INTRODUCTION

We have shown previously that parathyroid hormone (PTH)¹ stimulates c-fos transcription in the osteoblastic UMR 106-01 cell line through a cAMP-mediated pathway (1). However, the events that follow cAMP induction are less clear. The present work was undertaken to describe the nuclear mechanisms involved in PTH-mediated c-fos induction in osteoblasts. Many PTH-responsive genes in osteoblasts are thought to be secondary responses due to their delayed nature and requirement for ongoing protein synthesis (2, 3). By definition, these genes require the expression of primary response genes, such as c-fos, for their induction.

Several in vivo models have identified Fos as a player in bone biology. This factor was first linked to bone when it was discovered in a mouse osteosarcoma as the product of v-fos, the viral homolog of c-fos (4). Similarly, several groups have engineered mice that overexpress c-fos and display bone abnormalities including non-malignant bone neoplasms and collagenase-producing bone tumors (5, 6). Conversely, Fos null mice exhibit osteopetrosis and disorganized bone growth (7). Transgenic mice, which express a fos-lacZ fusion gene, also identified bone as one of the major sites for c-fos expression (8). In agreement with the rodent models, evidence for Fos involvement in human bone disease has been provided by high c-fos expression in Pagetic bone (9) and human osteosarcomas (10).

c-fos is regulated in a cell-specific manner through a variety of mechanisms. These signaling pathways most likely act through different combinations of highly conserved sites within the promoter region (11). The mechanism for c-fos induction in osteoblasts by the bone resorptive agent, PTH, has not yet been determined. Observations described in this and previous studies conducted in our laboratory (1) demonstrate that PTH activates c-fos transcription independent of de novo protein synthesis and mainly through increased intracellular cAMP. We therefore hypothesized that cAMP response element-binding protein (CREB) becomes phosphorylated by PKA allowing transcriptional activation through a cAMP response element (CRE). Current information supports this model because CREB is a constitutively expressed protein, which becomes activated upon phosphorylation within its P-box domain. CREB is then able to trans-activate genes that possess the CRE sequence. The c-fos 5 regulatory region has several imperfect CRE sequences. However, only one CRE markedly causes cAMP-dependent c-fos promoter activation in Balb/3T3 cells (12). This site has therefore been referred to as the major CRE and was the focus of our experimental approach.

We present a dissection of the PTH-responsive segment within the c-fos 5 regulatory region as well as describe the transcription factor activation mechanism. Specifically, the major CRE is required for PTH-mediated c-fos induction. This site confers basal expression through unphosphorylated CREB. However, PTH causes CREB to become phosphorylated within its PKA recognition site resulting in enhanced promoter function. These observations led to the following model. In unstimulated cells, CREB participates in basal c-fos expression through the major CRE. Upon PTH exposure, PKA phosphorylates, and thereby activates, CREB trans-activation functions. Increased Fos protein then participates in gene regulation in bone.

EXPERIMENTAL PROCEDURES

Radiolabeled [¹⁴C]chloramphenicol and enhanced chemiluminescence (ECL) reagents were obtained from Amersham Corp. DuPont NEN supplied [-³²P]ATP. Polyvinylidene difluoride membrane was purchased from Millipore (Bedford, MA). Oligonucleotides were synthesized by Midland Certified Reagent Co. (Midland, TX). Poly(dI-dC) was a product of Boehringer Mannheim. Tissue culture media and reagents were obtained from the Washington University Tissue Culture Center (St. Louis, MO). All other chemicals were obtained from Sigma or Fisher Scientific.

Anti-Fos antibodies were either purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) or were a gift from Dr. Natalie Teich (Imperial Cancer Research Fund, London, United Kingdom). Anti-CREB antibody, raised against the P-box region of CREB (13), was a gift of Dr. Joel Habener (Massachusetts General Hospital and Howard Hughes Medical Institute, Harvard Medical School, Boston, MA). Antibodies to ATF-1 (hybridoma) and ATF-2 (mouse ascites) were gifts from Dr. James Hoeffler (University of Colorado Health Sciences Center, Denver, CO). Anti-phosphoCREB was purchased from Upstate Biotechnology Inc. (Lake Placid, NY). Horseradish peroxidase-conjugated goat anti-rabbit IgG was purchased from both Bio-Rad and Santa Cruz Biotechnology, Inc.

Mouse fos-CAT 5 deletion constructs 56 fos-CAT, 71 fos-CAT, 151 fos-CAT, and 356 fos-CAT (14, 15, 16) were provided by Dr. Michael Gilman (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). KCREB expression plasmid (17) was a gift from Dr. John Chrivia (Saint Louis University School of Medicine, St. Louis, MO). pSV--galactosidase control vector, pSV2CAT, and pSV0CAT plasmids were purchased from Promega Corp. (Madison, WI).

UMR 106-01 cells were cultured as described previously (18). The treatment medium contained either 2% fetal bovine serum or 0.1% bovine serum albumin.

For Fos detection, cell monolayers were washed with PBS (pH 7.6), scraped into PBS, and pelleted by centrifugation (200 × g, 10 min, room temperature). All further steps were conducted at 4 °C or on ice as rapidly as possible and in the presence of 0.2 phenylmethylsulfonyl fluoride because Fos protein is extremely labile in our hands. The cell pellets were resuspended in lysis buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Triton X-100, 0.5% sodium deoxycholate, 0.2% SDS) and sonicated. Equal amounts of total protein as determined by the Bradford (19) dye binding (Bio-Rad reagent) method were separated by SDS-polyacrylamide gel electrophoresis, electroblotted to polyvinylidene difluoride membrane, and blocked overnight in 1% bovine serum albumin, Tween-Tris-buffered saline (TTBS) (0.1% Tween 80, 138 mM NaCl, 5 mM KCl, and 25 mM Tris base). For phosphoCREB detection, cells were lysed in the culture dishes by scraping directly into sample buffer, centrifuged briefly, boiled, and loaded onto an SDS-polyacrylamide gel since phosphorylation was so labile. Antibodies were diluted in 1% TTBS. Dr. Teich's Fos antibody was diluted to 1:1000, Santa Cruz Fos antibody 1:500, Dr. Habener's CREB antibody 1:2000, and phosphoCREB 0.5 µg/ml and secondary antibody, in all cases, 1:5000. Exposure to primary antibody was for 1 h except for phosphoCREB antibody, which required only 30 min. Longer incubation with this antibody results in cross-reactivity with unphosphorylated CREB. The membranes were then washed with 0.1% TTBS three times for 5 min. Secondary antibody was applied for 30 min in all cases. Proteins were detected using ECL according to manufacturer's directions.

Cells were seeded at 10⁶ cells/100-mm diameter Petri dish and transiently transfected the following day using a calcium phosphate coprecipitation method modified from Rosenthal (20). Briefly, DNA was added to 550 µl of a 0.25 M CaCl₂ solution and mixed dropwise into 550 µl of 2 × HEPES buffered salt solution (560 mM NaCl, 100 mM HEPES, 3 mM Na₂HPO₄, pH 7.1) per dish. The solution was added to the media in the culture dishes, which were then incubated in 8% CO₂ for 4 h. Following glycerol shock, the cells were returned to maintenance medium overnight, and then treated with the appropriate agent(s). A preliminary time course indicated that maximal c-fos promoter activity was detectable after 24 h of PTH treatment. Therefore, all treatments were for this duration. To harvest, the cells were washed four times with PBS and scraped into TEN (40 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl). The cells were then pelleted by centrifugation (1000 × g, 10 min, 4 °C) and resuspended in 150 µl of 0.25 M Tris, pH 8.0. Cell lysis was achieved by three freeze-thaw cycles. Endogenous acetylases were inactivated by incubating samples at 60 °C for 10 min (except samples to be assayed for -galactosidase activity) and debris removed by centrifugation (12,000 × g, 10 min, 4 °C). The supernatant was assayed for CAT activity by a modification of the Seed and Sheen (21) procedure. Separate plates were transfected with a -galactosidase expression plasmid to verify transfection uniformity from experiment to experiment and -galactosidase activity assayed. Transfection efficiency was so uniform that CAT activities from separate, individual experiments have been pooled and are shown in many figures as mean ± S.E. Background was defined as the activity of pSV0CAT and subtracted from that of the fos-CAT constructs. The pSV2CAT plasmid was transfected into separate plates as a positive control in all experiments.

Cell lysates were prepared by washing the cell monolayer with PBS and scraping into lysis buffer (buffer C from Dignam et al. (22), plus 100 nM okadaic acid, 100 nM phenylmethylsulfonyl fluoride, and 0.5 mM dithiothreitol). Cell debris was removed by brief centrifugation. Lysates (5 µl, approximately 7 µg of protein) were incubated in final volume of 20 µl containing binding buffer (final concentrations: 4% glycerol, 1 mM MgCl₂, 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, 10 mM Tris-HCl, pH 7.5), 100 ng of poly(dI-dC), and antisera or competitor DNA at room temperature for 10 min. Double-stranded, -³²P-labeled oligonucleotide was added to the reaction immediately following the above reagents. A preliminary time course demonstrated that protein:DNA interaction occurs instanteously. Therefore, 2 µl of 10 × gel loading dye was added immediately following the probe. The reactions were then applied to a 6% polyacrylamide gel. After electrophoresis in 0.25 mM Tris, 190 mM glycine, 1 mM EDTA running buffer, the gel was dried and complexes were visualized by autoradiography. The sequences of the oligonucleotide probes were as follows, with CREs underlined and mutations in bold. The Primer Designer program was used to choose flanking sequences for the double-stranded oligonucleotides for mutagenesis.

Mutant 356 fos CAT constructs were generated with the Chameleon double-stranded, site-directed mutagenesis kit (Stratagene) according to manufacturer's instructions. Mut 1 was created using a 20-mer mutagenesis oligonucleotide complementary to the major CRE-containing region with a single point mutation: 5-TGGACTTCCTGG-3. As above, the CRE is underlined and the substituted base in bold. Mut 3 was similarly developed with the following oligonucleotide: 5-GGATGGACTTCCTGGGCGGAACTGG-3. The selection primer (Stratagene) changed the AlwNI site to NruI.

RESULTS

We have previously demonstrated that the c-fos gene transcription rate and steady state mRNA levels increase in UMR 106-01 cells after PTH treatment in a time- and dose-dependent manner (1). However, Fos protein is the final determinant in AP-1-mediated gene activation. To assess the time course of PTH-induced Fos accumulation, UMR 106-01 cells were treated for the indicated time periods with 10⁸ M PTH (Fig. 1A). Fos protein levels were noticeably elevated after 30 min, becoming maximal by 60 min. The effect was transient with Fos returning to basal levels approximately 2 h after PTH treatment. Previously, work in our laboratory revealed that c-fos mRNA accumulation was dependent on PTH concentration (1). In order to determine whether Fos protein accumulated in the same PTH dose-dependent manner, UMR cells were treated with media containing the indicated PTH concentrations (Fig. 1B) and Fos protein levels measured. Fos protein accumulation was found to be PTH dose-dependent, with an increase first observed following 10¹⁰ M PTH treatment for 1 h. Maximal stimulation was achieved with 10⁸ M PTH treatment. This relationship is in agreement with that of c-fos mRNA accumulation.

8

8

3

7

PTH increases intracellular cAMP (23) and calcium (24) as well as phosphatidylinositol hydrolysis (25) in osteoblastic cells. In order to dissect the second messenger(s) responsible for PTH-induced Fos in this cell type, UMR cells were treated with PTH, 8-Br-cAMP, or PMA, either individually or combined, as indicated in Fig. 1C. All treatments caused an increase in Fos protein. Alone or combined, 8-Br-cAMP and PTH-induced Fos to the same extent, while PMA exceeded the maximal PTH response. We also noted that either PTH or 8-Br-cAMP combined with PMA resulted in a synergistic effect on Fos protein accumulation. However, the response is in excess of the maximal PTH-induced Fos level, indicating an alternate signaling mechanism. It should be noted that we have shown previously that activation of the protein kinase C pathway is not necessary for PTH stimulation of c-fos mRNA (1). We postulate that the synergism results from the combined effect of PKA- and protein kinase C-mediated phosphate addition to a transcription factor or factors.

We have thus far made it clear that PTH increases Fos levels in UMR cells through a cAMP-dependent pathway. However, the mechanistic description is incomplete without mapping the minimal PTH-responsive region within the c-fos promoter region and identifying the cis- and trans-acting elements responsible for transcriptional activation. To this end, mouse c-fos promoter 5 deletion constructs 356, 151, 71, and 56 fos-CAT (Fig. 2A) (14, 15, 16) were transiently transfected into UMR cells. These constructs include several highly conserved regions, which respond to different agonists in other cell types. Upon PTH treatment, 356, 151, and 71 were stimulated 2.6-, 2.8-, and 4.6-fold, respectively (Fig. 2B). Conversely, 56 fos-CAT, which is deleted 3 to the main CRE was unaffected by PTH treatment. It is worth emphasizing that the shortest of the three CRE-containing constructs (71 fos-CAT) gave the highest fold stimulation although basal expression sharply decreased (Fig. 2B, inset). However, there was no difference between PTH induction of 151 and 356 fos-CAT. Human c-fos promoter CAT constructs responded similarly to PTH when transfected into UMR cells (data not shown). PTH dose-response experiments showed that 71 fos-CAT displayed a PTH concentration dependence similar to the endogenous gene (Fig. 2C). Increased promoter activity was first detected following 10¹⁰ M PTH treatment. Maximal stimulation was detected after 10⁸ M PTH, and half-maximal response was calculated to occur with 5.5 × 10¹⁰ M PTH.

PTH activation of mouse c-fos-CAT 5 promoter deletion constructs.

8

8

3

7

Elegant experiments utilizing mice that express a fos-lacZ fusion gene established that different combinations of regulatory elements interact to control c-fos in different systems (8). In order to determine if the same element and second messenger pathway activate 71 as do 356 fos-CAT, these constructs were transiently transfected into UMR 106-01 cells followed by treatment with PTH, 8-Br-cAMP, and PMA either alone or in combination (Fig. 2D). An identical pattern would signify that both constructs activate transcription through the same region. The response pattern was similar for both enhancer regions. Like the Fos protein measurements, the PTH effect was mimicked by 8-Br-cAMP with 71 fos-CAT producing the highest fold stimulation, although lowest basal expression. PTH in combination with 8-Br-cAMP activated both constructs to nearly the same extent as either agent alone, suggesting a common signaling pathway. While 356 fos-CAT was slightly responsive to PMA, this agent alone had no effect on 71 fos-CAT. However, analogous to Fos protein measurements, PMA, when combined with either PTH or 8-Br-cAMP, increased CAT activity synergistically. This synergism was not abolished by pretreating with either PMA or myristoylated protein kinase C inhibitor peptide, possibly due to the extended length of time that transfected cells must be treated with the agents (data not shown). However, as stated above, we have previously ruled out diacylglycerol-inducible protein kinase C as a participant in PTH-induced c-fos expression (1).

In order to more directly assess the impact of the c-fos major CRE, a single point mutation was created in this element within 356 fos-CAT. Shown as Mut 1 in Fig. 3A, this base pair change has been shown previously to inhibit CREB interaction with a consensus CRE (26). The point mutation resulted in near abolition of the PTH effect on the promoter (from 3.2- to 1.7-fold) (Fig. 3B). However, with a single point mutation, we speculated that the remaining PTH-induced activity could be mediated through residual protein:CRE interaction. We therefore substituted 3 bases within the major CRE of 356 fos-CAT, hence creating Mut 3. As with Mut 1, Mut 3 transfection into UMR 106-01 cells reduced fold stimulation in response to PTH to 1.7-fold. However, this mutation drastically reduced basal expression (Fig. 3B, inset). Mut 3 had lower basal activity than Mut 1, illustrating that the more bases altered the less basal activity remains.

PTH activation of 356 fos-CAT with mutations in the major CRE.

8

Having defined the major CRE as necessary for PTH-stimulated c-fos activation, gel retardation techniques were employed to assess the interaction of proteins from UMR 106-01 cells at this site. Proteins in cell lysates from both control and PTH-treated cells bound the mouse major CRE (TGACGTAG), creating an identical shift pattern (Fig. 4). Titration with cold probe demonstrated that a 5-fold molar excess efficiently competed for all protein-DNA complexes, while 25-fold excess oligonucleotide containing a CTF/NF1 element did not. This titration also revealed that the lower shifted band is less efficiently competed for by cold probe, indicating a lower relative affinity than the upper bands. Other experiments revealed that other non-CRE elements (AP-1, GRE, SP-1) were unable to compete (data not shown) for binding.

8

The unaltered shift pattern following PTH treatment is in agreement with current thinking that CREB binds to the consensus CRE with equal affinity regardless of its phosphorylation state but activates transcription only when phosphorylated (27, 28). To determine if our system fits this model, we included an antibody that interacts with CREB only if the latter is phosphorylated at Ser¹³³ within the PKA consensus phosphorylation site. This antibody supershifted the upper bands only when protein was derived from PTH-treated cells (Fig. 5A). A nondiscriminating anti-CREB antibody supershifted these bands independent of PTH treatment (data not shown). Likewise, phosphorylation without change in CREB protein abundance was confirmed by Western blot. In this experiment, PTH did not alter the total CREB signal as visualized with CREB antibody while phosphoCREB was detected only in PTH-treated cells (Fig. 5B). The band running below the CREB containing complex in gel shift experiments appears to be ATF-1 since it was abolished by antibody to ATF-1. c-Jun antibody had no effect on any protein-DNA complex.

8

8

During the course of this study, the rat c-fos promoter region was published (11) and shown to be very similar to the human and mouse promoters. Rat c-fos bears the same nonconsensus major CRE as mouse but with different flanking sequences. Because UMR 106-01 cells are rat-derived, we assayed UMR 106-01 proteins binding to rat fos CRE in its native context. Similar to the mouse sequence, this element bound protein from both control- and PTH-treated UMR 106-01 cells equally, creating a shifted complex that could be supershifted by CREB antibody (data not shown).

Gel mobility shift experiments, which included a mutant CRE-containing oligonucleotide probe, confirmed the loss of CREB-containing bands (Fig. 6). The mutations were those incorporated into Mut 1 and Mut 3. This experiment confirmed our speculation that a single point mutation within the major CRE allows residual protein binding as determined by overexposing the film (data not shown). Residual CREB interaction with the mutant CREs was inversely proportional to basal expression of Mut 1 and Mut 3 constructs. Conversely, ATF-1 binding was not altered.

8

KCREB (killer CREB) is a dominant inhibitor of CREB function (17). The protein product has an amino acid substitution within the leucine zipper, which results in KCREB-CREB heterodimers with no DNA binding function. We cotransfected a KCREB expression plasmid along with the mouse fos-CAT constructs into UMR 106-01 cells and treated the cells with control- or PTH-containing media. KCREB inhibited PTH-mediated activation of both 71 and 356 fos-CAT promoter constructs by 60% (Fig. 7). Conversely, expression of 56 fos-CAT, which lacks the major CRE as well as PTH responsiveness, was unaffected by KCREB (data not shown).

8

DISCUSSION

The data presented in this work identify three steps in the mechanism through which PTH induces c-fos gene expression in osteoblastic cells. (i) PTH induces c-fos expression using cAMP as a second messenger. (ii) Activated PKA catalytic subunit phosphorylates CREB at Ser¹³³ within the P-box. (iii) Phosphorylated CREB bound to the major CRE in the c-fos 5 regulatory region results in transcriptional activation.

Despite causing increased intracellular calcium and phosphoinositide turnover, PTH induces many osteoblastic responses through cAMP-dependent mechanisms (3). Accordingly, our observation that a cAMP analog mimics the effect of PTH on both c-fos promoter activation and Fos protein accumulation is in agreement with this pattern. In addition, failure of maximal concentrations of 8-Br-cAMP and PTH in combination to significantly increase their effect on c-fos further supports this interpretation. This is because each agent maximally activates the same pathway making additional c-fos expression through this mechanism impossible. Although PMA induced Fos protein both alone and combined with PTH or 8-Br-cAMP, the response was always in excess of the maximal PTH induction. This point, coupled with convincing experiments performed by Clohisy et al. (1), refutes any protein kinase-C involvement in the PTH-induced transduction of c-fos. Therefore, all available evidence identifies cAMP as the major functional second messenger in this and many other PTH-mediated responses in osteoblasts.

We have previously defined PTH-induced c-fos expression as a primary response (1). De novo synthesized transcriptional activators are, by definition, not participants in this mechanism. Therefore, post-translational modification must be involved. Partridge et al. (29) have demonstrated PKA activation in response to PTH. These findings combined with the current data are in agreement with the classical CREB-mediated pathway. Specifically, constitutively expressed CREB is phosphorylated at Ser¹³³ by PKA. CREB is then activated, and as a dimer bound to a CRE, can activate gene transcription (for review, see Ref. 30). Our observations that CREB is necessary for the full PTH response coupled with PTH-induced PKA activity and CREB phosphorylation at the PKA consensus site are in complete support of this mechanism. It should be noted that Ca²⁺/calmodulin-dependent protein kinases types II and IV have also been shown to phosphorylate CREB at Ser¹³³. However, we have clearly eliminated calcium involvement in induction of c-fos by PTH (1).

While Ser¹³³ phosphorylation is required for CREB activity, additional kinases may also participate in CREB activation (28). Accordingly, PKA-mediated phosphorylation is required to create a casein kinase II consensus site, which, upon phosphorylation, activates CREB in vitro (31). The latter and similar observations involving other kinases have resulted in a hierarchical phosphorylation model (32). This states that, although CREB remains inactive without phosphorylation at the PKA site, the differential action of kinases on additional residues is required for its full activation and allows for cell-specific regulation. In this mechanism, Ser¹³³ phosphorylation may cause a conformational change in the CREB protein, making other phosphorylated residues available for protein-protein interaction or exposing other residues to kinases. Work is currently in progress to identify any additional phosphorylation events that may be part of the PTH-mediated signal transduction pathway in osteoblasts. In fact, we postulate that this mechanism causes the synergism that we observed between PMA and 8-Br-cAMP.

Proteins other than CREB are able to interact with the CRE sequence. However, this report clearly establishes CREB as one, if not the only, of these factors active and required at the c-fos major CRE. Gel shift assays combined with transfected fos-CAT constructs containing mutant CREs provide the most direct proof. Although ATF-1 interaction with the major CRE was visualized by gel mobility shift, we reject this factor as a candidate for four reasons. (i) ATF-1 binding was not altered by CRE mutations that inhibited PTH induction of 356 c-fos-CAT. (ii) Although the P-CREB antibody cross-reacts with a homologous site on ATF-1, we did not see ATF-1 phosphorylation. (iii) ATF-1 bound the major CRE with a lower affinity than CREB. (iv) No CREB-ATF-1 dimers bound to the major CRE.

CREB activity at the major CRE is in agreement with earlier c-fos promoter characterization, which assigns most of the cAMP-mediated c-fos induction to this element (12). Similar responses of 71 and 356 fos-CAT refute the possibility of other PTH responsive elements 5 to the major CRE. Basal activity accompanying induced promoter activation is also characteristic of the CRE. For example, CREB has been shown to enhance both basal and glucocorticoid-induced phosphoenolpyruvate carboxykinase gene expression through a 5 CRE (33).

In summary, we demonstrate that PTH induces c-fos transcription primarily through the cAMP/PKA pathway. This event requires that CREB becomes phosphorylated by PKA at Ser¹³³ and acts at the major CRE within the c-fos promoter region. This is the first demonstration that PTH induces gene transcription through phosphorylation of a constitutively expressed transcription factor and completes the signaling pathway from the cell membrane to the nucleus.