS3D in other species.
Since the initial characterization of the CRE and the CREB/ ATF family of transcriptional regulatory proteins, an array of nucleotide sequences have been implicated in modulating gene expression through cAMP-activated signaling pathways (29, 54 -57). The consensus CRE (5Ј-TGACGTCA-3Ј) was recognized following comparison of DNA sequences from several cAMP-inducible genes (54). However, a growing list of diverse nucleotide sequences have been identified as putative CREs for various genes in a variety of cellular contexts. Greater than 60 CRE entries are present in the Wisconsin Sequence Analysis Package (GCG DNA sequence data base program), and a nucleotide sequence comparison indicates similarities between the rat IGF-I CRE we now describe and the rat osteocalcin cAMP response element (58), a CRE-like sequence present in the hepatitis B virus enhancer (59), and the consensus CCAAT box-like DNA sequence, first identified in several viral promoters and more recently in genes highly expressed in liver and in adipose tissue (56, 60 -63). Interestingly, excess doublestranded unlabeled osteocalcin cAMP response element oligonucleotide competes with labeled HS3D probe for binding to osteoblast nuclear proteins. 3 These results indicate the potential utilization of common factors for activation of IGF-I and osteocalcin genes by cAMP in osteoblasts.
One pathway for activation of gene transcription by cAMP has been described in detail. Upon binding of cAMP to the regulatory subunit of PKA, the heterotetrameric holoenzyme dissociates, and the catalytic subunit translocates to the nucleus (56,57,64,65), where it stimulates the transcriptional potential of the CREB/ATF proteins, CREB-1, ATF-1, and CREM, through reversible serine phosphorylation (56,66). These proteins typically are bound to their cognate DNA sites as homo-or heterodimers in the basal state. The function of PKA-induced phosphorylation is to activate transcription, possibly by stimulating binding between CREB/ATF proteins and a member of the CREB binding protein family (67)(68)(69). In contrast to this pathway, our results demonstrate inducible binding of nuclear factors to a CRE that comprises the HS3D site in the 5Ј-UTR of IGF-I exon 1. The identity of the factor(s) is unknown, although a CRE consensus oligonucleotide can compete for binding, suggesting that the osteoblast nuclear protein(s) may be related to the CREB/ATF family. In this regard, Nichols et al. (70) have found that an asymmetric nonconsensus CRE in the tyrosine aminotransferase gene (5Ј-CTGCCTCA-3Ј) showed inducible binding to CREB-1 and have discovered that in vitro phosphorylation of serine 133 by PKA enhanced the affinity of the CREB homodimer for this sequence, although the dissociation constant (K d ) for phosphorylated CREB-1 remained severalfold higher than the K d of CREB for a symmetric CRE. The CRE in rat IGF-I promoter 1 also is asymmetric, but it is otherwise unrelated to the tyrosine aminotransferase CRE. Experiments are now in progress to determine if a member of the CREB/ATF family and/or C/EBP family (56) binds to the newly identified IGF-I CRE. Preliminary studies indicate that antibodies to CREB-1 did not modify the gel shift seen with a labeled HS3D probe and nuclear extracts from PGE 2 -treated osteoblast cultures. 3 In summary, we have identified a mechanism through which cAMP stimulates IGF-I gene transcription in osteoblasts and have defined the nucleotide sequences responsible for mediating hormonally activated transcription through inducible nuclear protein binding. Further efforts will focus on characterizing the proteins that bind to this site and discerning the mechanisms of their regulation and action.