Fibroblast Growth Factor Receptor 3 Gene Transcription Is Suppressed by Cyclic Adenosine 3′,5′-Monophosphate

Signaling through fibroblast growth factor receptors (FGFRs) is critical for the development and patterning of the vertebrate skeleton. Gain-of-function alleles of fgfr2 andfgfr3 have been linked to several dominant skeletal disorders in humans, while null mutations in fgfr3 result in the overgrowth of long bones in a mouse model system. Interestingly, the expression pattern of fgfr3 in growth plate chondrocytes overlaps that of the parathyroid hormone (PTH)-related peptide (PTHrP) receptor, a signaling molecule that also regulates endochondral ossification. The coincident expression of these two receptors suggests that their signaling pathways may also interact. To gain insight into the regulatory mechanism(s) that govern the expression of the fgfr3 gene in chondrocytes, we have identified a cell-specific transcriptional regulatory element (CSRh) by measuring the activity of various promoter fragments in FGFR3-expressing (CFK2) and nonexpressing (RCJ) chondrocyte-like cell lines. Furthermore, we demonstrate that activation of PTH/PTHrP receptors, either by stimulation with PTH or through the introduction of activating mutations, represses CSRh-mediated transcriptional activity. Finally, the transcriptional repression of the CSRh element was mimicked by treatment with forskolin, 8-bromo-cAMP, and 3-isobutyl-1-methylxanthine or by overexpression of the catalytic subunit of protein kinase A. Together, these data suggest that protein kinase A activity is a critical factor that regulates fgfr3gene expression in the proliferative or prehypertrophic compartment of the epiphyseal growth plate. Furthermore, these results provide a possible link between PTHrP signaling and fgfr3 gene expression during the process of endochondral ossification.

Signaling through fibroblast growth factor receptors (FGFRs) is critical for the development and patterning of the vertebrate skeleton. Gain-of-function alleles of fgfr2 and fgfr3 have been linked to several dominant skeletal disorders in humans, while null mutations in fgfr3 result in the overgrowth of long bones in a mouse model system. Interestingly, the expression pattern of fgfr3 in growth plate chondrocytes overlaps that of the parathyroid hormone (PTH)-related peptide (PTHrP) receptor, a signaling molecule that also regulates endochondral ossification. The coincident expression of these two receptors suggests that their signaling pathways may also interact. To gain insight into the regulatory mechanism(s) that govern the expression of the fgfr3 gene in chondrocytes, we have identified a cellspecific transcriptional regulatory element (CSRh) by measuring the activity of various promoter fragments in FGFR3-expressing (CFK2) and nonexpressing (RCJ) chondrocyte-like cell lines. Furthermore, we demonstrate that activation of PTH/PTHrP receptors, either by stimulation with PTH or through the introduction of activating mutations, represses CSRh-mediated transcriptional activity. Finally, the transcriptional repression of the CSRh element was mimicked by treatment with forskolin, 8-bromo-cAMP, and 3-isobutyl-1-methylxanthine or by overexpression of the catalytic subunit of protein kinase A. Together, these data suggest that protein kinase A activity is a critical factor that regulates fgfr3 gene expression in the proliferative or prehypertrophic compartment of the epiphyseal growth plate. Furthermore, these results provide a possible link between PTHrP signaling and fgfr3 gene expression during the process of endochondral ossification.
Elaboration of the vertebrate skeleton occurs via two overlapping, yet distinct developmental pathways. Intramembranous ossification, the primary pathway for flat bone development, relies upon the direct differentiation of condensed mesenchyme into osteoblasts. These osteoblasts then secrete various matrix components until they are encapsulated by calcified bone, thus linking the rate of intramembranous bone growth to the rate of osteoblast differentiation. Alternatively, endochondral ossification, the primary pathway for formation of the axial and appendicular skeleton, differs from intramembranous ossification in that condensed mesenchymal cells (chondrocytes) elaborate a complex cartilaginous template as they progress through a series of developmental stages at the epiphyseal growth plate. The epiphyseal growth plate is organized into distinct cellular compartments: the resting zone, which serves as a renewable source of chondrocytes; the proliferative zone, where rapid cell division results in stacked columns of chondrocytes; and the hypertrophic zone, where the cells terminally differentiate, hypertrophy, and secrete a specialized matrix. After the encapsulated hypertrophic chondrocytes mature, the associated extracellular matrix is rapidly invaded by blood vessels and bone-forming osteoblasts, which synthesize trabecular bone. Thus, the overall rate of longitudinal bone growth is regulated by the progression of chondrocytes through these distinct developmental stages within the epiphyseal growth plate.
In vitro and in vivo model systems have been utilized to study both the commitment and subsequent differentiation of chondrocytes (1)(2)(3)(4)(5). These studies demonstrate that many physiological effectors are capable of modulating chondrocyte differentiation in vitro and that mutations or alterations in gene expression can result in skeletal defects in vivo. Recent genetic evidence has linked eight different craniosynostosis and chondrodysplasia syndromes to mutations in three of the FGFRs 1 (6,7). These autosomal dominant mutations affect both intramembranous and endochondral ossification, suggesting that FGF signaling is an essential regulator of skeletal development.
Mutations in the fgfr3 gene have been linked to several skeletal dysplasias in humans, including achondroplasia (8,9), thanatophoric dysplasia types I and II (10,11), and hypochondroplasia (12). Achondroplasia is the most common genetic form of dwarfism in humans and results from a mutation in the transmembrane domain (G380R) of FGFR3, while thanato-phoric dysplasia is the most common neonatal lethal skeletal dysplasia in humans and results from any of three independent point mutations in the fgfr3 gene. Clinically, all of these mutations result in a characteristic disruption of growth plate architecture and disproportionate shortening of the proximal limbs. Biochemically, these mutations activate FGFR3 signaling (13,14). In contrast, inactivation of FGFR3 signaling in mice results in an increase in the size of the hypertrophic zone, as well as a coincident increase in bone length (15,16). Such complementary phenotypes suggest that FGFR3 signaling is an essential component of the hierarchy of regulatory mechanisms that regulate endochondral bone growth.
Recent studies have demonstrated that PTH/PTHrP receptor signaling, like FGFR3 signaling, regulates the process of endochondral bone growth (17)(18)(19)(20). Inactivation of either PTHrP or the PTH/PTHrP receptor in mice results in the premature ossification of proliferating chondrocytes as well as a marked decrease in the size of the proliferative zone (17,18), a phenotype resembling that seen with constitutive activation of FGFR3 signaling. The hypothesis that PTHrP signaling regulates the proliferation of chondrocytes by modulating FGFR3 activity provides an explanation for the similarity in these phenotypes. Such control may involve transcriptional, translational, and/or post-translational mechanisms. To this end, we have examined whether PTH/PTHrP receptor signaling regulates the transcriptional activity of the fgfr3 gene.
To examine the mechanisms governing expression of the fgfr3 gene in chondrocytes, an "enhancer" element capable of recapitulating the chondrocytic expression pattern of fgfr3 in vitro was identified and shown to be suppressed by PTH/ PTHrP receptor signaling. Furthermore, activation of protein kinase A (PKA) signaling, either through the elevation of cAMP levels or through the overexpression of the catalytic subunit of PKA, also suppressed the transcriptional activity of the enhancer element and decreased the steady state levels of the endogenous fgfr3 gene. These results suggest a model whereby the attenuation of PKA signaling, by effectors such as the PTH/PTHrP receptor, serves to regulate endochondral ossification by modulating the chondrocytic expression of fgfr3.
Electrophoretic mobility shift assays (EMSAs) were used to identify a single hexameric "orphan" receptor half-site in the enhancer element that interacts with nuclear protein(s). Elevation of cAMP levels failed to disrupt this DNA-protein complex; however, mutations in this motif that inhibit nuclear protein(s)/DNA interactions do dramatically impair chondrocytic transcriptional activity. Nevertheless, these same mutations did not prevent the cAMP-mediated suppression of transcriptional activity, thus suggesting that there are additional DNA/protein interactions within the enhancer element that are required for the cAMP-dependent regulation of fgfr3 expression.
To generate pCRE-luc, CREB-A (5Ј-GGT TGC CTG ACG TCA GAG A-3Ј) and CREB-B (5Ј-CCT CTC TGA CGT CAG GCA A-3Ј) synthetic oligonucleotides (Life Technologies, Inc.) were 5Ј-phosphorylated with T4 kinase and annealed. The annealed oligonucleotides were concatamerized by ligation with T4 ligase, blunted with Klenow and deoxynucleotide triphosphates, and then subcloned into the SmaI site of RSV-luc. Clones were sequenced to verify insert identity and orientation. pCRE-luc (TL IV 179.22) contains three copies of CREB-A in a head-to-tail array upstream of the RSV minimal promoter.
Expression Plasmids and Other DNA Reagents-The rat FGFR3 RNase protection probe was generated by amplifying rat cDNA with the following primers: DO80, 5Ј-GTC ATG GAA AGT GTG G-3Ј; and DO113, 5Ј-GCT CCG ACA CAT TGG-3Ј. The resulting product was cloned into pGEM-T (Promega Corp.), and both the identity and orientation of the insert was confirmed through sequencing. The plasmid pTRI-␤-Actin-125-Rat was purchased from Ambion, Inc. The expression plasmid for the catalytic subunit of protein kinase A (MT-C␣) was obtained from S. McKnight (23), while empty vector control (pEV142) was provided by R. Palmiter (24). Expression plasmids for wild-type (HKrk) (25) and mutated versions (HKrk-H223R (25) and HKrk-T410P (25)) of the human PTH/PTHrP receptor were provided by H. Jü ppner.
Cell Culture and Transfection-CFK2 (26) and RCJ (27) (clone 3.1C5.18) cell lines were obtained from J. Henderson and J. Aubin, respectively. Cell lines were maintained subconfluent in Dulbecco's modified Eagle's medium (Life Technologies) supplemented with 10% fetal calf serum (Sigma), 2 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml). All cells were transfected in triplicate using a modified calcium phosphate precipitant (28). The day prior to transfection, cells were plated at a density of 8 ϫ 10 4 cells/well in a 12-well plate (Corning). Four hours prior to transfection, cells were provided with 0.8 ml of fresh media. Five micrograms of each DNA construct (double-banded on a CsCl 2 gradient) and 0.5 g of pCScyto␤gal (29,30) were diluted to a final volume of 20 l with distilled, deionized water. To that was added 250 l of 0.272 M CaCl 2 and 270 l of 2ϫ BBS (50 mM N,N-bis[2-hydroxyethyl]-2-aminoethanesulfonic acid (BES), 280 mM NaCl, 1.5 mM Na 2 HPO 4 ). The DNA mixture was then allowed to precipitate at room temperature for 15 min, and 80 l of the precipitant was then added directly to the cells. Cells were incubated (5% CO 2 , 37°C) with the precipitate for 20 h, at which point they were then washed with Dulbecco's modified Eagle's medium and returned to fresh media. When cells were exposed to various pharmacological agents, cells were serum-starved (0.5% serum/Ham's F-12 medium) for 12 h prior to treatment. Sixty hours post-transfection, cells were washed with 1 ml of 1ϫ PBS and subsequently lysed in the well with 250 l of luciferase lysis buffer (25 mM Tris⅐HCl (pH 7.8), 2 mM dithiothreitol (DTT), 2 mM trans-1,2-diaminocyclohexane-N,N,NЈ,NЈ-tetraacetic acid , 10% glycerol, 1% Triton X-100). Insoluble materials were removed from all lysates by centrifugation at 14,000 rpm for 4 min, and 50 -100 l of the cleared lysate was used to analyze luciferase reporter expression as described previously (31). ␤-Galactosidase activity was determined with the Galacto-Light Plus system, as described by the manufacturer (Tropix, Inc.).
Forskolin (Sigma) was dissolved in Me 2 SO (Sigma) to generate a 1000ϫ stock. The stock solution was diluted to 100ϫ with Ham's F-12 prior to addition to cells. Synthetic human PTH-(1-34) (Sigma) was resuspended in 0.01 M acetic acid and stored at Ϫ80°C. A 2000ϫ stock of PTH was diluted 1:20 with Ham's F-12 medium prior to the stimulation of PTH/PTHrP receptor-transfected cells.
RNase Protection Analysis-Gene expression was determined by RNase protection analysis. Briefly, total cellular RNA was purified from a confluent 10-cm dish using the RNeasy kit (Qiagen, Inc.). Twenty micrograms of RNA was lyophilized prior to resuspension in 30 l of hybridization buffer (40 mM PIPES (pH 6.4), 400 mM NaCl, 1 mM EDTA, 80% formamide), which contained 0.5-5ϫ10 5 cpm of the appropriate [␣-32 P]UTP (800 mCi/mmol; NEN Life Science Products) transcribed probes. Samples were denatured at 95°C for 5 min and then incubated at 50°C for 8 h. Unhybridized probe was removed by adding 350 l of RNase digestion buffer (10 mM Tris⅐HCl (pH 7.5), 300 mM NaCl, 5 mM EDTA, 14 g of RNase A, 248 units of RNase T1) and incubating the reaction for 30 min at 30°C. Duplex RNA hybrids were then purified with RNAzol B (Teltest, Inc.), dried under vacuum, and resuspended in 8 l of RNA loading buffer (1 mM EDTA, 0.1% bromphenol blue, 0.1% xylene cyanole FF, 80% formamide). Samples were denatured for 5 min at 95°C, chilled on ice, and resolved on a 5% denaturing polyacrylamide gel (Long Ranger; J. T. Baker). Gels were dried under vacuum and exposed to Kodak X-Omat film. When RNase protection assays were quantified, gels were analyzed with a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA) and quantified with ImageQuant software (version 3.3, Molecular Dynamics). The sizes for the protected fragments are as follows: rat FGFR3, 291 nt; rat ␤-actin, 126 nt.
Nuclear Extracts and EMSAs-Nuclear extracts were prepared according to the protocol of Dignam et al. (32). Briefly, 10 10-cm or six 15-cm plates of confluent cells were washed three times with 1ϫ PBS and scraped into 15 ml of 1ϫ PBS. The cells were collected by centrifugation at 1600 rpm for 5 min at 4°C. The supernatant was discarded, and the cells were washed with 5 cell volumes of 1ϫ PBS and pelleted as described above. The supernatant was discarded, and the pellet was resuspended in 5 ml of buffer A (10 mM HEPES (pH 7.9, at 4°C), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT). The cell suspension was then incubated for 10 min at 4°C, transferred to an ice-cold sterile Dounce homogenizer, and lysed with 30 strokes of a tight fitting pestle. Nuclei were collected by centrifugation at 1600 rpm for 5 min at 4°C, and the pellet was resuspended in 2 ml of buffer C (20 mM HEPES (pH 7.9 at 4°C), 25% (v/v) glycerol, 420 mM NaCl, 1.5 mM MgCl 2, 0.2 mM EDTA, 0.5 mM phenylmethanesulfonyl fluoride, 0.5 mM DTT, protease inhibitors). The nuclear suspension was transferred to a sterile 50-ml tube, and the proteins were extracted for 1 h at 4°C with gentle stirring. The insoluble material was removed by centrifugation at 14,000 rpm for 10 min at 4°C, and the extract was dialyzed against two 100-volume changes of buffer D (20 mM HEPES (pH 7.9 at 4°C), 20% (v/v) glycerol, 100 mM KCl, 0.2 mM EDTA, 0.5 mM phenylmethanesulfonyl fluoride, 0.5 mM DTT, protease inhibitors). Finally, aliquots of the nuclear extracts were transferred to silanized tubes and snap frozen in liquid nitrogen. Extracts were stored at Ϫ80°C.
To analyze DNA binding proteins from treated cells, micronuclear extracts were isolated as described (33). Briefly, subconfluent 35-mm dishes were starved in 0.5% fetal calf serum/F-12 medium for 12 h, followed by treatment with various agents for 12 h (see the legend for Fig. 5). Cells were washed twice with ice-cold 1ϫ PBS and collected by scraping into 1.5 ml of ice-cold 1ϫ PBS. Cells were collected by centrifugation for 20 s at 14,000 rpm and resuspended in 100 l of modified buffer C (17.5 mM HEPES (pH 7.9 at 4°C), 22% (v/v) glycerol, 368 mM NaCl, 1.3 mM MgCl 2, 175 M EDTA, 0.5% Triton X-100, 0.5 mM DTT, 0.5 mM phenylmethanesulfonyl fluoride, 350 M Na 3 VO 4 , and protease inhibitors) with gentle pipeting. Proteins were eluted for 1 h on ice, and the extract was cleared by centrifugation at 14,000 rpm for 2 min at 4°C. Protein concentrations of the extracts were determined with a modified Bradford assay (Bio-Rad) and used directly.
Complementary oligonucleotides encompassing the desired binding site sequences were synthesized locally. All oligonucleotides used were gel-purified and subsequently annealed in 1ϫ oligo buffer (10 mM Tris⅐HCl (pH 7.5), 1 mM EDTA, 50 mM KCl) by mixing in a 1:1 molar ratio, heating to 95°C for 2 min, 97°C for 30 s, and cooling to 25°C over a period of 2 h. Duplex oligonucleotide probes were then end-labeled with exo Ϫ Klenow (Stratagene, Inc.) and purified over an S-200 HR size exclusion column (Amersham Pharmacia Biotech). The specific activity of the probe was estimated and ultimately diluted to 25,000 cpm/0.1 pmol prior to hybridization. Hybridizations were performed at room temperature for 20 min in a volume of 20 l, which consisted of 0.1 pmol of probe (25,000 cpm), 3 g of bovine serum albumin (Fisher), 1.25 g of poly(dI⅐dC) (Amersham Pharmacia Biotech), 4 g nuclear extract, and buffer D. Duplex oligonucleotide competitors were preincubated with the nuclear extract for 5-10 min at 25°C. The resulting complexes were resolved on a 4 -20% gradient gel (Novex) prerun in 0.375ϫ TBE (33 mM Tris borate (pH 8.3), 75 M EDTA) at 100 V for 100 min as described (31). Gels were dried under vacuum and subsequently exposed to Kodak X-Omat film.
The sense strand of each oligonucleotide used for EMSAs are shown in Table I. To demonstrate the functionality of CAAT/enhancer-binding protein (C/EBP), thyroid hormone response element (TRE), and retinoid-like orphan receptor ␣ family 1 (ROR␣1) competitors, the duplex DNA binding sites were radiolabeled and used to probe CFK2 nuclear extracts. All competitors were able to form gel shift complexes (data not shown).
FGFR Expression-Expression of fgfrs was determined by reverse transcriptase-polymerase chain reaction as described previously (34).

Identification and Localization of Chondrocytic Regulatory
Elements-FGFR3 is an essential regulator of endochondral ossification and is expressed in a subset of chondrocytes found within the developing epiphyseal growth plate (15,16,35). The specific expression pattern of fgfr3 suggests that it has stagespecific effects on the process of chondrocyte maturation. To examine the regulation of fgfr3 gene expression, two chondrocytic cell lines, CFK2 and RCJ, were utilized. CFK2 cells and RCJ cells, both derived from fetal rat calvarial mesenchyme, express chondrocyte-specific genes and elaborate a cartilagelike matrix (26,27). CFK2 cells express the fgfr1 (Fig. 1A, lanes 1 and 2), fgfr2 (Fig. 1A, lane 2), and fgfr3 (Fig. 1A, lane 1) genes, while RCJ cells only express fgfr1 (Fig. 1A, lanes 4 and 5) and fgfr2 (Fig. 1A, lane 5). Neither cell line expresses fgfr4 (Fig. 1A,  lanes 3 and 6). These data demonstrate that expression of the endogenous fgfr3 gene is regulated in a cell-specific manner and may reflect differences in the stage of chondrocyte differentiation.
To define the cis-acting sequences that are responsible for the CFK2-specific expression of the fgfr3 gene, we compared the activities of various fgfr3 promoter fragments in both the fgfr3-expressing (CFK2) and nonexpressing (RCJ) chondrocytelike cell lines (Fig. 1B). Previous work has demonstrated that basal promoter elements from the fgfr3 gene function equally well in both cell lines (21). However, by exploiting cell-specific differences, we demonstrated that the sequences located between Ϫ2311 and Ϫ1537 selectively enhanced the transcription of luciferase reporter constructs in the CFK2 cell line and failed to alter basal transcriptional activity in the RCJ cell line (Fig.  1B), thus recapitulating the cell-specific expression pattern of the endogenous fgfr3 gene in these cell lines.
To verify that the chondrocyte cell-specific region (CSR; nucleotides Ϫ2311 to Ϫ1537) alone was responsible for the observed CFK2-specific transcriptional activity, the CSR sequences were multimerized into a three-copy, head-to-tail array and placed 5Ј of the Rous sarcoma virus minimal promoter (pRSV-luc; Ref. 22). The transcriptional activity of the pCSR-luc reporter construct was examined in both the CFK2 and RCJ cell lines (Fig. 2). The CSR sequence elements selectively enhanced the transcriptional activity of the RSV minimal promoter (230-fold relative to pRSV-luc) in the CFK2 cell line, while only modestly affecting the transcriptional activity of the RSV promoter in RCJ cells (25-fold). These data, as well as the data presented in Fig. 1B, demonstrate that enhancer cAMP-mediated Repression of fgfr3 Gene Expression elements capable of recapitulating the chondrocytic expression of the fgfr3 gene in vitro are located between nucleotides Ϫ2311 and Ϫ1537. Furthermore, the CSR sequences alone could confer selective transcriptional activity upon a heterologous, minimal promoter and do not require additional fgfr3 promoter elements to selectively activate transcription in CFK2 cells.
To identify the chondrocyte-specific regulatory element(s) found within the Ϫ2311 to Ϫ1537 region, various pieces of the CSR sequences were multimerized into three-copy, head-to-tail arrays and placed 5Ј relative to the RSV minimal promoter (Fig. 3A). The transcriptional activity of reporter constructs bearing the 5Ј-most 50 nucleotides of the CSR (pCSRa-luc, pCSRc-luc, pCSRe-luc, pCSRf-luc, pCSRg-luc, and pCSRh-luc) was dramatically enhanced in fgfr3-expressing CFK2 cells (Fig.  3A), while little or no enhancement of RSV-mediated transcriptional activity was observed in RCJ cells (data not shown). Therefore, the position of chondrocytic "enhancer" activity was further refined to nucleotides Ϫ2311 and Ϫ2263 of the fgfr3 gene. Analysis of these sequences, using the MatInspector program (36,37), identified binding sites for several known transcription factors (Fig. 3B). Of interest were putative binding sites for the cAMP response element-binding protein (CREB), thyroid hormone receptor (T3R), C/EBP, and the ROR␣1/2, all components of signaling pathways that are known to affect chondrocyte development (2, 38 -43).
PTH/PTHrP Receptor Signaling Regulates CSR Activity in Vitro-Recent studies have demonstrated that PTH/PTHrP receptor signaling, like FGFR3 signaling, regulates the process of endochondral bone growth (17)(18)(19). fgfr3 is expressed throughout the proliferative zone, while peak levels of fgfr3 expression are found in the prehypertrophic region of the epiphyseal growth plate. PTH/PTHrP receptor is also expressed predominately in prehypertrophic chondrocytes (19,44,45). The overlapping patterns of fgfr3 and PTH/PTHrP receptor expression suggested that PTH/PTHrP receptor signaling might regulate fgfr3 gene expression in vivo.
PTH/PTHrP receptor signaling is known to modulate chondrocyte development, presumably through the activation of cAMP-and/or phospholipase C-associated intracellular signal-  1-3) and RCJ (lanes 4 -6) cell lines. The kinase domains of fgfrs 1-4 were amplified simultaneously using a single oligonucleotide primer pair (34). The resulting mixture of polymerase chain reaction-amplified products was digested with PstI (lanes 1 and 4), PvuII (lanes 2 and 5), and EcoRI (lanes 3 and 6) to determine the expression of the fgfr1/fgfr3, fgfr2, and fgfr4 genes, respectively. Fragments were resolved on 2% agarose gels and visualized by silver staining. Predicted sizes for PstI digests (lanes Data are presented as in Fig. 2 (pRSV-luc ϭ 1). B, genomic sequence encompassing the Ϫ2311 to Ϫ2263 chondrocytic region of the FGFR3 gene. Shown both above and below the sequence are binding sites for known transcription factors: thyroid hormone response element (T3R), ROR␣1/2, CREB, and C/EBP. All sites show a greater than 90% similarity to known binding sites.
ing cascades (46). Given the overlap in expression patterns and the presence of a CREB motif in the CSRh element, it was important to determine whether signals emanating from the PTH/PTHrP receptor were capable of regulating the transcriptional activity of the CSRh enhancer element. To mimic PTH/ PTHrP receptor signaling in CFK2 cells, expression vectors encoding either the wild-type or one of two constitutively active mutant isoforms of the human PTH/PTHrP receptor (T410P and H223R; Ref. 25), both associated with Jansen's metaphyseal chondrodysplasia (JMC), were co-transfected along with pCSRh-luc. Co-expression of the wild-type PTH/PTHrP receptor, as well as both mutant isoforms, resulted in a 37-40% decrease in CSRh-mediated transcriptional activity (Fig. 4A). Furthermore, the repression mediated by both the wild-type and T410P mutant PTH/PTHrP receptors could be enhanced by the addition of 50 nM PTH, while treatment of the H223Rtransfected cells with PTH failed to augment the observed transcriptional suppression (Fig. 4A, compare wild type, T410P, and H223R transfections with and without PTH). Taken together, these data suggest that CFK2 cells express low levels of a PTHrP-like ligand and that signaling through the PTH/PTHrP receptor can regulate fgfr3 gene expression through the CSRh enhancer element.
To confirm that PTH/PTHrP receptors were functioning as expected in CFK2 cells, both wild-type and JMC-associated PTH/PTHrP receptor expression constructs were co-transfected with a cAMP-responsive (pCRE-luc) reporter gene. Relative to empty vector, co-expression of the wild-type PTH/ PTHrP receptor resulted in a 2.2-fold induction of CREmediated transcriptional activity, while stimulation of wildtype receptor-expressing cells with 50 nM human PTH-(1-34) resulted in a 10-fold increase in transcriptional activity (Fig.  4B, compare wild-type receptor transfected cells with or without PTH with vector controls). Unlike the modest stimulation observed as a result of expression of the wild-type receptor, co-expression of either the T410P or H223R mutant PTH/ PTHrP receptors resulted in a PTH-independent 5.6-or 6.3-fold increase in CRE-mediated transcriptional activity, respectively (Fig. 4B, compare T410P and H223R mutant receptor-transfected cells with vector transfection). Stimulation of the T410P mutant receptor with 50 nM PTH resulted in a significant increase in CRE-mediated transcriptional activity (Fig. 4B, compare T410P transfected cells with and without PTH), while treatment of the H223R-transfected cells with PTH failed to augment CRE-mediated transcriptional activity.
cAMP-mediated Repression of the Chondrocytic Regulatory Element and Endogenous fgfr3-To determine whether the PTH/PTHrP receptor-mediated transcriptional repression of the CSRh enhancer element depends upon cAMP-mediated signaling pathways, pCSRh-luc-transfected CFK2 cells were serum-starved for 12 h to lower background levels of cAMP. Transfected cells were then exposed to various pharmacological agents known to elevate the intracellular concentration of cAMP. Activation of adenylate cyclase with 10 M forskolin (FSK) resulted in a 90% decrease in the transcriptional activity of pCSRh-luc (Fig. 5A). Moreover, similar decreases in the CFK2-specific transcriptional activity of pCSRh-luc were observed when cAMP levels were elevated either directly with 8-bromo-cyclic adenosine 3Ј,5Ј-monophosphate (8-Br-cAMP; a cell-permeable cAMP analog) or indirectly through the inhibition of cAMP phosphodiesterase activity with 3-isobutyl-1-methylxanthine (IBMX) (47) (Fig. 5A, 81 and 95%, respectively).
To examine the specificity of this transcriptional inhibition, the regulatory effects of AMP, the noncyclic precursor to cAMP, were compared with 8 Br-cAMP. The transcriptional activity of pCSRh-luc was unaffected by as much as 1 mM AMP, while a dose-dependent repression of CSR-mediated transcriptional ac-  Fig.  4A. D, FSK-induced repression of CFK2-specific endogenous rat fgfr3 gene expression. Cells were serum-starved for 12 h in 0.5% serum prior to treatment for 12 h with either 0.1% Me 2 SO (DMSO) or 10 M forskolin. RNA was harvested, and the relative levels of rat FGFR3 mRNA and ␤-actin were determined simultaneously using RNase protection. The rat FGFR3 probe protects a 291-nt fragment, and the ␤-actin probe protects a 126-nt fragment. tivity was observed with 8-Br-cAMP (Fig. 5B). A 50% decrease (IC 50 ) in CSRh-mediated transcriptional activity resulted from treatment with 70 M 8-Br-cAMP, while maximal inhibition (approximately 90%) was obtained with 300 M 8-Br-cAMP. Together, these data demonstrate that the transcriptional potential of the CSRh enhancer element is regulated by cAMP.
In addition, we examined the transcriptional response of both the full-length promoter construct (p(Ϫ2311/Ϫ27)FR3-luc) and the endogenous fgfr3 gene to increasing concentrations of cAMP. Treatment of serum-starved CFK2 cells with 10 M forskolin specifically repressed CFK2-cell specific transcriptional activity of p(Ϫ2311/Ϫ27)FR3-luc while having little or no effect upon the basal transcriptional activity of p(Ϫ2311/ Ϫ27)FR3-luc in RCJ cells (Fig. 5C). To examine the effects of cAMP on the expression of the endogenous gene, serum-starved CFK2 cells were stimulated for 12 h with either 0.1% Me 2 SO or 10 M forskolin, and the relative levels of fgfr3 mRNA were determined by RNase protection (Fig. 5D). Band intensities were quantified, and the decrease in fgfr3 mRNA levels was determined relative to an internal ␤-actin control. Treatment with Me 2 SO failed to repress the steady state levels of fgfr3 mRNA, while treatment with 10 M forskolin resulted in a 46% decrease in fgfr3 mRNA levels. A second experiment was performed following a 24-h induction with forskolin. At this time point, no change in fgfr3 mRNA levels was observed relative to ␤-actin. Together, the transcriptional response of the fulllength promoter and of endogenous fgfr3 to forskolin support the hypothesis that fgfr3 can be regulated by the intracellular concentration of cAMP.
Protein Kinase A Overexpression Represses CSRh-mediated Transcription-cAMP levels can affect gene transcription by selectively activating PKA (48). Through its activation, PKA has been shown to augment the transcription of genes that contain a CRE(s) by selectively phosphorylating CREB at Ser-133 (49). Phosphorylated CREB is then capable of enhancing transcription by interacting with the CBP/p300 family of transcriptional co-activators (50,51).
In the inactive form, PKA exists as a tetramer of two catalytic (C␣) and two regulatory subunits. Binding of cAMP by the regulatory subunits results in the dissociation of the tetramer complex and release of the active C␣ subunit (52). To assess the ability of C␣ overexpression to mimic the effects of PKA activation, the transcriptional response of pCRE-luc was examined after co-transfection with the Zn 2ϩ -inducible metallothionine promoter-C␣ (MT-C␣) expression vector (23). As expected, cotransfection of MT-C␣ resulted in a 3-fold increase in the transcriptional activity of pCRE-luc, while the Zn 2ϩ -dependent overexpression of C␣ led to a 12-fold enhancement in CREmediated transcriptional activity (Fig. 6A, compare vector with and without Zn 2ϩ with MT-C␣ with and without Zn 2ϩ) . Therefore, overexpression of the catalytic domain of PKA mimics cAMP-mediated transcriptional activation of CRE-containing genes.
To determine if PTH/PTHrP receptor-mediated repression of the CSR sequences also depends on PKA activation, the MT-C␣ expression vector was co-transfected with the pCSRh-luc reporter construct. Although treatment of the control vectortransfected cells with Zn 2ϩ resulted in a 27% decrease in transcriptional activity (Fig. 6B, compare vector co-transfections with and without Zn 2ϩ ), an additional 50% decrease in CSRhmediated transcriptional activity was observed when C␣ activity was induced with 80 M Zn 2ϩ (Fig. 6B, compare MT-C␣ transfections with and without Zn 2ϩ treatment). These data suggest that the repression of CSRh-mediated transcriptional activity by cAMP results from the activation of PKA signaling cascades.

Nuclear Factor(s) Interact with 5Ј-PuGGTCA-3Ј Core and
Regulate the Transcriptional Potential of the CSRh Sequences-Recruitment or stabilization of the transcriptional initiation complex by DNA-binding proteins and/or associated accessory factors is thought to be the underlying mechanism that regulates gene expression. To identify the DNA-protein interaction(s) that direct the observed chondrocytic transcriptional activity, the CSRh fragment was radiolabeled and used as a probe in EMSAs. Incubation of radiolabeled CSRh with extracts from CFK2 cells identified one major complex (Fig. 7A,  lane 2); however, a complex with a similar mobility was also obtained with RCJ nuclear extracts (Fig. 7A, lane 3). To assess the specificity and identity of the CSRh-binding protein(s) (CSR-BP), nuclear extracts from CFK2 cells were preincubated with various unlabeled competitors. The CSR-BP⅐CSRh complex was completely disrupted by competition with a 10-fold molar excess of unlabeled probe (Fig. 7B, lanes 7-9), while a C/EBP consensus binding site failed to compete efficiently for the binding of CSR-BP even at a 100-fold molar excess (Fig. 7B, lanes 10 -12). Significant competition for the binding of CSR-BP was also observed with TRE and ROR␣1 consensus binding sites (Fig. 7C, lanes 14 -16 and 17-19, respectively), while, surprisingly, a known CREB binding site failed to compete effectively even at a 100-fold molar excess (Fig. 7C, lanes  20 -22).
A number of different Zn 2ϩ finger-containing transcription factors recognize DNA motifs similar to the 5Ј-PuGGTCA-3Ј hexamer element found in the CSRh fragment (53). To determine if this motif is involved in the binding of the CSR-BP complex, nuclear extracts derived from CFK2 cells were preincubated with several CSRh-like competitors (Fig. 8A). Significant competition for the binding of radiolabeled CSRh was observed when CFK2 cell extracts were preincubated with either DM201/2, a duplex DNA fragment corresponding to the sequences between Ϫ2311 and Ϫ2276 of the mouse fgfr3 gene, or DM221/2, a duplex DNA fragment containing a tetranucleotide substitution outside of the Zn 2ϩ finger DNA binding motif (Fig. 8B, lanes 2-4 and 8 -10, respectively). However, mutations that disrupt the 5Ј-PuGGTCA-3Ј motif (DM223/4) failed to compete for the formation of the CSR-BP⅐CSRh complex, even at a 100-fold molar excess (Fig. 8B, lanes 5-7). Identical results were obtained when DM201/2, DM223/4, and DM221/1  ϭ 1). B, repression of pCSRh-luc by co-expression of C␣. CFK2 cells were transfected with combinations of pCSRh-luc and either pEV142 (vector) or MT-C␣ expression plasmids. Transfected cells were then treated as described above; however, data are presented as percentage of control (untreated vector ϭ 100%).
were radiolabeled and used as EMSA probes directly (data not shown). Incubation of CFK2 nuclear extracts with DM201/2 and DM221/2 resulted in the formation of a stable DNA-protein complex, while DM223/4 failed to form a resolvable DNA-protein complex (data not shown).
To determine whether the loss of protein binding activity observed with the DM223/4 mutations correlated with a loss in CSRh-mediated transcriptional activity, mutations in the Zn 2ϩ finger DNA binding motif (Fig. 8A, DM223/4) were introduced into pCSRh-luc (pmCSRh-luc), and the resulting transcriptional activity was examined in CFK2 cells. Disruption of the 5Ј-PuGGTCA-3Ј motif resulted in a 75% decrease in CSRhmediated transcriptional activity (Fig. 8C). These data suggest that the 5Ј-PuGGTCA-3Ј sequence motif regulates the transcriptional activity of the CSRh fragment by facilitating the formation of a CSR-BP⅐CSRh transcriptional regulatory complex.
To determine if cAMP levels affect the transcriptional activity of the CSRh element by regulating the formation of the CSR-BP⅐CSRh complex, modified nuclear extracts were prepared from cells treated with 0.1% Me 2 SO, 10 M forskolin, or 0.5 mM IBMX. Although treatment with either forskolin or IBMX resulted in a 90 -95% decrease in transcriptional activity (Fig. 5A), the CSRh binding activity found in CFK2 cells was not affected by elevations in the concentration of cAMP (Fig. 8D).
The 5Ј-PuGGTCA-3Ј Motif Is Not Sufficient for cAMP-mediated Suppression of the CSRh Element-To determine whether the Zn 2ϩ finger binding motif is both necessary and sufficient for the cAMP-mediated suppression of the CSRh enhancer element, the transcriptional response to forskolin of the fulllength promoter containing the Zn 2ϩ finger mutation (pm-(Ϫ2311/Ϫ27)FR3-luc) was examined. Like p(Ϫ2311/Ϫ27)FR3luc, the transcriptional activity of pm(Ϫ2311/Ϫ27)FR3-luc was suppressed by treatment with forskolin, whereas the activity of the promoter lacking the CSR element (p(Ϫ1537/Ϫ27)FR3-luc) was largely unaffected (Fig. 9). Despite the marked decrease in basal activity, pmCSRh-luc also remained responsive to cAMP (Fig. 9). Together, these results suggest that another CSRh binding activity (or activities) is required for cAMP-mediated suppression of the CSRh enhancer element. DISCUSSION Recently, both gain-of-function and loss-of-function mutations in fgfr3 have revealed unique roles for this receptor dur- S/THR identifies the complement of the hexameric half-site. DM201/2 corresponds to sequences between Ϫ2311 and Ϫ2276 of the mouse fgfr3 gene, while mutations in DM223/4 (5Ј-PuGGTCA-3Ј core) and DM221/2 (AT-rich sequences) are underlined. B, competition for binding of radiolabeled CSRh fragment with DM201/2, DM223/4, and DM221/2 duplex DNA sequences. Extracts were preincubated with 10-, 50-, or 100-fold molar excess of unlabeled competitor DNA prior to incubation with labeled CSRh fragment. The resulting complexes were resolved on a 4 -20% nondenaturing gradient gel. C, transcriptional activity of pmC-SRh-luc reporter construct. Mutations corresponding to those in DM223/4 were introduced in the pCSRh-luc construct to generate pmC-SRh-luc. The transcriptional potential of pmCSRh-luc was then compared with pCSRh-luc in CFK2 cells. Data are plotted as the mean Ϯ S.D. of the ratio of luciferase to ␤-galactosidase for three independent assays. D, CSR-BP⅐CSRh complex formation following cAMP induction. CFK2 cells were serum-starved for 12 h prior to treatment as follows: no treatment, 0.1% Me 2 SO (DMSO), 10 M FSK, or 0.5 mM IBMX. Modified nuclear extracts (see "Experimental Procedures") were prepared from the treated cells, and protein concentrations were quantified. Equal amounts of protein were complexed with the radiolabeled CSRh fragment and resolved on a 4 -20% nondenaturing gradient gel. ing development (6,15,16,54). Loss-of-function alleles of fgfr3 lead to an increase in the size of the hypertrophic zone and the subsequent overgrowth of long bones (15,16), while gain-offunction mutations in fgfr3 have been genetically linked to autosomal dominant disorders where both the size and architecture of the epiphyseal growth plate are altered (8 -11). Together, these observations demonstrate that FGFR3-mediated signaling is an essential regulator of endochondral ossification.
Previous studies that initially characterized the fgfr3 promoter identified sequences between Ϫ220 and ϩ612 of fgfr3 that were capable of promoting efficient transcription in vitro, as well as supporting tissue-specific expression in vivo (21). Nevertheless, these minimal elements failed to recapitulate the entire tissue-specific expression pattern of fgfr3 and suggested the existence of other "enhancer" elements. A chondrocytic "enhancer" capable of recapitulating the in vitro expression pattern of the endogenous fgfr3 gene was identified by examining the transcriptional activity of a number of different reporter constructs in fgfr3-expressing (CFK2) and nonexpressing (RCJ) cell lines. The sequences found between Ϫ2311 and Ϫ1537 (CSR) of fgfr3 enhanced the transcriptional activity of the fgfr3 promoter in CFK2 cells, yet did not affect basal transcriptional activity in RCJ cells. The inability of the CSR enhancer to stimulate transcriptional activity in RCJ cells demonstrated that this element is not required for fgfr3 basal promoter activity and that it functions in a cell-specific manner. It should also be noted that sequences located 5Ј to the CSR suppressed CFK2-specific transcriptional activity; however, the nature of this repression remains to be identified and characterized.
In many cases, transcription factor binding sites are modular and can confer transcriptional activity upon heterologous promoters. Using the RSV minimal promoter, sequences between Ϫ2311 and Ϫ1537 of fgfr3 were found to selectively enhance the RSV-mediated transcriptional activity in CFK2 cells, thus demonstrating that these sequences are both necessary and sufficient for chondrocytic cell-specific regulation in vitro. Additional deletion constructs were used to map the chondrocytic "enhancer" activity to sequences between Ϫ2311 and Ϫ2263 (referred to as CSRh). Analysis of this sequence identified overlapping binding sites for two steroid/thyroid hormone receptor superfamily members and a CREB-like binding site.
Like FGFR3, both gain-of-function and loss-of-function alleles of the PTH/PTHrP receptor have been linked to defects in endochondral bone growth (17,25,55). These results demonstrate that PTHrP signaling, like FGF signaling, regulates endochondral ossification. In situ hybridization studies have demonstrated that the expression pattern of the PTH/PTHrP receptor and fgfr3 overlap in the epiphyseal growth plate and suggest that PTHrP signaling may modulate the chondrocytic expression of fgfr3 in vivo.
Both PTH and PTHrP, through their interaction with a common G protein-coupled receptor (PTH/PTHrP receptor), activate cAMP and Ca 2ϩ second messenger signaling pathways by stimulating adenylate cyclase and/or phospholipase C activity, respectively (46). To determine if PTH/PTHrP receptor signaling could affect the transcriptional activity of the CSRh enhancer element, both wild-type and the constitutively active JMC isoforms of the human PTH/PTHrP receptor were cotransfected with pCSRh-luc. Although ligand-independent suppression of the CSRh enhancer was observed with all receptor isoforms, only cells expressing either the wild-type or the T410P mutant PTH/PTHrP receptors responded to PTH stimulation by further suppressing the transcriptional activity of the CSR element.
Biochemical characterization of the JMC mutations has demonstrated that both the T410P and H223R mutations induce the ligand-independent production of cAMP (55); however, PTH/PTHrP receptors harboring the H223R mutation fail to stimulate inositol phosphate turnover in response to ligand binding (25). Therefore, these results, as well as preliminary studies looking at the role of intracellular Ca 2ϩ , suggest that maximal repression of the CSRh enhancer element requires both the cAMP-and Ca 2ϩ -dependent signaling pathways. 2 Such synergistic "cross-talk" between cAMP and Ca 2ϩ second messenger signaling cascades may result from a positive feedback loop involving the activation of Ca 2ϩ /calmodulin-stimulated adenylate cyclases (56). Additional studies will be required to elucidate the transcriptional role of Ca 2ϩ . Forskolin, a naturally occurring diterpene, elevates intracellular cAMP levels by directly activating adenylate cyclase. Unlike activation of the well characterized CREB signaling pathway, forskolin suppressed the CFK2-specific transcriptional activity of the p(Ϫ2311/Ϫ27)FR3-luc and pCSRh-luc reporter constructs, as well as expression of the endogenous fgfr3 gene. These observations, as well as the dose-dependent repression of pCSRh-luc activity by 8-Br-cAMP, demonstrated that the chondrocyte-specific transcriptional activity of the CSRh enhancer element can be down-regulated by cAMP.
Enhancer elements influence transcription by facilitating and/or stabilizing the assembly of the transcriptional initiation complex. Therefore, EMSAs were used to identify potential protein/CSRh DNA interactions. These experiments identified one DNA-protein complex in CFK2 nuclear extracts (CSR-BP⅐CSRh). Proteins in the CSR-BP⅐CSRh complex were recognized by steroid/thyroid hormone (S/THR) binding site (T3R, ROR␣1, and rOCRE (data not shown)) competitors but did not interact with known cAMP-responsive consensus binding sites (C/EBP and CREB), even at a 100-fold molar excess. The interaction of steroid/thyroid hormone transcription factors with the DNA helix relies upon the conserved 5Ј-PuGGTCA-3Ј hexamer motif (53). Although most of the S/THRs bind as homo-or heterodimers to variably spaced direct or indirect repeats of this motif (57), several of the orphan S/THRs have been shown to bind as monomers to single half-sites (58,59). Previously, S/THRs have been implicated in the cAMP-dependent transcriptional regulation of various cytochrome P450 (steroid hydroxylase) genes (60). Characterization of the transcriptional regulation of the P450c17 and P450c21 genes identified cAMPresponse elements that differ from the classic CRE-motif (61). These sites resemble the classic 5Ј-PuGGTCA-3Ј hexamer orphan receptor half-site and bind various members of the S/THR superfamily (SF-1 and COUP-TF (62-64) and NGFI-B (65), respectively). Together, these results suggest that the factor(s) binding to the CSRh enhancer element belong to the S/THR superfamily and are not related to the CREB/CREM/ATF family of basic leucine zipper transcription factors.
Because the CSRh sequence appears only to contain a single half-site, the effects of elevated cAMP levels and disruption of the 5Ј-PuGGTCA-3Ј sequence motif on CSR-BP⅐CSRh complex formation were examined. First, treatment of cells with forskolin does not appear to alter the formation of the CSR-BP⅐CSRh complex, suggesting that additional elements not identified by EMSAs interact with the CSRh sequence. Second, such factors are required to modulate the cAMP response, and the action of these elements is independent of CSR-BP⅐CSRh complex formation. Third, a mutation that altered three of the six consensus nucleotides inhibited formation of the CSR-BP⅐CSRh complex and significantly reduced the CSRh-mediated transcriptional activation of pCSRh-luc; however, it does not abolish the cAMP-induced suppression of the CSRh enhancer element or the full-length promoter. Together, these results suggest that the orphan receptor half-site is required for maximal CFK2-specific transcriptional activity, but other undefined DNA/protein interactions are required for cAMP responsiveness. Additional experiments will be required to determine which portion of the CSRh sequence is important for mediating the cAMP response and the proteins that mediate the response.
The data presented in this study suggest that signaling cascades in the epiphyseal growth plate capable of modulating the amount of cAMP and the activity of PKA may affect chondrocyte growth and differentiation by regulating fgfr3 gene expression in proliferating chondrocytes. Recently, interactions between the signaling pathways involving Indian hedgehog and PTHrP have been reported in the growth plate (17,45). Coordination of these signals appears to be necessary to maintain the orderly transit of chondrocytes through the growth plate. Animals deficient in either PTHrP or the PTHrP receptor have decreased chondrocyte proliferation (17,18). Ligand-independent activating mutations in FGFR3 (G380R, R248C, or K650E) (13,14) also decreases chondrocyte proliferation (44,66). 3 Therefore, we hypothesize that one pathway by which PTHrP can stimulate chondrocyte proliferation may involve down-regulation of fgfr3 expression. This would link the Indian hedgehog/PTHrP signaling pathway to the FGFR3 signaling pathway in the epiphyseal growth plate. Interestingly, recent in vivo studies have demonstrated that FGFR3 signaling can repress Indian hedgehog activity in the growth plate (44). This completes a potential feedback loop that may coordinate endochondral bone growth.