The Zinc Finger Transcription Factor Gli2 Mediates Bone Morphogenetic Protein 2 Expression in Osteoblasts in Response to Hedgehog Signaling

Abstract

Bone morphogenetic protein 2 (BMP-2) plays a critical role in osteoblast function. In Drosophila, Cubitus interruptus (Ci), which mediates hedgehog signaling, regulates gene expression of dpp, the ortholog of mammalian BMP-2. Null mutation of the transcription factor Gli2, a mammalian homolog of Ci, results in severe skeletal abnormalities in mice. We hypothesize that Gli2 regulates BMP-2 gene transcription and thus osteoblast differentiation. In the present study, we show that overexpression of Gli2 enhances BMP-2 promoter activity and mRNA expression in osteoblast precursor cells. In contrast, knocking down Gli2 expression by Gli2 small interfering RNA or genetic ablation of the Gli2 gene results in significant inhibition of BMP-2 gene expression in osteoblasts. Promoter analyses, including chromatin immunoprecipitation and electrophoretic mobility shift assays, provided direct evidence that Gli2 physically interacts with the BMP-2 promoter. Functional studies showed that Gli2 is required for osteoblast maturation in a BMP-2-dependent manner. Finally, Sonic hedgehog (Shh) stimulates BMP-2 promoter activity and osteoblast differentiation, and the effects of Shh are mediated by Gli2. Taken together, these results indicate that Gli2 mediates hedgehog signaling in osteoblasts and is a powerful activator of BMP-2 gene expression, which is required in turn for normal osteoblast differentiation.

Bone morphogenetic proteins (BMPs), which are structurally related to the transforming growth factor beta superfamily, were originally identified by their capacity to induce ectopic bone formation in rodents (50, 55). Members of the BMP family have diverse functions in embryonic development (26, 57) and have been demonstrated to play crucial roles in osteogenesis (56, 62). Among BMP family members, BMP-2 has been extensively studied for its variety of functions, particularly in embryonic skeletal development and postnatal bone remodeling and bone repair (8, 40).

BMP-2 has been shown to play a major role in chondrogenic and osteogenic differentiation. BMP-2 is involved in almost every aspect of chondrogenesis, including commitment to the chondrogenic lineage and development of the growth plate (11, 51, 59). BMP-2 also promotes commitment of pluripotent mesenchymal cells to the osteoblast lineage by regulating signals that stimulate specific transcriptional programs required for bone formation (7, 39, 43). The bone-inducing function of BMP-2 has also been demonstrated in studies of subcutaneous or intramuscular implantation of BMP-2 into rodents, where BMP-2 induces ectopic cartilage and bone formation in a similar manner to endochondral bone formation (55, 61). This activity has been applied to repair bone defects in humans (4, 20).

BMP-2 is an autocrine and paracrine growth factor. The BMP-2 gene has been mapped to chromosome 3 and contains an 11-kb transcription unit and three exons (12, 21). BMP-2 is expressed from the early stages of embryonic development and in adulthood, primarily in bone-forming tissues (2). During osteoblastic differentiation, BMP-2 mRNA is induced and maintains the sustained phenotype of mature osteoblasts (17, 23). It has been found that BMP-2 expression in bones of aged animals is significantly decreased (34, 49), suggesting that BMP-2 expression may play a role in bone remodeling in aging. A recent human genetic study indicated that polymorphisms of BMP-2 gene expression are linked to a high risk for osteoporosis (47). Previous studies have indicated that the regulation of BMP-2 gene expression during limb morphogenesis and osteoblast differentiation may involve multiple signaling pathways (1, 3, 15, 19, 24, 25, 53), suggesting considerable complexity in BMP-2 gene regulation. Our group has previously characterized the 5′-flanking region of the mouse BMP-2 gene and identified several cis elements in the BMP-2 promoter which are likely required for its transcriptional regulation (12, 13, 14, 19). However, the precise mechanisms responsible for BMP-2 gene regulation during osteoblast differentiation and bone formation are not well elucidated.

Previous studies of Drosophila melanogaster have found that Sonic hedgehog (Shh) signaling, which is important in the development of a wide variety of tissues (29, 52), enhances decapentaplegic (dpp) gene expression by the transcription factor Cubitus interruptus (Ci) (38). dpp is the ortholog of mammalian BMP-2 and -4. In mammals, Shh signaling has been shown to be necessary for skeletogenesis, as deletion of the Shh gene causes severe distal anterior/posterior skeletal abnormalities (9). Shh signaling in vertebrates is mediated by the Gli family of zinc finger proteins, the mammalian homologues of Drosophila Ci. Three Gli family members, Gli1, Gli2, and Gli3, have been identified (28, 32), and these Gli proteins play an essential role in embryonic development by regulating Shh target genes through Gli binding sites (38, 45). Disruption of Gli genes results in diverse developmental defects and abnormalities in multiple tissues and organs (5, 9, 36, 37). The Gli family is thought to have similar functions in mammalian cells to those of Ci. In the absence of Shh, these proteins undergo proteolytic processing, by which Gli3 is cleaved to form C-terminally truncated Gli3, while Gli2 is completely degraded. Available data suggest that the predominant functional form of Gli3 is truncated Gli3, which functions as a transcriptional repressor. In contrast, the Gli2 protein is a major transactivator of many Shh target genes (16, 46, 54). Genetic studies have shown that null mutations of Gli2 and/or Gli3 result in severe defects in skeletal development in mice and humans (6, 22, 35, 36).

Based on this information, we hypothesized that the Shh signaling mediators Gli2 and Gli3 play important roles in regulating BMP-2 gene expression for osteoblast differentiation. In the present study, we determined the function of Gli2 on BMP-2 gene transcription. We found that Gli2 is a powerful transactivator of the BMP-2 gene in vitro and in vivo and that overexpression of Gli2 in osteoblast precursor cells induces osteoblast differentiation. These results provide new insights into the molecular mechanisms responsible for BMP-2 gene regulation during osteogenesis.

MATERIALS AND METHODS

Growth factors and compounds.

Escherichia coli-expressed BMP-2, which was described previously (61), was dissolved in phosphate-buffered saline (PBS)-0.1% bovine serum albumin at 1 mg/ml as a stock solution. Mouse Sonic hedgehog (Shh-N), a product of R&D Systems (Minneapolis, MN), was dissolved in PBS-0.1% bovine serum albumin at 20 μg/ml as a stock solution. The Shh signaling inhibitor cyclopamine and the protein kinase A (PKA) inhibitor KT5720 were purchased from Calbiochem (San Diego, CA) and dissolved in dimethyl sulfoxide at 1 mM as stock solutions.

Cell culture and transfection.

C3H10T1/2 and 2T3 cells were cultured in α-minimal essential medium (α-MEM), and C2C12 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS). cDNA expression plasmids were transiently transfected into cells by using Lipofectamine Plus reagents following the manufacturer's instructions (Invitrogen, Carlsbad, CA).

DNA constructs and promoter mutation.

For construction of mouse BMP-2 promoter reporters, a series of deletion cassettes of the 5′ promoter region of the mouse BMP-2 gene (16, 19), including −2712/+165, −1997/+165, −1803/+165, −838/+165, −310/+165, −212/+165, and −150/+165, were linked to firefly luciferase in the pGL3 vector. Mutant promoter constructs (ΔBS1, ΔBS2, and ΔBS3) were made in which the putative Gli binding sites in the BMP-2 promoter were mutated, using the GeneEditor in vitro site-directed mutagenesis system (Promega) and the mutant oligonucleotides 5′-AAGCCAGCAGGCACGCGTCAAGGTGGAGTAAC, 5′-CTTCGGAGCGCGAGGATCCCGGTTTGGCAACCCGAG, and 5′-GAGCGCTGATGGGGGCCCGCCAGAGTCAGGC for BS1, BS2, and BS3, respectively. A mouse BMP-4 promoter reporter construct was made by inserting a 3.6-kb BMP-4 5′-flanking sequence into the pGL3 vector. The Gli-responsive reporter construct 8×3′Gli-BSδ5/LucII, which contains multiple copies of Gli response elements, and the expression plasmids for Gli1, Gli2, and Gli3 were kindly provided by Hiroshi Sasaki (46). Gli1 was cloned into pcDNA3. Gli2, Gli2ΔN2, and Gli2ΔC4 were cloned into the pcDNA3.1/His vector. Gli3 and trGli3 were cloned into the pact vector under the control of the β-actin promoter.

Promoter reporter luciferase assay.

C2C12 cells were plated into 24-well plates at 4 × 10⁴ cells per well in DMEM with 10% FCS 18 to 24 h prior to transfection. Cells were incubated for 4 h at 37°C with 250 μl of Opti-MEM transfection solution containing Lipofectamine Plus reagent, 0.5 μg of reporter plasmids, 0.1 μg of pSV-β-galactosidase (β-Gal) expression vector (Promega, Madison, WI), and 0 to 0.5 μg of different Gli expression constructs. After 4 h of incubation, 250 μl of fresh DMEM containing 20% FCS was added. The cells were cultured for 24 to 48 h and lysed with 100 μl of reporter lysis buffer (Promega). The luciferase activities of cell lysates were measured by a luciferase assay kit (Promega) and normalized with β-Gal activity (63).

Reverse transcription-PCR (RT-PCR).

Total RNAs of the cells receiving different treatments were extracted using the RNA STAT-60 reagent (TEL-TEST, Inc., Friendswood, TX). The purified RNAs were reverse transcribed into cDNAs by using the SuperScript III first-strand synthesis system (Invitrogen), and the synthesized cDNAs were amplified by PCR for 35 cycles (94°C for 1 min, 56°C for 1 min, and 72°C for 1 min). The mouse primers used were as follows: for Ptc1, 5′-AACAAAAATTCAACCAAACCTC and 5′-TGTCTTCATTCCAGTTGATGTG; for Smo, 5′-TGCCACCAGAAGAACAAGCCA and 5′-GCCTCCATTAGGTTAGTGCGG; for BMP-2, 5′-TGAGGATTAGCAGGTCTTTG and 5′-CACAACCATGTCCTGATAAT; for Gli2, 5′-GATATCTCCTTGATGCGACTT and 5′-TGCTACTGCTGCTGAGTTGGG; for Gli3, 5′-TCCATGGCTCTCTACCACATC and 5′-GTGGCAGCTGAGGGAAGGAT; and for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-CACCATGGAGAAGGCCGGG and 5′-GACGGACACATTGGGGGTAG.

Real-time PCR.

Quantitative PCR was performed using the QuantiTect SYBR green PCR system (QIAGEN, Valencia, CA). Mouse BMP-2 primers (5′-TGAGGATTAGCAGGTCTTTG and 5′-CACAACCATGTCCTGATAA) were added to 25 μl of QuantiTect SYBR green PCR master mix containing test cDNA, PCR was performed, and the quantitative real-time PCR value was analyzed using an ABI sequence detection system (Applied Biosystems, CA) following the manufacturer's instructions. The BMP-2 mRNA signal was normalized with GAPDH.

RNA interference (RNAi).

Both sense and antisense DNA oligonucleotides (56 nucleotides), each containing two 21-nucleotide copies of palindromic sequence identical to that of Gli2 cDNA, were designed with the siRNA Wizard program (Invivogen, San Diego, CA). The synthesized and annealed short double-stranded DNA was inserted into a small interfering RNA (siRNA)-expressing vector carrying a neomycin resistance selection cassette (Invivogen). The psiRNA-Gli2 plasmid, which expresses Gli2 siRNA hairpins in cells, was transfected into the target osteoblasts with Lipofectamine Plus reagents (Invitrogen), and the cells were selected with G418. Reduced expression of Gli2 in the cell colonies was confirmed by both RT-PCR and Western blotting.

Gli2 null mutant mice.

Gli2^wt/zfd heterozygous knockout mice were obtained from Alexandra Joyner's lab. Mice were maintained as heterozygotes and crossed to generate mutant combinations. Fetal mice (embryonic day 20.5 [E20.5]) were genotyped by PCR analysis of tail genomic DNA with Gli2 sense (5′-AAACAAAGCTCCTGTACACG), Gli2 antisense (5′-CACCCCAAAGCATGTGTTTT), and pPNT (5′-ATGCCTGCTCTTTACTGAAG) primers (36). The homozygous mutants (Gli2^zfd/zfd) and their wild-type littermates (Gli2^wt/wt) were used for in situ hybridization and immunohistochemistry (IHC).

Immunoblotting.

In RNAi studies, C2C12 cells expressing Gli2 siRNA were cultured in six-well plates and transfected with His-tagged Gli2 expression plasmid and cultures for 30 h. In PKA experiments, C3H10T1/2 cells were transfected with His-Gli2 and treated with compound KT5720 at 2 μM for 30 h. After that, the cells were lysed with sodium dodecyl sulfate (SDS)-radioimmunoprecipitation assay lysate buffer containing the protease inhibitors phenylmethylsulfonyl fluoride (1 mM), aprotinin (10 μg/ml), and leupeptin (10 μg/ml). SDS sample buffer (2×) containing 0.5 M β-mercaptoethanol was added. The sample was loaded into a Mini-Protein II SDS-PAGE Ready gel (Bio-Rad, Hercules, CA). Proteins were transblotted from the gel onto a polyvinylidene difluoride membrane (Bio-Rad) in transblotting buffer (25 mM Tris, 192 mM glycine, and 20% [vol/vol] methanol, pH 8.3) at 4°C for 1 h. The membrane was blocked with 5% milk in Tris-buffered saline with 0.1% Tween 20 (TBS-T) for 1 h at room temperature and incubated with rabbit Omni Probe anti-His antibody (Santa Cruz, CA) diluted 1:2,000 in 5% milk in TBS-T at 4°C overnight. After being washed, the membrane was incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit immunoglobulin G (IgG) antibody (Amersham Biosciences, Buckinghamshire, United Kingdom) diluted 1:5,000 at room temperature for 1 h. The membrane was then washed six times with TBS-T for 5 min each time. Immunostaining was detected using an enhanced chemiluminescence (ECL) system (Amersham) with exposure on X-ray film. The His-Gli2 protein signal was normalized to the GAPDH protein level.

Immunohistochemistry.

Formalin-fixed, paraffin-embedded embryos were sectioned at 4 to 6 μm and applied to slides. After deparaffinization, the slides were treated with 10 mM sodium citrate, pH 6.0, at 95°C for 5 min and then immunostained using the ImmunoCruz staining system (Santa Cruz), which utilizes an HRP-streptavidin complex. Goat anti-BMP-2 antibody (sc-6895; Santa Cruz) was used as the first antibody at a 1:200 dilution. The visible brown staining of the HRP substrate was analyzed by the automated image analysis program Image-Pro Plus (Media Cybernetics, Silver Spring, MD).

In situ hybridization.

Embryos were fixed overnight in fresh 10% paraformaldehyde in RNase-free PBS, transferred via an ethanol series to 100% ethanol, followed by methanol and xylene, and finally embedded in wax. Ten-micrometer-thick sections were placed on microscope slides, dewaxed, and utilized for in situ hybridization. A 1.0-kb mouse BMP-2 RNA probe was hybridized to these sections, and the hybridized tissues were subsequently stained with HRP as described previously (19). The stained images were captured with a Nikon E400 microscope.

ALP activity assay.

C2C12 cells, 2T3 cells, or primary calvarial cells isolated from newborn Gli2^zfd/zfd mice were seeded into 48-well plates in DMEM or α-MEM with 10% FCS. The cells were treated with Shh or cyclopamine or transfected with Gli2 expression plasmid for 2 to 5 days in medium with 2.5% FCS. The cells were then lysed with 0.05% Triton X-100 buffer. The cell lysate was analyzed for alkaline phosphatase (ALP) activity in a 96-well plate. Ten microliters of lysate was incubated with 90 μl of fresh AMP solution containing p-nitrophenol phosphate substrate at 37°C for 30 to 60 min. One hundred microliters of 0.5 N NaOH was added to stop the reaction. The plates were read at 405 nm. ALP activity was determined by using a p-nitrophenol standard curve and normalized to the amount of cell protein (16, 63, 64).

Mineralized matrix formation.

2T3 cells or calvarial cells from Gli2^zfd/zfd mice were cultured in 12-well plates in α-MEM with 10% FCS, and the cells were treated with BMP-2 or transfected with Gli2 expression vector. When the cells reached confluence, the medium was changed, and 100 μg/ml ascorbic acid and 5 mM β-glycerol phosphate were added. The medium was changed every other day. On day 15, the cultures were harvested. Cells were washed, fixed in phosphate-buffered formalin for 10 min, and then washed with water. For von Kossa staining, a 2% silver nitrate solution was added, and the plate was exposed to sunlight for 20 min and then rinsed with water. Five percent sodium thiosulfate was added for 3 min, followed by a rinse. The modified van Gieson stain was then used as a counterstain after the von Kossa stain. The area of von Kossa-stained matrix was quantified by automated image analysis using a video analysis program (62, 64).

ChIP.

C3H10T1/2 and C2C12 cells (1 × 10⁶) cultured in 100-mm petri dishes were transfected with Gli2 expression plasmid for 24 h and treated with 1% formaldehyde for 10 min to cross-link chromatin. Chromatin immunoprecipitation (ChIP) assays were performed following the protocol of a ChIP kit (Upstate, MA). Briefly, the cells were scraped and sonicated on ice to shear chromatin DNA down to 0.2- to 1.0-kb fragments. The sonicated cell supernatant was diluted and precleared with a protein A agarose-salmon sperm DNA slurry. The protein A bead pellet was discharged, and anti-Gli2 antibody (a gift provided by Philip Beachy) was added to the supernatant at 4°C overnight with rotation, followed by incubation with fresh protein A agarose beads for 1 h at 4°C for precipitation. The specific protein-DNA complex was reversely cross-linked, and DNA fragments were purified. With these DNAs as templates, PCRs were performed using primer set 1 (P1, TTTTAGCAGCACCTCTCTG; and P2, AACCATTGCTTCTGTCATC) and primer set 2 (P3, CTGCCACAAAAGACACTTG; and P4, AAGGGCGCAGCGAAGATCTG). The input was the positive control. No IgG or normal goat IgG (sc-2028; Santa Cruz) was the negative control.

EMSA.

Approximately 2 × 10⁶ to 4 × 10⁶ C3H10T1/2 cells transfected with His-Gli2 were collected by digestion, and nuclear extracts were prepared using an NE-PER nuclear extract reagent kit (Pierce, IL). Oligonucleotide DNA probes containing wild-type or mutant Gli binding sites were end labeled with biotin by using a biotin 3′-end DNA labeling kit (Pierce). Electrophoresis mobility shift assays (EMSAs) were performed using a chemiluminescent EMSA kit (Pierce) according to the manufacturer's instructions. Briefly, DNA-protein binding reactions were carried out by incubating 2- to 3-μl nuclear protein extracts with 20 fmol of labeled and/or 100- to 200-fold unlabeled probe in a 20-μl volume at room temperature for 20 min. The 20-μl binding reaction mixes were then subjected to gel electrophoresis on a 5% polyacrylamide gel (Bio-Rad) and transferred to a nylon membrane. Biotin-labeled DNA was detected by using a streptavidin-HRP conjugate and a chemiluminescent substrate. To examine supershifts of DNA-protein complexes, Omni Probe anti-His antibody (Santa Cruz) was added to DNA-protein reactions and loaded into the gel.

RESULTS

Overexpression of Gli2 promotes osteoblast differentiation.

Previous studies have demonstrated that the Shh signaling pathway plays an important role in the commitment of mesenchymal cells to the osteoblast lineage (30, 60). Since it is known that Gli2 is an important transcriptional activator in the Shh signaling pathway, we investigated the effects of Gli2 on osteoblast differentiation. Shh signaling is initiated by the binding of Shh to its receptor Patched (Ptc), which releases the receptor Smoothened (Smo) and thus activates an intracellular signaling cascade. We examined the gene expression of Shh receptors in C2C12, 2T3, and C3H10T1/2 osteoblast precursor cells. The RT-PCR results indicate that the Ptc1 and Smo genes are expressed in these cells (Fig. 1A). We then examined the effects of Shh and Gli2 on the osteoblastic maturation of these mesenchymal cells. C2C12 cells were treated with Shh at 200 ng/ml for 5 days, and ALP activity was determined. We found that Shh increased the ALP activity in these cells. The addition of cyclopamine (1 μM), an inhibitor of the Shh receptor, abolished the Shh-enhanced ALP activity (Fig. 1B). We utilized immortalized mouse osteoblast precursor 2T3 cells, which form mineralized bone nodules in culture (18), to assess the ability of Gli2 to increase mineralized bone nodule formation. von Kossa staining showed that transfection of 2T3 cells with Gli2 not only increased mineralized bone nodule formation at day 15 in the absence of BMP-2 but also enhanced BMP-2-induced mineralized matrix formation (Fig. 1C). Importantly, we further demonstrated that the Gli2-mediated enhancement of ALP activity in C2C12 cells was inhibited by the addition of Noggin to the cultures (Fig. 1D). Noggin is a known natural antagonist of BMP-2 and -4 (58). These results suggest that Gli2, as a mediator of Shh signaling, induces osteoblast differentiation and that this effect is BMP dependent. The data are consistent with the notion that the transcription factor Gli2 up-regulates BMP gene expression, which in turn regulates osteoblast differentiation.

FIG. 1.

Gli2 promotes osteoblast differentiation through the BMP-2 pathway. (A) Gene expression of hedgehog receptors in osteoblast cells. Total RNA was purified from C3H10T1/2, C2C12, and 2T3 cells, and RT-PCR was performed with primers for the receptors of Ptc1 and Smo. (B) Effects of Shh on ALP activity. C2C12 cells were plated in 96-well plates and treated with Shh at 200 ng/ml in the presence or absence of cyclopamine at 1 μM for 5 days. The cell lysates were analyzed for ALP activity using Sigma ALP testing reagents. The ALP activity was normalized to the amount of cell protein. *, P < 0.01 between vehicle and cyclopamine. (C) Effect of Gli2 on mineralized matrix formation. 2T3 cells cultured in 12-well plates (four parallel wells of each) were transfected with Gli2 expression plasmid or empty vector and treated with BMP-2 (100 ng/ml). The cells were cultured in α-MEM containing 100 μg/ml ascorbic acid and 5 mM β-glycerol phosphate for 15 days. The mineralized matrix formation was examined by von Kossa/van Gieson staining. (D) Effect of Noggin on Gli2-enhanced ALP activity. C2C12 cells were transfected with Gli2 plasmid or empty vector and treated with 500 ng/ml Noggin for 48 h. The ALP activity in the cell lysate was measured by an ALP assay kit (Sigma) and normalized to the amount of cell protein. #, P < 0.05; *, P < 0.01 (between vehicle and Noggin).

Gli2 is a transactivator for BMP-2 gene transcription.

To test the hypothesis that BMP-2 is regulated by Gli2, we examined the effects of Gli2 on the transcriptional activities of the BMP-2 and BMP-4 promoters. We have previously described the mouse BMP-2 and BMP-4 genomic structures and promoters (12, 13) and promoter reporter constructs for both (−2712/+165-Luc and 3.6-Luc). C2C12 cells were cotransfected with the −2712/+165-Luc BMP-2 reporter or 3.6-Luc BMP-4 reporter construct and the Gli2 expression plasmid. The promoter activity was determined by the luciferase activity and was normalized to β-Gal activity. We found that Gli2 markedly stimulated BMP-2 promoter activity but had no effect on the BMP-4 promoter (Fig. 2A). In the following experiment, we found that Gli2, at a range of 0 to ∼500 ng/well, increased BMP-2 promoter activity in a dose-dependent manner, with maximal enhancement of up to 10-fold (Fig. 2B). To determine if Gli2 regulates BMP-2 mRNA expression, we performed quantitative PCR experiments. C2C12 and 2T3 cells were transiently transfected with Gli2 plasmid or empty vector for 30 h. BMP-2 mRNA expression was analyzed by real-time RT-PCR using mouse BMP-2 primers. Figures 2C and D show that the BMP-2 mRNA level was significantly enhanced by Gli2 transfection. These results indicate that Gli2 specifically activates BMP-2 gene transcription and mRNA expression.

FIG. 2.

Gli2 enhances BMP-2 transcription in osteoblasts. (A) BMP promoter reporter assays. The luciferase reporters −2712/+165-Luc and BMP4-3.6-Luc were cotransfected with Gli2 expression plasmid into C2C12 cells. The relative promoter activity was measured after 30 h as the luciferase activity and normalized to β-Gal activity. (B) Dose-dependent enhancement by Gli2 of BMP-2 promoter activity. C2C12 cells carrying a BMP-2 promoter reporter gene, −2712/+165-Luc, were cultured in 24-well plates and transfected with Gli2 expression plasmid at a dose range from 20 to 500 ng/well or with empty vector pCDNA3 at the same range. The promoter activity was determined as described above. (C and D) Effects of Gli2 on BMP-2 mRNA expression. Total RNAs were extracted from C2C12 cells and 2T3 cells after the transfection of plasmids expressing Gli2 or empty vector for 30 h and were reverse transcribed to cDNAs. Real-time PCR (B) of C2C12 cells was performed with mouse BMP-2 primers and the SYBR green dye probe, which was detected with an ABI sequence detection system (Applied Biosystems, CA). The BMP-2 mRNA level was normalized to the GAPDH level. The BMP-2 PCR products from C2C12 and 2T3 cells were also analyzed by gel assays (C). *, P < 0.01.

The Gli family members, namely, Gli1, Gli2, and Gli3, are characterized by a highly conserved zinc finger domain (ZF), which is responsible for binding to target DNA. The carboxyl-terminal part of the ZF is a DNA activation domain, while the amino-terminal region is a repressor (46). The Gli2 and Gli3 proteins are modulated by proteolytic processing, which leads to complete degradation of Gli2 and the production of a C-terminally truncated Gli3 fragment. In this study, we also compared the transactivation activity of Gli2 in BMP-2 gene transcription with those of other Gli proteins. First, we determined the transactivation of Gli proteins with a known specific Gli-responsive reporter gene, 8x3′Gli-BSδ5/LucII, in C2C12 cells (46). Compared with the individual empty vector, we found that full-length Gli1 and Gli2 are powerful activators of the Gli reporter, while Gli3 is a weak activator. In similar studies, we also found that N-terminally truncated Gli2 (Gli2ΔN2) is more potent than full-length Gli2, an effect which is due to deletion of the repressor domain. In contrast, C-terminally truncated Gli2 (Gli2ΔC4) and Gli3 (trGli3) repressed the Gli reporter (Fig. 3B). Next, we examined the functions of these Gli proteins in BMP-2 gene transcription. Experiments with the mouse BMP-2 promoter-luciferase construct showed that Gli2 is the only member of the Gli family which enhances BMP-2 promoter activity in C2C12 cells (Fig. 3C). To verify the effects of the Gli proteins on BMP-2 expression, we transfected the expression plasmids and examined endogenous BMP-2 mRNA levels in osteoblastic 2T3 cells. Consistently, we found that Gli1 does not have an effect on BMP-2 gene regulation, while Gli2 enhances and trGli3 inhibits endogenous BMP-2 expression compared with individual vectors (Fig. 3D). These data suggest that Gli2 is a major activator of BMP-2 gene transcription among the Gli family.

FIG. 3.

Gli2 is a major activator of the BMP-2 promoter. (A) Sketches of Gli proteins. All full-length Gli proteins, Gli1, Gli2, and Gli3, have an N-terminal repressive domain (dark gray boxes), a central zinc finger DNA binding domain (gray boxes), and a C-terminal activation domain (open boxes). (B) Transactivation of Gli proteins on Gli-responsive reporter gene. The reporter gene, 8x3′Gli BSδ5/LucII, containing multiple copies of Gli response elements (left open box) and a TATA box, was cotransfected with full-length and truncated Gli expression plasmids or their empty vectors into osteoblast precursor C2C12 cells. Thirty hours after transfection, the reporter activity was determined by fluorescence reading of luciferase activity, which was normalized with the activity of cotransfected β-Gal. (C) Transactivation of Gli proteins on BMP-2 promoter reporter gene. The 5′ promoter region of the mouse BMP-2 gene spanning positions −2712 to +165 (gray box) was cloned in front of the luciferase reporter. The reporter gene was cotransfected with Gli expression plasmids into C2C12 cells, and the promoter activity was analyzed as described above. (D) Endogenous BMP-2 mRNA expression. 2T3 cells were transfected with Gli expression constructs and their empty vectors for 30 h. BMP-2 mRNA was examined by RT-PCR with GAPDH controls.

Loss of function of Gli2 inhibits BMP-2 gene expression and subsequent osteoblast differentiation.

To further confirm that Gli2 positively regulates BMP-2 gene expression, we carried out loss-of-function studies both in vitro and in vivo. In an RNA interference study, a DNA sequence corresponding to the sequence of Gli2 RNA (AAUGAUGCCAACCAGAACAAG) was synthesized. The annealed double DNA strands of this sequence were cloned into an siRNA expression vector. C2C12 and 2T3 cells were stably transfected with the Gli2 siRNA-expressing vector, pGli2-siRNA, and Gli2 gene expression was examined. The RT-PCR results indicated that Gli2 siRNA substantially reduced Gli2 mRNA in both C2C12 and 2T3 cells (Fig. 4A). The data also showed that Gli3 mRNA expression was not affected by Gli2 siRNA, indicating that the RNA interference was specific to Gli2 mRNA. Gli2 knockdown was also further validated by a Western blot assay with C2C12 cells, in which ectopically expressed Gli2 was markedly inhibited (Fig. 4B). Following validation of the Gli2 gene knockdown, we determined the effects of Gli2 siRNA on BMP-2 gene expression. By using quantitative RT-PCR, we found that Gli2 siRNA significantly reduced BMP-2 mRNA levels in C2C12 cells (>60%) compared to those with a control siRNA vector (Fig. 4C). The Gli2 siRNA-mediated inhibition of BMP-2 transcription was also observed in BMP-2 promoter assays with both C2C12 and 2T3 cells. As shown in Fig. 4D, Gli2 siRNA markedly inhibited BMP-2 promoter activity, with or without Gli2 transfection. The results indicate that knocking down the Gli2 gene significantly decreased BMP-2 gene transcription and mRNA expression in osteoblasts. Finally, we examined the effect of Gli2 siRNA on osteoblast differentiation and found that ALP activity was significantly attenuated by Gli2 siRNA compared to that with the control siRNA in both C2C12 and 2T3 cells (Fig. 4E).

FIG. 4.

Gli2 gene knockdown reduces BMP-2 gene expression in osteoblasts. (A and B) Gli2 siRNA knocks down Gli2 gene expression in osteoblast cells. The siRNA expression vector expressing Gli2 siRNA hairpins and a control siRNA vector were stably transfected into C2C12 and 2T3 cells. Total RNA was extracted from the stable cell colonies, and the mRNAs of Gli2, Gli3, and GAPDH were examined by RT-PCR (A). The Gli2 knockdown was also confirmed by Western blotting, in which the cotransfected His-Gli2 in the cell lysate of C2C12 cells was blotted by Omni Probe anti-His antibody (B). (C) Quantitative real-time RT-PCR of BMP-2. Total RNA was purified from the cell lysates of C2C12 cells carrying Gli2 siRNA or control siRNA. The mouse BMP-2 primers were used to amplify BMP-2 mRNA by real-time PCR with an ABI sequence detection system. The BMP-2 mRNA signal was normalized with GAPDH. *, P < 0.01. (D) BMP-2 promoter activity of Gli2-silent cells. The C2C12 and 2T3 cells that expressed Gli2 siRNA were cotransfected with a BMP-2 promoter reporter gene, −2712/+165-Luc, and the Gli2 expression plasmid or its empty vector. Thirty-six hours later, the cell lysate was analyzed for relative BMP-2 promoter activity, which was normalized with β-Gal activity. *, P < 0.01 between Gli2 siRNA and control. (E) Effects of Gli2 RNAi on ALP activity of osteoblasts. The C2C12 and 2T3 cells with Gli2 gene knockdown were cultured in 96-well plates for 2 days, and ALP activity was measured as described in the legend to Fig. 1. *, P < 0.01 between Gli2 siRNA and control.

We next performed loss-of-function studies using Gli2 knockout mice. Gli2^zfd/zfd mice are functionally null mutant mice in which part of the zinc finger DNA binding domain of the Gli2 gene has been removed. A homozygous Gli2 knockout in mice (Gli2^zfd/zfd) is perinatally lethal, causing severe abnormalities in cartilage and bone development (22, 35, 36). We examined skeletal BMP-2 mRNA expression by in situ hybridization with these mutants. E20.5 embryos were processed and hybridized with a mouse BMP-2 probe. We found that the hybridized BMP-2 mRNA signal was markedly decreased in the area of the growth plate and the trabecular bone of the tibias of Gli2^zfd/zfd embryos compared to that in wild-type littermates (Fig. 5A). We then performed BMP-2 IHC on these homozygous Gli2 embryos to examine the expression of BMP-2 protein. Consistent with the results from in situ hybridization, we found that BMP-2 protein expression in the growth plate, trabecular bones (Fig. 5B), and vertebrae (Fig. 5C) was decreased compared with that in wild-type controls. We also examined BMP-2 gene expression at an early embryonic stage. The reduced IHC stains of BMP-2 in the areas of developing spine of E10.5 embryos indicated that BMP-2 gene expression is also decreased in the Gli2 null mutants during an early embryonic stage when chondrocytes and osteoblasts start to form (Fig. 5D). To quantify BMP-2 gene expression in Gli2 null mutants, the primary osteoblasts were dispersed from the calvariae of wild-type and Gli2^zfd/zfd embryos (E20.5), and BMP-2 mRNA expression was measured by real-time RT-PCR. The results indicated that the BMP-2 mRNA level was significantly reduced in Gli2^zfd/zfd osteoblasts (Fig. 5E). ALP activity and mineralized matrix formation were also determined using the primary osteoblasts. Similar to the results obtained from the Gli2 RNAi experiment, we found that ALP activity was significantly reduced in the primary Gli2^zfd/zfd osteoblasts compared to that in wild-type primary cells (Fig. 5F). In the assay of mineralized matrix formation, we found that the wild-type primary calvarial osteoblast cells formed a slight mineralized matrix in 15 days which was greatly enhanced by BMP-2. However, in Gli2-deficient cells, both spontaneous and BMP-2-induced mineralized matrixes were inhibited (Fig. 5G). The results of Gli2 loss-of-function experiments further confirmed a role for Gli2 in the regulation of BMP-2 gene expression and osteoblast functions.

FIG. 5.

Gli2 knockout reduces BMP-2 gene expression in bone. (A) In situ hybridization of BMP-2 on tibias of Gli2 null embryos. The embryos were obtained at E20.5 after breeding of the heterozygous Gli2 null mice and were genotyped. The processed embryos were hybridized with a mouse BMP-2 RNA probe and subsequently stained with HRP. The BMP-2 mRNA signal in the area of the growth plate and the trabecular bone of the tibia was analyzed. (B, C, D) IHC of BMP-2 in Gli2 null embryos. Both Gli2^+/+ and Gli2^zfd/zfd embryos at the ages of E10.5 (D) (n = 6 for wt and n = 5 for null group) and E20.5 (B and C) (n = 7 for wt and n = 5 for null group) were processed for IHC. The paraffin-embedded slices were incubated with goat anti-BMP-2 antibody and stained with a goat immunostaining kit (Santa Cruz). Brown BMP-2 staining is indicated by arrows. Panels B and C show staining of tibias and vertebrae (E20.5), respectively, at magnifications of ×100 (c, d, g, h) and ×400 (e, f, i, j). Panel D shows staining in the area of the developing spine (E10.5) at a magnification of ×200. (E) Quantitative real-time RT-PCR of BMP-2 from Gli2-deficient osteoblasts. Primary osteoblast cells were isolated from the calvariae of Gli2^+/+ and Gli2^zfd/zfd embryos (E20.5) and cultured in α-MEM. Total RNAs were extracted from the cell lysates, and the BMP-2 mRNA level was quantitated by real-time RT-PCR with mouse BMP-2 primers and normalized with GAPDH. *, P < 0.01. (E and F) Effects of Gli2 null mutation on osteoblast differentiation, as determined by ALP activity (E) and mineralized matrix formation (F). The primary osteoblast cells were isolated from the calvariae of Gli2^+/+ and Gli2^zfd/zfd embryos (E20.5) and cultured in α-MEM. The ALP activity and mineralized matrix formation (von Kossa staining) were examined at 48 h and 15 days, respectively, as described in the legend to Fig. 1. *, P < 0.01.

Gli2 binds to specific sites in the BMP-2 promoter.

Previous investigations have demonstrated that Gli proteins transactivate target genes through a DNA consensus sequence, GACCACCCA. This generic Gli binding site has been identified in several Shh/Gli target genes (38, 45). To identify the responsive elements for Gli2 in the BMP-2 promoter, we performed promoter deletion analysis, ChIP assays, and EMSAs. A series of BMP-2 promoter constructs were tested in the present studies. These deletion constructs, including −2712/+165, −1997/+165, −969/+165, −838/+165, −310/+165, −212/+165, and −150/+165-Luc, were cotransfected with the Gli2 expression plasmid into C2C12 cells. β-Gal-normalized relative activities of the BMP-2 promoters demonstrated declines within the two promoter regions, −2712/−1997 and −310/−212 (Fig. 6A). The results suggest that these areas are the potential Gli2 interacting regions containing positive cis-regulating elements. Sequence analysis demonstrated that three putative Gli binding sites are located in these two regions, i.e., BS1 (−2528/−2537), BS2 (−302/−310), and BS3 (−198/−207) (Fig. 6B).

FIG. 6.

Gli2 transactivation is mediated through Gli binding sites in the BMP-2 promoter. (A) Promoter deletion assay. A series of deletion fragments of the BMP-2 promoter, from −2712/+165 to −150/+165 (open boxes, promoter region; dark gray boxes, exon 1), were linked to the pGL3 luciferase (gray boxes) reporter. The deletion reporter genes were cotransfected with Gli2 expression plasmid into C2C12 cells. The increase in the promoter activity responding to Gli2 was determined relative to that in cells with empty vector. The β-Gal activity was used to normalize the luciferase reading. (B) Scheme of Gli binding sites (Gli-BS) in the BMP-2 promoter. The sequence of the BMP-2 promoter from −2712 to +165 was analyzed for Gli binding elements, using the GenePro sequence analysis program. The gray box shows exon 1, and the dotted boxes represent the locations of three Gli binding sites. P1, P2, P3, and P4 are PCR primers for ChIP assays. (C) ChIP. C2C12 cells (left) and C3H10T1/2 cells (right) were transfected with Gli2 expression vector for 24 h and treated with 1% formaldehyde to cross-link chromatin. The harvested cells were sonicated to shear chromatin down to 0.2- to 1.0-kb fragments. Anti-Gli2 antibody and protein A agarose were added for precipitation. The specific protein-DNA complex was reversely cross-linked, and DNA fragments were purified. With these DNA templates, PCR was performed using the P1/P2 or P3/P4 primers. Input, positive control. No IgG or mouse IgG was the negative control. (D) EMSA. The ∼30-bp DNA oligonucleotide probes containing BS1, BS2, BS3, and mutant binding sites and their flanking sequences, identical to those in the BMP-2 promoter, were labeled with biotin and incubated with the nuclear proteins extracted from Gli2-transfected C3H10T1/2 cells as well as with anti-His antibody. The DNA-protein binding reactions were recognized by gel shifts and supershifts. A 200-fold excess of unlabeled probes was added to observe specific competition (lanes 1 to 3, 6 to 8, and 11 to 13). With mutant probes, the binding ability was analyzed by interaction with nuclear extract (lanes 4, 5, 9, 10, 14, and 15). (E) Reporter assay with mutant BMP-2 promoter. The promoter reporter constructs containing mutant Gli binding sites (ΔBS1, ΔBS2, and ΔBS3) in the BMP-2 promoter were cotransfected with Gli2 expression vector into C2C12 cells for 30 h. Their luciferase activities were determined and normalized with β-Gal. *, P < 0.01 for inhibition of Gli2 transactivation on the mutant promoter compared with that on the wild-type promoter.

To determine whether the Gli2 protein interacts directly with these putative binding regions in the BMP-2 promoter, we performed ChIP assays. Chromatin from C2C12 and C3H10T1/2 cells transfected with Gli2 expression vector was cross-linked to stabilize protein-DNA complexes and sheared by sonication into small fragments. The protein-DNA complexes were then immunoprecipitated by incubation with anti-Gli2 antibody and protein A agarose beads. De-cross-linked and purified DNA was used as a template for PCR, using the primer sets P1/P2 and P3/P4 to amplify two fragments (−2567/−2233 and −353/−133) of the mouse BMP-2 5′-flanking promoter. Two PCR products, of 334 bp and 220 bp, were amplified with the P1/P2 and P3/P4 primers, respectively, from the DNA fragments isolated from the protein-DNA complexes which were specifically immunoprecipitated by anti-Gli2 antibody. Total DNA before immunoprecipitation was used as a positive control for PCR, and the negative controls were set up as IgG or no antibody (Fig. 6C). These results indicate that the Gli2 protein physically binds with the BMP-2 promoter at the specific regions which contain putative Gli-responsive elements.

In EMSA experiments, about 30-bp DNA probes for the BMP-2 promoter containing the sequence of each putative Gli binding site or mutant binding site were synthesized and, after being labeled with biotin, were incubated with nuclear proteins extracted from osteoblast precursor C3H10T1/2 cells in which His-Gli2 was ectopically expressed. By using a gel shift assay, we found that Gli2 proteins bound directly to each probe (BS1, BS2, and BS3), as shown by shifted bands which were further supershifted by adding anti-His antibody, and an excess of unlabeled probes abolished both shifted and supershifted bands (Fig. 6D, lanes 1 to 3, 6 to 8, and 11 to 13). Mutations of these binding sites abolished the binding abilities of the probes with nuclear extracts, as shown with mutant probes ΔBS1, ΔBS2, and ΔBS3 (Fig. 6D, lanes 4, 5, 9, 10, 14, and 15). These results indicated that Gli2 interacts directly with the BMP-2 promoter through these Gli binding sites, which was further confirmed by the following promoter mutation experiments. We mutated BS1, BS2, and BS3 in the promoter and tested their responsiveness to Gli2. We found that individual mutation of each Gli binding site caused a partial inhibition of promoter activity (Fig. 6E), suggesting that all of these elements have a contribution to and may be required for Shh/Gli2 activation. This needs to be further verified by an assay of combined mutations.

Gli2 mediates the effects of Shh on BMP-2 gene transcription.

Gli proteins transduce Shh signals to target genes, and Gli2 has been shown to be an important activator in this pathway in many systems (10, 33, 46). In this study, we investigated the effect of Shh on BMP-2 gene expression in osteoblast precursor cells and determined whether this Shh function is mediated by Gli2. First, we examined the effects of Shh on BMP-2 promoter activity. C2C12 cells were transfected with the BMP-2 promoter reporter gene (−2712/+165-luc) and treated with Shh (200 ng/ml) for 48 h, with or without ectopically forced expression of Gli2. We found that the treatment of Shh significantly increased both basal and Gli2-enhanced BMP-2 promoter activity (Fig. 7A). In addition, pretreatment with cyclopamine (2 μM), a hedgehog signaling inhibitor, not only reduced basal BMP-2 promoter activity but also attenuated Shh-induced BMP-2 promoter activity (Fig. 7B). We also determined the effect of Gli2 RNAi on the Shh-induced enhancement of BMP-2 transcription in C2C12 cells. Knocking down Gli2 expression eliminated the enhancing effect of Shh on BMP-2 promoter activity (Fig. 7C). This suggests that the induction of BMP-2 expression by Shh is mediated by Gli2. We have previously shown that trGli3 is a powerful negative regulator of BMP-2 gene expression (16). Thus, we investigated the interaction of the Shh signaling mediators Gli2 and trGli3 in the regulation of BMP-2 gene transcription. When Gli2 and trGli3 were cotransfected, the Gli2-mediated enhancement of BMP-2 promoter activity was significantly antagonized by the repressor form of Gli3 (Fig. 7D). The activities of both Gli2 and Gli3 are known to be down-regulated by PKA phosphorylation, leading to proteolytic processing in the absence of the Shh signal. We next examined the effects of PKA on Gli2 regulation in osteoblasts. A Western blot of C3H10T1/2 cells showed that treatment with the PKA-specific inhibitor KT5720 substantially increased the Gli2 protein level (Fig. 7E). Consistently, in the BMP-2 promoter assay, we found that overexpression of PKA markedly attenuated Gli2-mediated activation of BMP-2 transcription (Fig. 7F). In summary, these data indicate that the activation of Shh signaling enhances BMP-2 promoter activity and that this effect is mediated by the transactivation of Gli2, and they suggest that Gli2 function in osteoblasts is regulated by the PKA pathway.

FIG. 7.

Gli2 mediates Shh effects on BMP-2 gene transcription in osteoblasts. (A) Effects of Shh and Gli2 on BMP-2 promoter activity. C2C12 cells carrying a BMP-2 promoter reporter gene, −2712/+165-Luc, were transfected with the Gli2 construct and treated with Shh at 200 ng/ml for 48 h. The relative promoter activity was determined by reading the luciferase-excited fluorescence, with normalization to β-Gal activity. *, P < 0.01 between Shh and control. (B) Effect of cyclopamine on Shh-mediated BMP-2 promoter activity. C2C12 cells with the BMP-2 promoter reporter were treated with cyclopamine (2 μM) or dimethyl sulfoxide vehicle control in the presence of Shh (200 ng/ml) for 48 h. The luciferase activity was measured. #, P < 0.05; *, P < 0.01 (between cyclopamine and control). (C) Effect of Gli2 RNAi on Shh-mediated BMP-2 promoter activity. C2C12 cells carrying Gli2 siRNA were transfected with the BMP-2 promoter reporter gene and treated with Shh at 200 ng/ml for 48 h, and the BMP-2 promoter activity was analyzed. *, P < 0.01 between Shh and control. (D) Effect of trGli3 on Gli2. C2C12 cells were cotransfected with Gli2 and/or the repressor form of Gli3 (trGli3). Thirty-six hours later, the promoter activity was determined as described above. *, P < 0.01 between trGli3 and control. (E) Effect of PKA inhibitor KT5720 on processing of Gli2 protein. C3H10T1/2 cells were transfected with His-Gli2 and treated with KT5720 at 2 μM for 30 h. The cell lysate were analyzed by Western blotting with anti-His antibody (Omni Probe; Santa Cruz). (F) Effect of PKA on BMP-2 promoter activity. C2C12 cells carrying the BMP-2 promoter reporter gene were cotransfected with Gli2 and PKA expression plasmids for 30 h, and the promoter activity was determined. *, P < 0.01 between the two gray bars.

DISCUSSION

Both Shh signaling and BMP-2 expression are required for osteoblast differentiation. In this study, we determined the effect of Gli2, a transcriptional mediator of Shh signaling, on BMP-2 gene expression in osteoblasts. We found that overexpression of Gli2 enhanced BMP-2 gene transcription in osteoblasts and that a loss of function of Gli2 reduced BMP-2 gene expression in osteoblasts and bone tissues. We further identified the Gli2 interaction elements in the BMP-2 promoter and demonstrated that Shh-induced BMP-2 gene transcription in osteoblasts is mediated by Gli2.

Shh-Gli2-BMP-2 pathway in osteoblast differentiation.

BMP signaling induces mesenchymal cells to differentiate into mature osteoblasts (42, 43, 56). Like the BMP signaling pathway, the hedgehog signaling pathway has been implicated in osteoblast differentiation. In the present studies, we found that the hedgehog receptor signaling pathway is intact in osteoblast precursor C3H10T1/2, C2C12, and 2T3 cells, as treatment with Shh triggers osteoblast differentiation in these cells. Using Gli2 gain- and loss-of-function approaches, we also demonstrated that the transcription factor Gli2 plays a critical role in mediating Shh stimulation of osteoblast differentiation (Fig. 1, 4, and 5). Importantly, this Gli2-induced osteoblast differentiation in osteoblast precursor cells is likely due to up-regulation of BMP-2 expression, since the addition of Noggin, a specific natural BMP-2 antagonist, blocked this effect of Gli2 (Fig. 1D). The data in Fig. 1 and 7 show that treatment with Shh has less potency than transfection of Gli2 on either ALP activity or BMP-2 promoter activity in C2C12 cells. This is a puzzle. It is known that Shh acts on cells through its receptors (Ptc and Smo) and that its signal is transduced to Gli proteins by a protein complex composed of Fu, SuFu, Cos-2, and cytoskeletons which regulates the proteolytic processing of Gli proteins. A possible explanation for the phenomenon we observed is that the receptor pathway for Shh in C2C12 cells may not function well, although we could detect the expression of these receptors in these cells. However, overexpression of Gli2 by transfection bypasses such a pathway and directly activates BMP-2 transcription.

In addition, our data also suggest that Gli2 has a synergistic effect with BMP-2 signaling in osteoblasts, since overexpression of Gli2 greatly enhanced BMP-2 stimulation of mineralized matrix formation in cultures of 2T3 cells (Fig. 1C) and since, in contrast, Gli2 deficiency reduced BMP-2 activity (Fig. 5G). We also found that Gli2 synergizes with Smad1 to stimulate a luciferase reporter gene that contains both Gli binding sites and Smad binding sites (data not shown). Thus, we think that the Shh-Gli2 pathway not only regulates BMP-2 gene expression, which we have demonstrated in the present studies, but also may regulate BMP-2 signaling by cross talk between Gli2 and Smad1 in osteoblasts. These results raise strong support for our notion that Shh/Gli2 signaling promotes the osteoblastic phenotype, at least in part, through the BMP-2 pathway.

Gli family regulates BMP-2 gene expression in osteoblasts.

Genetic and embryologic studies suggest that Gli proteins have distinct activities and are not functionally equivalent. Nevertheless, their partial redundancy and often overlapping domains of expression have made it difficult to define precisely their individual features and functions (27, 28, 31, 32, 44, 46). Gli family members have been characterized by their sequence homology, including a central DNA binding domain with five C-2-H-2 zinc fingers, a C-terminal transcription activation domain, and an N-terminal repression domain. In the promoter study (Fig. 3), we found that only Gli2 markedly enhanced BMP-2 promoter activity, while Gli1 and Gli3 had very weak effects. Deletion of the N-terminal domain (Gli2ΔN2) caused a significant increase in Gli2 transactivation of the BMP-2 promoter, and deletion of the C-terminal region turned Gli2 into a repressor of the BMP-2 gene. Consistent with our previous findings (21), C-terminally truncated Gli3 possessed a potent inhibitory effect on BMP-2 gene transcription. These results, along with those for Gli2 RNAi and Gli2 null mutation, demonstrate that Gli2 is a powerful activator of BMP-2 gene expression and indicate that Gli2 regulates target genes in a similar manner to that of Gli3.

Gli2 and Gli3 have important but distinct functions in bone, especially in skeletal development. Both Gli2 and Gli3 mutant mice have severe skeletal abnormalities. In Gli2 null mutant mice, the skeletal defects include cleft palate, the absence of vertebral bodies and intervertebral discs, and shortened tibias and sternums as well as cartilage defects (22, 35, 36). These Gli2 phenotypes are distinct from those of Gli3 mutants. Gli3 mutant mice exhibit polysyndactyly that is similar to several human skeletal syndromes, namely the Greig cephalopolysyndactyly syndrome and the Pallister-Hall syndrome (6). The exaggeration of skeletal defects in Gli2/Gli3 double mutant mice suggests some unique and redundant functions for these two genes in skeletal development. In characterizing Gli2 null mice, we found that in addition to the skeletal defects in embryos, one-third of heterozygous Gli2 null mice developed an osteopenic phenotype postnatally, as represented by a lower bone mineral density and markedly reduced trabecular bone volume (data not shown). These mice also had severe defects in cartilage development, including smaller stature, extremely short forelimbs, disorganized elbow joints, and enlarged growth plates in the long bones of the hindlimbs (data not shown). In this study, we also examined BMP-2 gene expression in bones of Gli2^zfd/zfd mice and found that BMP-2 mRNA and protein expression was decreased in osteoblasts on the trabecular bone surface as well in hypertrophic chondrocytes in the growth plates of long bones and developing vertebrae (Fig. 5). This suggests that the regulation of BMP-2 gene expression by Gli2 in these areas is important for normal endochondral bone formation.

Gli transcription factors act on target genes by binding to the specific Gli-responsive element(s) in promoters (38, 45). Through promoter deletion analysis, we mapped out two Gli2-responsive regions in the BMP-2 promoter. Using ChIP and EMSA, we identified three putative Gli binding sites in these areas (Fig. 6). Our data suggest that the Gli binding sites are responsible for direct transactivation of Gli2. However, the individual roles of each of the Gli binding sites in this promoter have yet to be defined by single and multiple mutations of these binding sites. Moreover, the possible implication of competition between Gli2 and trGli3 for binding to the BMP-2 promoter remains to be established, since Gli2 and Gli3 are known to occupy the same binding sites on other promoters. We have found that trGli3 dramatically inhibits Gli2 enhancement of BMP-2 promoter activity (Fig. 7D). The levels and activities of endogenous Gli2 and trGli3 during osteoblast differentiation need to be further investigated.

Processing of Gli proteins and BMP-2 gene regulation.

In cells, Gli2 and Gli3 share similar regulatory mechanisms in responding to Shh signaling. In the absence of Shh, both proteins are phosphorylated by PKA and undergo proteolytic processing (54). However, unlike that of Gli3, the processing of Gli2 was not well identified until a recent study conducted by Wang et al. (41). In this study, Wang et al. demonstrated that the phosphorylation of Gli2 is done sequentially by PAK, CK1, and GSK3 and that the E3 ubiquitin ligase β-TrCP is required for leading Gli2 to proteasomal degradation. Unlike Gli3 and Ci, only a minor fraction of Gli2 is proteolytically processed to form a transcriptional repressor, and in addition to being processed, the full-length Gli2 protein is readily degraded. Coincidentally, in our studies, we found that blocking PKA activity elevated the Gli2 protein levels in osteoblasts, suggesting that the inhibition of PKA prevents Gli2 from degradation (Fig. 7E). Furthermore, overexpression of PKA significantly inhibited Gli2-enhanced BMP-2 promoter activity (Fig. 7F). The data on PKA-mediated Gli2 processing provide a mechanism for Shh effects on BMP-2 gene expression in osteoblasts (Fig. 7A to C). These results are consistent with our previous findings with Gli3 in which we demonstrated that the effect of Gli3 on BMP-2 gene transcription is PKA dependent and that proteasome inhibitors prevent Gli3 proteolytic processing and consequently enhance BMP-2 gene expression as well as bone formation (16). Therefore, based on these observations, we suspect that proteasome inhibitors enhance BMP-2 gene expression, not only by reducing trGli3 production but also by preventing Gli2 degradation.

Proteasomal processing is mediated by members of the large family of PKA-dependent E3 ubiquitin ligases (48). Consistent with the finding of Wang et al., we recently found that Slimb, a Drosophila homologue of β-TrCP, induces proteolytic processing of Gli2 protein and inhibits BMP-2 promoter activity in osteoblasts (data not shown). However, it is very interesting that Smurf1, a Hect domain E3 ligase, also plays a role in Gli2 processing. We identified a “PY” motif at the N-terminal region that is known to be recognized by Smurf1. Previously, we demonstrated that Smurf1 has an important role in BMP-2 gene regulation, osteoblast differentiation, and bone formation by regulating proteolytic processing of Smad1 and Runx2 (63, 64). The additional requirement for Smurf1 in Gli2 processing may raise the possibility that Gli2 undergoes sequential proteolytic processing mediated by β-TrCP and then Smurf1, resulting in degradation of full-length Gli2. Taken together, these findings suggest that the PKA- and E3 ligase-dependent proteasomal pathway is involved in the regulation of Gli proteins and BMP-2 gene expression in osteoblasts.

In conclusion, we provide evidence that Gli2 is a transcriptional activator of BMP-2 gene transcription and plays a critical role in the Shh-Gli2-BMP-2 signaling pathway in osteoblast differentiation.