Signaling in Human Osteoblasts by Extracellular Nucleotides

Extracellular nucleotides acting through specific P2 receptors activate intracellular signaling cascades. Consistent with the expression of G protein-coupled P2Y receptors in skeletal tissue, the human osteosarcoma cell line SaOS-2 and primary osteoblasts express P2Y1 and P2Y2 receptors, respectively. Their activation by nucleotide agonists (ADP and ATP for P2Y1; ATP and UTP for P2Y2) elevates [Ca2+] i and moderately induces expression of the c-fos proto-oncogene. A synergistic effect on c-fos induction is observed by combining ATP and parathyroid hormone, a key bone cell regulator. Parathyroid hormone elevates intracellular cAMP levels and correspondingly activates a stably integrated reporter gene driven by the Ca2+/cAMP-responsive element of the human c-fospromoter. Nucleotides have little effect on either cAMP levels or this reporter, instead activating luciferase controlled by the full c-fos promoter. This induction is reproduced by a stably integrated serum response element reporter independently of mitogen-activated protein kinase activation and ternary complex factor phosphorylation. This novel example of synergy between the cAMP-dependent protein kinase/CaCRE signaling module and a non-mitogen-activated protein kinase/ternary complex factor pathway that targets the serum response element shows that extracellular ATP, via P2Y receptors, can potentiate strong responses to ubiquitous growth and differentiative factors.

Extracellular stimuli regulate gene expression via the activation of intracellular signaling systems that transduce the signal from membrane-bound receptors to transcription factors. This results in the rapid, transient induction of genes at the transcriptional level whose products will mediate the appropriate cellular response. One of the best characterized of these immediate early genes is the c-fos proto-oncogene, whose activation can be linked to different signaling cascades target-ing distinct promoter elements through the phosphorylation of transcription factors (1,2).
In the c-fos promoter, the calcium/cAMP-responsive element (CaCRE) 1 (1), located just upstream of the TATA box, can mediate activation following elevation of intracellular cAMP, as well as Ca 2ϩ in certain instances (3,4). Proteins of the CREB/activating transcription factor family recognize this element, and transcriptional induction arises via their phosphorylation on Ser 133 (5). The serum response element (SRE), at position Ϫ300 in the promoter, mediates induction by many extracellular signals via a ternary complex composed of a dimer of serum response factor (SRF) together with one molecule of the TCF family of Ets proteins (Elk-1, SAP-1a, or ERP/NET/ SAP-2 (6 -10)). TCFs, particularly Elk-1 and SAP-1a, are important nuclear targets of various MAPK cascades (11), while SRF is apparently sufficient for activation by certain Ca 2ϩ signals and signals emanating from the Rho/Rac/CDC42 family of small GTPases (12)(13)(14). Upstream of the SRE is the v-sisinducible element, the binding site for homo-and heterodimers of signal transducer and activator of transcription 1 and 3 upon their cytoplasmic activation by cytokines and certain growth factors (15,16). While transient transfections have proven useful in attributing a role to each element, the results from mice containing c-fos transgenes (17) and more recent data in vitro indicate that multiple elements are necessary for a strong response (18), 2 thereby implying that they are targeted simultaneously by intracellular signals. c-fos induction plays an important role in vitro in driving immortalized fibroblasts to enter the cell cycle and plays an important role in vivo in the skeletal system. Mice lacking the c-fos gene fail to develop osteoclasts and thus show an osteopetrotic phenotype in which the dynamic process of bone remodeling has been shifted toward bone accumulation (19,20). Conversely, constitutive overexpression of c-fos in the bone environment of transgenic mice leads to the development of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Parathyroid hormone (PTH) is essential for the modeling and remodeling of the skeleton. In bone-derived cell lines as well as in primary osteoblasts in culture, PTH strongly induces transcription of the c-fos gene (23,24). This occurs via the cAMP/ cAMP-dependent protein kinase/CREB pathway in the osteosarcoma cell line SaOS-2 (25) but may involve protein kinase C and the ERK pathway in other cells (2). Because PTH is a systemic factor and the process of bone remodeling is essentially a focal phenomenon, cellular responsiveness to PTH in vivo is likely to be modulated by other factors. We wondered whether ATP might play such a modulatory role, since ATP can be released from osteoblasts into the local bone microenvironment via a nonlytic mechanism (26) and since ATP synergizes with mitogens to enhance DNA synthesis in a variety of cells (27)(28)(29).
Extracellular nucleotides, such as ATP, exert stimulatory effects on cells at micromolar concentrations through the P2 family of membrane-bound receptors (30 -33). Two major classes of P2 receptors have been delineated: P2X receptors, which are ligand-gated ion channels, and P2Y receptors, which are coupled to G proteins (32). More pertinent to our hypothesis, P2Y receptors are expressed in osteoblastic cells of rat (34,35) and human origin (36,37). Two major subtypes of P2Y receptor, P2Y 1 and P2Y 2 , are coupled through G q to phosphatidylinositol 4,5-bisphosphate hydrolysis and hence Ca 2ϩ mobilization from intracellular stores (32).
Here we show that both the osteosarcoma cell line SaOS-2 and primary cells in culture express P2Y receptors that functionally couple to c-fos activation. This involves increased intracellular Ca 2ϩ and a signaling pathway that can activate an SRE-driven reporter gene independently of the predominant ERK/TCF signaling module. Co-activation of this pathway and that induced by PTH increases c-fos mRNA levels well above those induced by either stimulus alone, thereby providing a novel example of synergy between a cAMP-dependent protein kinase and a Ca 2ϩ -triggered signaling system not involving MAPK. These data demonstrate that extracellular nucleotides can strongly potentiate the response of bone cells to systemic factors and suggest that this may be a common mechanism to generate strong localized responses to systemic growth and differentiation factors.
Cell Culture-Human bone-derived cells (HBDC) were isolated and cultured from explants of human bone as described previously (38). In brief, specimens of human bone in Dulbecco's modified Eagle's medium were finely minced with a scalpel and then washed free of marrow cells with several volumes of medium. The minced bone was cultured in 9-cm Petri dishes containing Dulbecco's modified Eagle's medium supplemented with 10% FCS, 100 g/ml streptomycin, 100 units/ml penicillin, and 2 mM L-glutamine. The cells were maintained at 37°C in a humidified atmosphere of 95% air and 5% CO 2 for 3 weeks. Confluent cells were either seeded onto glass coverslips or induced for subsequent preparation of whole cell extracts or RNA. SaOS-2 cells were maintained in ␣-modified Eagle's medium supplemented as above, and seeded onto glass coverslips or harvested upon confluence.
Construction of c-fos-luciferase Reporter Genes for Stable Transfection-The fragment containing the c-fos promoter (spanning positions Ϫ721 to Ϫ1, accession number M16287) linked to the luciferase gene was subcloned from pUC19fosluc1 (kindly provided by Dr. L. Runkel) into pSV2neo, to create pfoslucneo1. A multimerized SRE-luciferase reporter gene vector was obtained by ligating the oligonucleotide CCG-CAGGATGTCCATATTAGGACATCTGTGTGCCGTCCTACAGGTATA-ATCCTGTAGACACGCCGG that was used to replace the promoter upstream of the c-fos TATA box in the same vector. Clones containing three copies of the SRE were identified and verified by sequencing. The vector containing a single copy of the c-fos CaCRE was generated by deleting sequences upstream of position Ϫ88 in the fos promoter in pfoslucneo1. Its identity was verified by sequencing.
Isolation and Culture of Stable Cell Lines Incorporating the c-fosluciferase Reporter Gene-SaOS-2 cells (ATCC HTB 85) were maintained in RPMI 1640 medium supplemented with 10% (v/v) FCS. These cells were transfected with pfoslucneo1 or CaCRE-luc using Lipofectin reagent according to the protocols of the manufacturer, and the cells were maintained in RPMI/FCS for 24 h. Stably transfected cell pools were selected for resistance to G418 following standard protocols (39). UMR-106 cells were cultured in Ham's F-12 and Dulbecco's modified Eagle's medium (1:1) supplemented with 10% FCS at 37°C in a humidified atmosphere containing 95% air and 5% CO 2 . Stable pools of UMR-106 cells containing the SRE-luciferase reporter were generated as described above.
Luciferase Reporter Gene Assays-For reporter gene assays, the cells were seeded into 96-well plates at a density of 96,000 cells/well. At near confluence, the medium was replaced with one containing 0.5% FCS for 16 h, followed by serum-free RPMI containing 0.1% BSA for 24 h. Agonists were added to cells as a 10ϫ stock solution prepared in the same medium to the final concentrations indicated in the figures. After a 4-h incubation, cells were washed twice in cold PBS and lysed in luciferase cell culture lysis reagent for 15 min at room temperature (25 l/96-well plate). The plates were subsequently centrifuged at 3000 rpm for 5 min at room temperature before transferring 20 l of each sample to a 96-well microtiter plate (Dynatech) for analysis. A microtiter plate luminometer (ML 3000, Dynatech Laboratories) was set up in the enhanced flash mode (maximum sensitivity, delay time ϭ 2 s, integration time ϭ 10 s). Luciferase reagent (100 l) was automatically added, and light emission was measured every 10 ms during a period of 10 s. Data were recorded as the peak value of relative light units.
Measurement of [Ca 2ϩ ] i -SaOS-2 cells and HBDC were grown to confluence on 22-mm diameter glass coverslips. [Ca 2ϩ ] i was measured after 2 h of serum deprivation. Cells were loaded with fura-2 by incubation with fura-2 acetoxymethyl ester (5 mM) for 20 min at 37°C in HEPES buffer (10 mM HEPES, 121 mM NaCl, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 2 mM CaCl 2 , 5 mM NaHCO 3 , 10 mM glucose, pH 7.2) containing 2% BSA. Cells were subsequently washed three times in the same buffer containing 0.2% BSA. Measurements were performed with a photon-counting spectrophotometer on a Nikon TM Diaphot microscope with a ϫ 40 oil immersion objective. The cell-coated coverslip was attached with silicone grease to form the base of a stagemounted, thermostatically regulated chamber maintained at 37°C. Groups of 6 -8 cells were illuminated with excitation light (340 and 380 nm) at a rate of 32 times/s, and the emission measurements (at 510 nm) were integrated into 1-s averages and stored. Agonists, in HEPES buffer with 0.2% BSA, were added for 60 -120 s, followed by at least 10 min of recovery prior to further stimulation. R min , R max , and autofluorescence values were obtained in situ using ionomycin, as described previously (40). [Ca 2ϩ ] i was calculated from the ratio of fluorescence at the two excitation wavelengths after subtraction of autofluorescence (41). The results were evaluated statistically using the Student's t test, assuming a significance of p Ͻ 0.05.
RNA Isolation and cDNA Synthesis-Total RNA was extracted from control and stimulated cells with 4 M guanidine thiocyanate, 0.5% sarkosyl, 0.1 M mercaptoethanol, 25 mM sodium citrate, pH 7.0, followed by acid phenol/chloroform extraction (39). RNA was stored in ethanol at Ϫ20°C. Prior to first strand cDNA synthesis, RNA was DNase-treated with RNase-free DNase I (35 units/ml). 5 g of DNase-treated total RNA was used as template for first strand cDNA synthesis in a 50-l reaction containing 0.5 mM dNTPs, 1.25 g of oligo(dT), 20 units of RNase inhibitor, 10 mM dithiothreitol, 6 mM MgCl 2 , 40 mM KCl, 50 mM Tris-HCl, pH 8.3, and 1000 units of Moloney murine leukemia virus reverse transcriptase. After 1 h at 37°C, the reaction was frozen at Ϫ20°C.
Northern Analysis-5-10 g of total RNA was denatured and electrophoresed through a 0.8% (w/v) agarose gel containing 3.7% formaldehyde (v/v), followed by transfer to Zetabind hybridization membrane as described by Evans et al. (25). Blots were prehybridized at 42°C in 40% formamide, 5ϫ SSC, 10ϫ Denhardt's reagent, 1% SDS, 200 g/ml denatured salmon sperm DNA, 200 g of tRNA and probed with a 487-base pair fragment spanning exons 3 and 4 of the human c-fos gene labeled with [␣-32 P]dCTP (3000 Ci/mmol) by random priming. Membranes were washed for 30 min at 65°C with 0.2ϫ SSC, 1% SDS solution, and mRNAs were visualized using phosphor storage technology and autoradiography with Kodak XAR film and intensifying screens at Ϫ70°C. RNA loading and integrity were followed by ethidium bromide staining or hybridization with a random primed fragment purified from a cDNA clone containing rat GAPDH. The blot in Fig. 6 was hybridized with c-fos and GAPDH riboprobes in the same mix as described previously (42).
cAMP Measurements-SaOS-2 cells were serum-starved as above and then pretreated with 0.1 mM 3-isobutyl-1-methylxanthine for 30 min at 37°C. Agonist was then added as described in the legend to Fig.  7. After a 20-min incubation, the cells were covered with 1 ml of ice-cold 65% ethanol and scraped off with a rubber policeman. Cells were collected by centrifugation at 8000 ϫ g for 3 min and stored at Ϫ20°C until cAMP levels were determined using a radioimmunoassay (43). The statistical significance was assessed by a one-way analysis of variance followed by the Tukey-Kramer multiple comparisons post-test.
Gel Retardation Assay-Reactions (7.5 l) contained the following components: 2.5 g of poly(dI-dC)(dI-dC), 250 ng of calf thymus DNA, 5% (v/v) glycerol, 66 mM NaCl, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.35% (v/v) Triton X-100, 0.05% (w/v) low fat milk, 15 mM dithiothreitol, 15,000 cpm of 32 P-labeled probe (0.2 ng/4 fmol), and 10 g of whole cell extract. After a 30-min incubation at room temperature, the entire reaction was loaded on a 4.5% polyacrylamide gel containing 45 mM Tris borate, 1.5 mM EDTA (pH 8.3) and run at 1 mA/cm for 3-4 h. Gels were dried, and the complexes were visualized by autoradiography using intensifying screens or phosphor storage technology. Core SRF 90 -244 was produced in HeLa cells using a recombinant vaccinia virus (8). The probe corresponded to the c-fos SRE (see above) subcloned in front of a G-free cassette plasmid (44). After EcoRI and NarI digestion, the ends were labeled by a Klenow fill-in reaction containing [␣-32 P]dATP and cold dCTP, dGTP, and dTTP. Fragments were isolated from polyacrylamide gels by electroelution.

Differential Expression of P2Y 1 and P2Y 2 Receptors by
SaOS-2 and Primary Human Bone-derived Cells-Since osteoblasts have been described to express different purinergic receptors of the P2Y family, we used RT-PCR to analyze which subtypes are expressed in the human osteosarcoma cell line SaOS-2. In addition, we tested two populations of primary osteoblastic cells derived from explants of human bone in vitro culture, termed HBDC1 and HBDC2. RT-PCR on cDNA from SaOS-2 cells gave rise to an intense signal for P2Y 1 receptor transcripts but only a weak signal for P2Y 2 receptor (Fig. 1). In contrast, only P2Y 2 receptor transcripts were visualized in the HBDC1 cDNA, while amplification of HBDC2-derived cDNA showed similar levels of amplification for both P2Y 1 and P2Y 2 . The signal with primers specific for GAPDH was similar in all reactions (Fig. 1), confirming the integrity and amount of cDNA in each sample.

Elevation of [Ca 2ϩ ] i in SaOS-2 Cells and Primary Human Osteoblasts following P2 Receptor Stimulation-P2Y
receptors are coupled to heteromeric G proteins intracellularly, and have been reported to activate, via G q , phosphatidylinositol 4,5bisphosphate hydrolysis and Ca 2ϩ mobilization from intracellular stores (32). Since SaOS-2 cells and HBDC1 express predominantly the P2Y 1 and P2Y 2 receptors, respectively, we tested whether different P2Y 1 and P2Y 2 receptor agonists could mobilize intracellular Ca 2ϩ , measured using fluorescence increases in groups of 6 -8 fura-2-loaded cells. ATP consistently induced a rise in [Ca 2ϩ ] i in both cell types ( Fig. 2A and Table I).
In SaOS-2 cells, ATP was effective at concentrations ranging from 1 to 100 M ( Fig. 2A). In these cells, the P2Y 2 agonist UTP evoked only a minor increase in [Ca 2ϩ ] i at 10 M and a more significant increase at 100 M (Fig. 2B) that nevertheless was smaller than that induced by 1 M ATP (Fig. 2, compare A and  B). This difference in responsiveness was also observed upon the sequential addition of UTP and ATP, thus indicating that it did not reflect decreased sensitivity of the cells. More importantly, UTP induced the same increase as ATP in HBDC1 ( Fig.  3 and Table I), clearly showing that the difference is due to the P2Y receptor subtype expressed by the cells.
The effects of other agonists confirmed this differential responsiveness between the primary cells and the established cell line. In SaOS-2 cells, the ATP analogues ATP␥S and 2-me-SATP, as well as ADP (all 10 M), induced [Ca 2ϩ ] i increases comparable with ATP (Table I). The primary cells did not show the same behavior, since these P2Y 1 agonists did not elevate [Ca 2ϩ ] i in HBDC1 (Fig. 3). These differences in the functional response to extracellular nucleotides are consistent with and thereby confirm the receptor profiles obtained by RT-PCR (Fig. 1).
Induction of c-fos mRNA in SaOS-2 Cells following P2 Receptor Stimulation-Since nucleotide addition elevated intracellular calcium levels, we tested whether this second messenger pathway activated nuclear signaling, as measured by the induction of the proto-oncogene c-fos. This event is particularly relevant in bone-derived cells, since c-fos has been implicated in many of the processes that govern skeletal tissue remodeling (21,22).
ATP␥S stimulation of quiescent SaOS-2 cells led to a dosedependent induction of c-fos mRNA, measured by Northern blotting (Fig. 4A), which was maximal at 10 M ATP␥S. This represented a typical transient activation in which fos mRNA levels peaked 45 min after stimulation and then rapidly decayed (Fig. 4B). In addition, we tested the same range of P2Y receptor agonists used to analyze Ca 2ϩ mobilization. 10 M ATP␥S, ATP, and ADP stimulated c-fos expression, whereas UTP and 2-meSATP did not (Fig. 4C). While this generally reflects the activation of the P2Y 1 receptor, this correlation is not universal, as shown by the lack of induction by 2-meSATP and the decrease between 10 and 100 M ATP␥S. This was not due to differing levels of RNA, since rehybridization of the blots with a GAPDH probe and/or ethidium bromide staining confirmed equal loading of mRNA in all lanes. This apparent discrepancy between increased [Ca 2ϩ ] i and c-fos induction will be discussed in more detail below.
Synergistic Induction of c-fos by the Combination of P2 Re-FIG. 1. Differential P2 receptor expression by SaOS-2 and primary human bone-derived cells. PCR amplification of cDNAs with P2Y 1 , P2Y 2 , or GAPDH primer pairs. cDNAs were synthesized from RNA templates prepared from SaOS-2 cells, two different primary osteoblast populations (HBDC1 and -2), and an osteoclastoma tumor. The tumor expresses both receptor subtypes and thus serves as the positive control (37). As indicated on the left, the bands of 259 and 362 base pairs (bp) correspond to amplification products specific for P2Y 1 and P2Y 2 receptor cDNAs, respectively, and GAPDH yields a larger 519-base pair band. The control reaction contained H 2 O instead of template cDNA, and the unmarked lanes contain molecular size markers. ceptor Agonists and PTH in SaOS-2 Cells-While P2Y 1 receptor agonists induced c-fos, the level of expression was low relative to serum (Fig. 4C). Since nucleotides enhance the proliferative response to mitogens in other cell types (27,28), we tested whether they might enhance c-fos induction in SaOS-2 cells by PTH, a potent stimulator of bone growth in vivo. Quiescent SaOS-2 cells were treated with different nucleotides alone or in combination with PTH and mRNA levels analyzed by Northern blotting. Cotreatment with PTH and either ATP or ADP resulted in a synergistic induction of c-fos mRNA relative to either inducer alone (Fig. 5A). In contrast, UTP did not augment PTH-induced c-fos activation in SaOS-2 cells, which is again consistent with its inability to stimulate the P2Y 1 receptor.
To confirm that this synergy arose from direct transcriptional activation and facilitate further characterization of this effect, we transfected SaOS-2 cells with a reporter gene in which the full c-fos promoter, spanning positions Ϫ711 to Ϫ1, was linked to the firefly luciferase gene. To avoid the problems inherent in assaying signaling by transient transfection of nonphysiological quantities of DNA, we selected pools of stably transfected cells using the neomycin resistance gene present in the same construct. Both serum and PTH induced a robust response of this reporter construct, elevating luciferase activity 15-20-fold in the stable transfectants (Fig. 5B). ATP showed only a weak effect on its own, consistently severalfold above the background level. However, the combination of ATP and PTH cooperatively induced the c-fos reporter, which is in direct contrast to the failure of UTP to significantly augment luciferase activity in combination with PTH (Fig. 5B). Thus, we are able to reproduce the synergy between ATP and PTH using the c-fos promoter in stable transfection assays, which indicates that this effect arises primarily from increased transcriptional activation and not from stabilization of the fos mRNA. Furthermore, it suggests that ATP and PTH activate intracellular signals that target regulatory elements in the c-fos promoter.
Synergistic Induction of c-fos by the Combination of P2 Receptor Agonists and PTH in HBDC-To determine whether extracellular nucleotides might also potentiate c-fos activation by PTH in primary osteoblasts, we treated HBDC with PTH and a range of P2Y receptor agonists. As above, ATP cooperated with PTH to strongly enhance c-fos mRNA levels above those induced by either factor alone (Fig. 6). In these cells, the P2Y 2 receptor agonist also synergized with PTH, whereas the P2Y 1 agonist ADP was inactive (Fig. 6). Thus, the primary cell population shows a response consistent with their receptor subtype and that contrasts with the response of the P2Y 1 / SaOS-2 combination to the same nucleotides. (32); therefore, we tested if the synergy between ATP and PTH resulted from enhanced Ca 2ϩ mobilization. In neither SaOS-2 cells nor the primary osteoblasts did PTH, at 20 -500 ng/ml, generate a significant calcium response, nor did we ever observe that PTH enhanced the effect of either ATP or UTP on [Ca 2ϩ ] i . 3 In SaOS-2 cells, PTH-(1-34) induces c-fos expression through cAMP-dependent phosphorylation of CREB independently of protein kinase C activation, TCF phosphorylation, or signal transducer and activator of transcription induction (25). Therefore, we tested if the synergy on c-fos expression following nucleotide/PTH stimulation might reflect modulation of cAMP levels. A radioimmunoassay was used to assess the effects of nucleotide stimulation, alone and in combination with PTH, on cAMP accumulation in SaOS-2 cells. Forskolin, a potent activator of adenylyl cyclase at 10 g/ml, elevated intracellular cAMP to 170 pmol/ml above the basal level. On the other hand, the addition of 10 M ATP, ATP␥S, UTP, or ADP had no significant effect on cAMP levels either alone (Fig. 7A) or in combination with PTH (Fig. 7B).

PTH and ATP Utilize Distinct Second Messengers in Human Osteoblasts-In clonal rat osteoblastic cells, ATP increases PTH-induced [Ca 2ϩ ] i responses
The CaCRE in the c-fos Promoter Is Insufficient for a Synergistic Response-The CaCRE in the c-fos promoter, which is situated immediately upstream of the TATA box, can mediate activation of reporter genes by certain Ca 2ϩ signaling pathways (see Introduction). We used stably transfected SaOS-2 cell lines to test the possibility that the ATP/Ca 2ϩ and PTH/ cAMP pathways converge on this site to synergistically induce c-fos. A reporter driven by a truncated promoter, containing only the CaCRE linked to luciferase, was compared with the full-length reporter construct. Co-stimulation of the stable pools containing the truncated reporter with PTH and nucleotides resulted in luciferase expression levels slightly above those observed with PTH alone (Table II). Notably, this increase reflected an additive effect of ATP and PTH rather than the multiplicative effect seen on the endogenous gene and the full promoter-driven reporter. This strongly suggested that 3 Table I presents the statistics concerning these measurements.
synergistic activation did not arise by the convergence of these two signaling systems on the CaCRE alone.
ATP Activates the c-fos SRE Independently of ERK and TCF-These data implied that the ATP/Ca 2ϩ signaling pathway targeted another site in the c-fos promoter. The SRE seemed the most likely candidate, since it can also mediate activation by various calcium-dependent pathways. We again chose the stable transfection approach. Unfortunately, SaOS-2 cells have proven refractory to stable transfection with SREdriven reporter constructs, so we resorted to another bonederived cell line, UMR-106, that shows the same responses as SaOS-2 cells to ATP and PTH. 4 Pools of UMR-106 cells were generated that contain a reporter construct in which the luciferase gene is controlled by three c-fos SREs cloned in front of the c-fos TATA box. These cells responded strongly to EGF and slightly less so to serum, which increased luciferase activity 12and 6-fold, respectively (Table II) stimulate luciferase at all, while ATP caused a 4.6-fold increase in luciferase activity. Thus, the SRE can mediate the response to ATP-driven signals and is likely to be the promoter element targeted by the ATP/Ca 2ϩ pathway in bone cells.
The SRE integrates signals from many pathways but is particularly responsive to activated MAPK cascades via the phosphorylation of TCF, a major nuclear target for the MAPKs ERK, SAPK/c-Jun N-terminal kinase, and p38 MAPK (2). We wondered whether this synergy reflected the activation of one of these cascades by nucleotides, especially since other bone growth factors can induce ERK activity (45)(46)(47). To test for MAPK activation, we used antisera directed against the activated kinases, which are highly specific for the molecules phosphorylated on Thr and Tyr in their Thr-Xaa-Tyr activation motif. Western blots of SaOS-2 whole cell extracts, probed with anti-Thr(P) 202 -Tyr(P) 204 ERK antisera, revealed that ATP did not activate ERK on its own or together with PTH (Fig. 8). In contrast, EGF gave rise to a robust activation of ERK, as described previously in osteoblasts (48), which was partially diminished by coinduction with PTH (Fig. 8). UMR-106 cells show similar behavior (49), which may represent another example of antagonism between cAMP-dependent protein kinase and the ERK cascade apparent in some cell contexts (49 -52). Thus, the synergy between ATP and PTH does not involve the ERK pathway.
We used the same strategy to test whether ATP activated the stress-responsive MAPKs, namely SAPK/c-Jun N-terminal kinase and p38 MAPK, in these cells. Neither kinase was induced by ATP (Fig. 8), whereas they were weakly activated by EGF (Fig. 8) and strongly activated by the cellular stresses arsenite and anisomycin. 5 It remained possible that ATP activated a non-MAPK pathway that could nevertheless target TCF, as has been documented in a mouse macrophage cell line (53). TCF can be easily visualized in band shift assays, where it forms a ternary complex on a 32 P-labeled SRE probe together with recombinant SRF deletion mutant that spans the MADS box and neighboring amino acids (core SRF 90 -244 in Fig. 9). Phosphorylation of TCF leads to the slowed mobility of the ternary complexes, as can be readily seen in the complexes formed by the EGFtreated whole cell extract (Fig. 9). Antibody supershift experiments show that complex I1 contains the TCF Elk-1 and that I2 is formed by SAP-1a. 6 Neither ATP, PTH, nor the two together induced any change in TCF complexes compared with untreated, control extracts (complex U; Fig. 9). Thus, ATP acti-   8. ATP does not stimulate ERK, SAPK, or p38 MAPK phosphorylation in SaOS-2 cells. SaOS-2 cells rendered quiescent by serum deprivation were stimulated with 10 M ATP␥S, 50 ng/ml EGF, and 100 ng/ml PTH alone or in combination as indicated above each lane. After 10 min, whole cell extracts were prepared, and 15 g of protein was loaded per lane on 8.5% SDS-polyacrylamide gel electrophoresis minigels, run in triplicate. After transfer to polyvinylidene difluoride, the blots were immunodetected with the following antisera: anti-phospho-Thr 202 /Tyr 204 ERK1/2, anti-phospho-Thr 183 /Tyr 185 SAPK, and anti-phospho-Thr 180 /Tyr 182 p38, followed by peroxidase-coupled anti-rabbit antiserum and detection using enhanced chemiluminescence. The blots were stripped and reprobed with anti-pan-ERK, anti-SAPK, and anti-p38 MAPK antiserum to confirm the presence of similar levels of the kinases in each lane. 7 vates the SRE through a MAPK-and TCF-independent signaling pathway.

DISCUSSION
The major findings of this study are as follows: 1) bonederived cells, both the osteosarcoma cell line SaOS-2 and primary cultures of osteoblasts from human bone, express different subtypes of the purinergic receptor class P2Y; 2) extracellular nucleotides elevate [Ca 2ϩ ] i but not [cAMP] i in both cell types; 3) their response to different nucleotides reflects the receptor subtype identified by RT-PCR; 4) these receptors functionally couple to intracellular signaling pathways that weakly activate both the endogenous c-fos gene and stably transfected c-fos promoter-luciferase reporter constructs; 5) the P2Y signaling pathway combines with one induced by PTH to synergistically elevate c-fos mRNA levels and luciferase activity; 6) the P2Y component in this synergy involves the c-fos SRE in a MAPK-and TCF-independent manner.
We have investigated two different bone cell types in this study. The immortalized cell line SaOS-2 is derived from a human osteosarcoma and shows osteoblastic characteristics (54). To confirm that results obtained from these cells were representative of osteoblasts, we also used primary cultures of cells derived from human bone explants. These contain almost exclusively osteoblasts at different states of differentiation (38), which helps explain the slight differences in P2Y receptor expression between different cultures and between individual cells within a culture (55). This variability means that the responses measured using these cells represent that of a population of cells rather than a clone.
Purinergic receptors of the P2Y class are strongly implicated in mediating the response of these cells to NTPs. P2Y receptor expression has previously been demonstrated in rat osteoblastic cell lines (34,35) and human bone-derived cells (36,37).
Here we report that, in SaOS-2 cells, ADP, ATP, and 2-me-SATP induced increases in [Ca 2ϩ ] i of similar magnitude, data consistent with expression of the P2Y 1 receptor subtype. Only marginal levels of P2Y 2 mRNA could be detected, and accordingly the P2Y 2 receptor agonist UTP was only weakly effective at elevating [Ca 2ϩ ] i . In contrast, UTP and ATP strongly elevated [Ca 2ϩ ] i in primary osteoblast population HBDC1, which correlated with high levels of P2Y 2 receptor mRNAs. Consistent with the very low levels of P2Y 1 mRNA, these cells showed no response to either ADP or 2-meSATP, while intermediate levels of P2Y 1 receptor expression by HBDC2 were consistent with the ability of ADP/2-meSATP to mobilize intracellular calcium in a subpopulation of these cells as previously reported (55). As indicated above, we attribute this heterogeneity of receptor expression to the well characterized differentiationdependent heterogeneity of osteoblast phenotype within primary populations cultured from explants of bone. Notably, nucleotide treatment did not lead to increased [cAMP] i in either cell type, thus eliminating cAMP as an intracellular second messenger for P2Y receptors in these osteoblasts.
To assess the biological activity of NTPs in osteoblasts, we have analyzed the activation of the c-fos proto-oncogene, which is strongly implicated in controlling the proliferation and differentiation of bone cells (see Introduction). We find that P2 receptor agonists induce c-fos expression in correspondence with the elevation of [Ca 2ϩ ] i via specific receptor subtypes (discussed above). This correlation is upheld in the primary population of HBDC1; UTP and ATP induced c-fos, while ADP and 2-meSATP were ineffective. Similarly, in SaOS-2 cells ADP and ATP also led to increased c-fos mRNA levels, while UTP did not. However, while 2-meSATP effectively increased [Ca 2ϩ ] i in SaOS-2 cells, this agonist failed to induce c-fos gene expression. This indicates that increased [Ca 2ϩ ] i alone is not sufficient for induction of the c-fos gene in SaOS-2 cells. Many reports have demonstrated that single G protein-coupled receptor species can be linked to multiple effector systems (55,56). Occupation of the P2Y 1 receptor by 2-meSATP could induce a receptor conformation that activates G protein-dependent hydrolysis of phosphatidylinositol 4,5-bisphosphate, but not the G protein(s) that trigger other effector pathways leading to c-fos induction. Consistent with this notion, others have documented agonist-dependent coupling of G protein-coupled receptors to various intracellular effectors (57,58).
The relatively modest increase in c-fos mRNA levels induced by nucleotides contrasted with the robust increase generated by PTH treatment. Even more striking was the synergistic effect on c-fos by the combination of NTPs and PTH, which resulted from the activation of two distinct signaling pathways. While NTPs lead to increased [Ca 2ϩ ] i without affecting cAMP, PTH does the opposite, namely elevating the levels of cAMP, cAMP-dependent protein kinase activity, and CREB phosphorylation independently of increased Ca 2ϩ (25,59,60). CREB phosphorylation alone might suffice to explain the synergistic induction of c-fos, since its major binding site in the fos promoter, the CaCRE, can integrate both cAMP and Ca 2ϩ signals in transfection assays (61). However, the assays using stably transfected reporter genes suggest that the NTP-Ca 2ϩ and PTH-cAMP pathways do not converge on the same promoter element. The CaCRE reporter was strongly induced by PTH. This induction was not significantly increased by including promoter sequences out to position Ϫ711, ruling out the contribution of other cryptic cAMP response elements active in transient transfection analyses (3,4). On the other hand, the full promoter mediated the synergy between NTPs and PTH. NTPs alone consistently gave a severalfold induction of the fos Ϫ711-linked reporter gene and background levels with the CaCRE. There are several possible explanations for this observation. One is that the NTP-Ca 2ϩ pathway weakly targets the FIG. 9. TCF phosphorylation is not induced by ATP stimulation in SaOS-2 cells. The SaOS-2 whole cell extracts described in the legend of Fig. 8 cells were analyzed for TCF phosphorylation by band shift assay. Extracts were prepared from uninduced cells (vehicle) or cells induced with 10 M ATP␥S, 50 ng/ml EGF, 100 ng/ml PTH, or ATP␥S and PTH. In addition to 10 g of whole cell extract, binding reactions also contained 32 P-labeled c-fos SRE and a SRF deletion mutation, core SRF 90 -244 . Core SRF readily allows visualization of induced related changes in TCF independently of other cellular proteins. After a 30-min incubation, the complexes were resolved by prolonged electrophoresis on 5% polyacrylamide gels. After drying, the complexes were visualized by autoradiography. Phosphorylation of TCF changes the mobility of its ternary complexes with core SRF and the SRE from the uninduced position (U) to the slowed, induced positions (I1 and I2). This mobility change reflects hyperphosphorylation of TCF resulting from activation of the ERK signaling cascade (42). Complex I1 contains hyperphosphorylated Elk-1, while complex I2 contains predominantly hyperphosphorylated SAP-1a (53). 6 CREB/activating transcription factor complexes bound to cryptic cAMP response elements located elsewhere in the c-fos promoter. This seemed unlikely, since NTPs activated neither the CaCRE reporter gene nor one driven by an array of six cAMP response elements. 7 The other is that synergism arises from the effects of the NTP-Ca 2ϩ pathway targeting another promoter element, either the SRE or the v-sis-inducible element. Ca 2ϩ signals can selectively target the SRE in neuronal (62,63), mesangial (64), and T cells (65). Accordingly, we observed that NTPs induced luciferase expression from a SREdriven reporter gene stably transfected into bone cells.
The SRE is bound and activated in vitro by a ternary complex, containing SRF and TCF, that reproduces the pattern observed in genomic footprints (6,7,11). Transactivation by TCF is strongly potentiated through its phosphorylation by the MAPKs, and the MAPK ERK can be activated by P2Y receptor agonists, as well as other Ca 2ϩ signals, in different cell types (29, 66 -70). Surprisingly, band shift and Western blotting show that ATP treatment of SaOS-2 cells does not lead to activation of any MAPK pathway or to detectable levels of TCF phosphorylation, thus ruling out a contribution by this signaling module (Figs. 8 and 9). The SRF⅐SRE complex can also mediate the calcium-driven activation of transiently introduced reporter genes, either alone (12,13) or together with the FAP element located immediately downstream of the SRE (64,65). Taken together, these observations suggest that synergy represents the combined effect of the NTP-Ca 2ϩ pathway targeting the SRE via SRF independently of TCF and the PTH-cAMP pathway targeting CREB. Notably, a similar synergism between CREB and SRF has been proposed to mediate c-fos induction by neurotrophins in neuronal cells (18). It will be interesting to determine the mechanism behind this effect and whether the MAPK/TCF signaling module can also contribute to it. Considering the strong effect on the endogenous c-fos gene, it is also possible that other regulatory elements, located intragenically, have a role in this synergy (65,71).
The fact that NTPs and PTH cooperated in both osteoblastlike osteosarcoma cell lines and primary osteoblasts suggests that this phenomenon is physiologically relevant. PTH is a systemic hormone that functions as a principal regulator of mineral homeostasis that affects primarily the skeletal system and kidney. In bone, the PTH-responsive cells are osteoblasts (72), which, following activation by this hormone, produce factors that drive osteoclastic differentiation and subsequent bone resorption (73). These data suggest that NTPs amplify the responsiveness of bone cells to PTH, thus providing a means of mounting a localized response to a systemic factor like PTH. This seems reasonable, since ATP can exist transiently in the extracellular environment following both cell damage and release via a regulatory nonlytic mechanism and, furthermore, can act synergistically with factors to enhance DNA synthesis and promote cell proliferation in Swiss 3T3 and 3T6 fibroblasts, A431 cells (27), porcine aortic smooth muscle cells (28), and thyroid cells (29). Thus, the localized release of nucleotides as a result of different physiological stresses, along with the differential expression of P2 receptors, provides the means for a highly targeted response in vivo to ubiquitous extracellular factors like PTH. This raises the possibility that this type of signal integration represents a universal means for generating a selective response in many in vitro and in vivo situations.