CCAAT/Enhancer-binding Proteins (C/EBP) β and δ Activate Osteocalcin Gene Transcription and Synergize with Runx2 at the C/EBP Element to Regulate Bone-specific Expression

CCAAT/enhancer-binding proteins (C/EBP) are critical determinants for cellular differentiation and cell type-specific gene expression. Their functional roles in osteoblast development have not been determined. We addressed a key component of the mechanisms by which C/EBP factors regulate transcription of a tissue-specific gene during osteoblast differentiation. Expression of both C/EBPβ and C/EBPδ increases from the growth to maturation developmental stages and, like the bone-specific osteocalcin (OC) gene, is also stimulated 3–6-fold by vitamin D3, a regulator of osteoblast differentiation. We characterized a C/EBP enhancer element in the proximal promoter of the rat osteocalcin gene, which resides in close proximity to a Runx2 (Cbfa1) element, essential for tissue-specific activation. We find that C/EBP and Runx2 factors interact together in a synergistic manner to enhance OC transcription (35–40-fold) in cell culture systems. We show by mutational analysis that this synergism is mediated through the C/EBP-responsive element in the OC promoter and by a direct interaction between Runx2 and C/EBPβ. Furthermore, we have mapped a domain in Runx2 necessary for this interaction by immunoprecipitation. A Runx2 mutant lacking this interaction domain does not exhibit functional synergism. We conclude that, in addition to Runx2 DNA binding functions, Runx2 can also form a protein complex at C/EBP sites to regulate transcription. Taken together, our findings indicate that C/EBP is a principal transactivator of the OC gene and the synergism with Runx2 suggests that a combinatorial interaction of these factors is a principal mechanism for regulating tissue-specific expression during osteoblast differentiation.

The CCAAT/enhancer-binding proteins (C/EBPs) 1 comprise a family of transcription factors that are critical for normal cellular differentiation and metabolic functions in a variety of tissues. There are currently six members of the C/EBP family designated as C/EBP␣, -␤, -␦, -␥, -⑀, and -(1), most of which are expressed in liver, spleen, and adipocytic tissues. However, more selective expression in other tissues has been observed among the family members (2)(3)(4)(5)(6)(7). Isoforms of the C/EBP proteins are known, and all function by homo-or heterodimerization with one another and interaction with other transcriptional activators or co-activators such as NF-B, Stat3, c-Myb, PU.1, SP-1, ATF-2, PPAR␥, and Runx-1 (8 -14).
Very little is known about the role of C/EBP factors in osteogenesis. Targeted disruptions of C/EBP genes have been performed, but in none of the studies were gross abnormalities of the skeleton observed (6,(15)(16)(17). However, recent studies have identified C/EBP regulation of genes expressed in osteoblasts. The insulin-like growth factor 1 is a key regulator of osteoblast growth and differentiation (18). C/EBP␦ enhances either basal or prostaglandin E 2 -activated transcription of the insulin-like growth factor 1 promoter in osteoblasts (19,20). Expression of COX-2 and the ␣1 subunit of type I collagen is also regulated in osteoblasts by C/EBP factors (21,22). The interaction of C/EBP␣ with a Runx1 factor (23) is also particularly relevant for postulating a role for C/EBP factors in osteoblast differentiation.
The Runt-related transcription factors (Runx/AML/CBF␣/ PEBP2␣) represent essential gene regulatory proteins that control lineage commitment for hematopoiesis (24 -26) and osteogenesis (27,28). Runx2 (AML3/Cbfa1/PEBP2␣A) is the most abundant Runt-related protein in osteogenic and chondrogenic cell lineages (29 -31). Genetic ablation of the Runx2 gene causes developmental defects in osteogenesis (27), and hereditary mutations in the Runx2 gene are linked to specific ossification defects as observed in cleidocranial dysplasia (32). Runx2 is essential for osteoblast differentiation (27,29,30) and regulates expression of numerous bone-related genes (29,31,(33)(34)(35). The importance of Runx2 in expression of the bonespecific osteocalcin (OC) gene is well documented (36,37). Thus, Runx2 performs specialized functions during bone-tissue development and differentiation in vivo. However, it is noteworthy that osteoblast-specific transcription of osteocalcin occurs even in the absence of Runx sites in the rat OC promoter (38), suggesting a tissue-specific role for other regulatory factors in osteoblasts.
Activation of tissue-specific genes is controlled by combinatorial mechanisms that rely on local features of the promoters, including organization of control elements in the target genes and/or the interplay between DNA-binding proteins and various transcriptional co-regulators (39). Both C/EBP␤ and Runx factors have been shown to cooperate with chromatin remodeling factors (p300, SWI/SNF) and other enhancer-binding proteins (40,41). For example, Ets-1, c-Myb, Sp1, and C/EBP, together with Runx factors, stimulate the transcription of hematopoietic and osteogenic genes (13,14,23,(42)(43)(44), whereas PPAR␥ and Stat3 interactions with C/EBP␣ are driving forces for adipocyte differentiation (45). Given these observations, i.e. the presence of both C/EBP␤ and -␦ and Runx2 in osteoblasts (29,30,46) and C/EBP␣-Runx1 protein-protein interactions in regulation of a hematopoietic specific gene (13), we addressed the possible role of C/EBP factors in osteoblasts and in the regulation of a bone-specific gene, osteocalcin.
Here we report that C/EBP␤ and -␦, but not -␣, are developmentally expressed during osteoblast differentiation and are up-regulated in response to 1,25(OH) 2 D 3 , a hormone that promotes osteoblast differentiation. We have identified a C/EBP-responsive regulatory element in the proximal promoter of the bone-specific osteocalcin gene. Deletion or mutation of this motif abrogates transcriptional enhancement by C/EBP factors. Furthermore, we provide the first demonstration that Runx2 and C/EBP␤ physically interact and that C/EBP and Runx proteins act synergistically to activate the OC promoter. Importantly, this functional synergism is mediated through the C/EBP element. These findings establish for the first time a role of C/EBP in the regulation of an osteoblastspecific gene and define a novel mechanism for C/EBP in the regulation of cell type-specific gene transcription.
The Ϫ208 OC-CAT Runx mt plasmid was generated by digestion of the mC CAT plasmid (36) with SphI-PpMU, followed by blunt ending and self-ligation. To generate the plasmid with mutations in both binding sites, two PCR reactions were set as above, but in this case Ϫ208 OC-CAT Runx mt was used as template. All plasmids were sequenced using the pUC/M13 forward primer. The expression constructs encoding the wild type Runx2 and Runx2 ⌬361 are as reported previously (29,31). Runx2 ⌬230 was prepared by PCR amplification of the coding sequences using the forward primer 5Ј-CGGGATCCATGCGTATTCC-3Ј and a reverse primer with an engineered stop codon 5Ј-GGCTCGAGT-CATTTAGAGTCATCAGGC-3Ј. Nonencoded nucleotides are underlined. The PCR fragment was digested with BamHI-XhoI and ligated to similarly digested pcDNA 3.1 His C vector (Invitrogen Inc., San Diego, CA). In-frame ligation of these constructs was confirmed by DNA sequencing. Expression constructs of C/EBP␣, -␤, and -␦ were obtained from Dr. Alan Friedman (Johns Hopkins Hospital, Baltimore, MD).
Cell Cultures, Transient Transfection, and Reporter Assays-Normal rat diploid osteoblasts obtained from 21-day fetal rat calvariae were isolated and maintained as described (49). HeLa cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Rat osteosarcoma (ROS 17/2.8) cells were grown in F-12 supplemented with 5% fetal calf serum. HeLa and ROS 17/2.8 cells plated in six-well plates were transiently transfected with 1-2.5 g of OC promoter-CAT and 0.1-0.75 g of cytomegalovirus empty vector, C/EBP␤, C/EBP␦, or Runx expression constructs. Cells were transfected using SuperFect transfection reagent as described previously (50). RSVluciferase plasmid (100 ng) was included as an internal control for transfection efficiency. Cells were harvested 24 -36 h after transfection and assayed for CAT activity. The data were normalized to luciferase values obtained from the same samples.
Isolation of Total Cellular RNA and Northern Blot Analysis-Total cellular RNA was isolated from normal rat fetal calvarial osteoblasts or adult rat tissues as described previously (51). Cells at different stages of differentiation were washed and scraped in phosphate-buffered saline. Frozen cell pellets from each time point were thawed and processed together. RNA pellets were briefly air-dried and dissolved in 400 l of diethyl pyrocarbonate-treated water and stored at Ϫ70°C until further usage. RNA samples from different stages of differentiation were electrophoresed in 1% formaldehyde-agarose gels and transferred to Hybond-N ϩ membrane (Amersham Biosciences, Inc.) in 20ϫ SSC. Blots were hybridized with random primed (Prime-It kit; Stratagene, La Jolla, CA), 32 P-labeled cDNA probes for C/EBP␣ (NcoI fragment), C/EBP␤ (EcoRI-XhoI fragment), C/EBP␦ (BamHI-EcoRI fragment), human glyceraldehyde-3-phosphate dehydrogenase (EcoRI-HindIII fragment), human histone H4 (HindIII-XbaI fragment of FO108), alkaline phosphatase (EcoRI fragment), Runx2 (BamHI-XbaI fragment), and rat OC (EcoRI-BamHI fragment) at 42°C overnight. The blots were washed and subjected to autoradiography.
Oligonucleotides and Electrophoretic Mobility Shift Assay-Oligonucleotides spanning the C/EBP motif from the rat osteocalcin promoter (Ϫ117/-88) were synthesized: wild type, 5Ј-GGTTTGACCTATTGCG-CACATGACCCCCAA-3Ј; and mutant, 5Ј-GGTTTGACCTAgactagt-CATGACCCCCAA-3Ј. The C/EBP motif is shown in bold, and the mutated sequences in lowercase. Nuclear extracts from ROS 17/2.8, HeLa, and primary rat osteoblast cells at different stages of culture (days 7, 14, and 20) were prepared by a modified Dignam method (52) with 0.42 M KCl for extraction. ROS 17/2.8 and HeLa cells were plated in 100-mm plates at a density of 0.5 ϫ 10 6 . Cells were collected at 95% confluence by washing twice with ice-cold phosphate-buffered saline. The whole isolation procedure was carried out on ice. Cells from five plates were pooled into a 50-ml polypropylene tube and pelleted by centrifugation at 165 ϫ g for 5 min at 4°C. Cells were gently resuspended in 5-10 volumes of Nonidet P-40 lysis buffer (10 mM Tris, pH 7.4, 3 mM MgCl 2 , 10 mM NaCl, 0.5% Nonidet P-40) supplemented with 1ϫ Complete protease inhibitor mixture (Roche Molecular Biochemicals) and incubated on ice for 20 min. Aliquots (25 l) of the nuclear extracts were snap-frozen in liquid nitrogen and stored at Ϫ80°C until further use. Protein concentration of nuclear extracts was determined by Bradford assay. A polyclonal Runx2-specific antibody (Oncogene Research Products, Boston, MA) was used in supershift assays.
Immunoprecipitation-HeLa cells were transfected with Runx2 and C/EBP␤; ϳ10 7 cells/immunoprecipitation were lysed in 800 l of Nonidet P-40 buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 1% Nonidet P-40, 1ϫ Complete (Roche Molecular Biochemicals), 25 M MG132 (Sigma-Aldrich)) and extracted at 4°C for 15 min, followed by centrifugation at 16,000 ϫ g for 15 min. The supernatant was transferred to a clean microcentrifuge tube and precleared with 20 l of protein A/G Plusagarose beads (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), at 4°C for 30 min. The beads were collected by centrifugation at 1000 ϫ g for 5 min at 4°C. Xpress antibody (3 g, Invitrogen Corp., Carlsbad, CA) was added to the precleared cell lysate followed by incubation at 4°C for 1 h. To precipitate the immunocomplexes, 50 l of protein A/G Plusagarose beads were added and further incubated at 4°C with agitation for 1 h. The beads were washed twice with 1 ml of washing buffer (20 mM Tris, pH 8.3, 0.5% sodium deoxycholate, 0.5% Nonidet P-40, 50 mM NaCl, 2 mM EDTA, 1ϫ Complete , 25 M MG 132), suspended in 1ϫ SDS sample buffer, and analyzed by Western blotting.

C/EBP Family Members Are Expressed during Osteoblast
Differentiation and Are Regulated by Vitamin D 3 -We initially assessed the expression of C/EBP factors in various bone tissues and during development of the osteoblast phenotype. Fig.  1A shows that C/EBP␤ and -␦ mRNAs are present in calvarial tissue (lane 2) and in cortical and trabecular bone (data not shown) at levels similar to those in a representative soft tissue, muscle (lane 1). C/EBP␣ is not detected in bone tissue, consist-ent with its pivotal role for adipogenesis. Osteoblast markers (OC, Runx), which are not present in soft tissues, are shown for comparison. The abundance of C/EBP␤ and -␦ in bone prompted examination of expression of the C/EBP family members from growth to differentiation stages of osteoblasts in vitro, representing proliferation (days 3-5), matrix maturation (days 7-12), and the mineralization stage (days 19 -22), reflected by peak levels of histone H4, alkaline phosphatase (ALP), and OC, respectively (Fig. 1B).
Both C/EBP␤ and -␦ mRNAs are detected from the growth to maturation stages and are expressed in a biphasic pattern. In contrast, C/EBP␣ expression is not detected at any stage of osteoblast differentiation, consistent with its absence in bone tissue. Two sizes of C/EBP␤ mRNA are present with the larger species appearing more constitutive, whereas the smaller transcript is expressed during the growth period (days 3-5). The latter transcript is decreased markedly during the matrix maturation stage (days 7-12, when alkaline phosphatase-positive cells are forming nodules), followed by a 4 -5-fold increase in expression concomitant with mineral deposition and peak levels of osteocalcin and Runx2 expression (Fig. 1B). C/EBP␦ mRNA is expressed in a manner similar to that for C/EBP␤, but the larger transcript is detected at very low levels. A significant (5-fold) temporal increase in C/EBP␦ mRNA expression is observed during osteoblast differentiation from confluence (day 7) to the mature osteoblasts (day 22). Several Runx2 isoforms that result from utilization of alternative promoters and differential splicing are expressed (Fig. 1B). The increase in expression of the major Runx2 transcript during later stages of differentiation is consistent with increased Runx2 DNA binding activity in mature osteoblasts (29). Thus, the increases in C/EBP␤ and C/EBP␦ in the late stages of osteoblast differentiation appear to parallel peak expression levels for the osteoblast-related osteocalcin gene and the Runx2 transcription factor.
We further assessed the relationship of C/EBP factors to osteoblast differentiation. Vitamin D 3 , a hormone that promotes osteoblast differentiation, is a known enhancer of many osteoblast-related genes (53). Expression of C/EBP␤ is stimulated by vitamin D 3 at each stage of osteoblast maturation, 6-fold during growth (day 7) and nodule development (day 12) and 3-fold during mineralization (day 19). Vitamin D 3 -dependent enhancement of C/EBP␦ is similar to that of C/EBP␤ except the -fold stimulation is lower (3-fold at all stages). Treatment of cells with vitamin D 3 has no effect on the expression of C/EBP␣ at any stage of differentiation (data not shown). The differentiation-promoting properties of vitamin D 3 are reflected by the decrease in histone H4 and the increase in OC expression (Fig.  1C). Thus, expression of both C/EBP␤ and C/EBP␦ is strongly enhanced upon treatment with vitamin D 3 in relation to osteoblast differentiation. Taken together, these data demonstrate that C/EBP transcription factors are expressed at significant levels in bone tissue, they increase during osteoblast differentiation in vitro, and their expression is up-regulated by 1,25(OH) 2 D 3 .
C/EBP Proteins Activate the Osteocalcin Gene through a C/EBP-responsive Element in the Proximal Promoter-The enhanced expression of C/EBP␤ and -␦ during mineralization relative to the onset of OC transcription and in response to vitamin D 3 suggests that these C/EBP transcription factors may contribute to osteoblast-specific expression of the OC gene. Previous studies using promoter deletion constructs of the rat OC gene have shown that the initial 200 bp of the promoter can confer tissue-specific expression (38). This region contains a Runx-responsive motif and a homeodomain box that also binds an osteoblast-specific complex (54). Sequence analysis of this FIG. 1. Biphasic expression and vitamin D regulation of C/EBP transcription factors during rat osteoblast differentiation and presence in bone tissues. A, total RNA was isolated from various skeletal and soft rat tissues as described under "Materials and Methods." RNA (10 g) was resolved in a 1% formaldehyde gel and transferred to nylon membrane. The blots were hybridized with NcoI fragment for C/EBP␣, EcoRI-XhoI fragment for C/EBP␤, BamHI-EcoRI fragment for C/EBP␦, EcoRI fragment for alkaline phosphatase, Hin-dIII-XbaI fragment for histone H4, BamHI-XhoI fragment for Runx2, EcoRI-HindIII fragment for glyceraldehyde-3-phosphate dehydrogenase, and EcoRI-BamHI fragment for the detection of rat OC. C/EBP␤ and -␦ mRNAs are expressed in both soft and skeletal tissues. Lane 1, muscle; lane 2, day 21 fetal calvaria. B, primary rat osteoblasts were cultured for the indicated time (days) and total RNA (10 g) was resolved in a 1% agarose gel and probed with the above mentioned cDNA fragments. The relative positions of 28, 18, and 5 S ribosomal RNA are indicated for reference. C, enhancement of C/EBP factors by vitamin D during osteoblast differentiation. Primary rat calvarial osteoblasts were cultured in vitro and collected at three stages of development: proliferation (day 7), matrix maturation (day 12), and mineralization (day 19). Cells were treated with 10 Ϫ8 M vitamin D 3 for 24 h prior to harvesting. Total RNA was isolated as described under "Materials and Methods." The Northern blotting and probing were performed essentially as described above. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is shown as control for loading of RNA. region reveals the presence of a C/EBP motif ( Fig. 2A). To determine whether the C/EBP protein(s) can interact with this element, gel mobility shift analyses were performed using oligonucleotides (see "Materials and Methods") containing either wild type or mutated C/EBP binding sequences ( Fig. 2A). Using nuclear extracts from day 20 primary rat osteoblasts in which OC is actively expressed, we observed two major protein-DNA complexes (Fig. 2B). Specificity of these protein-DNA interactions was confirmed by oligonucleotide competition assays. Addition of unlabeled wild type oligonucleotide, but not mutant oligonucleotide, inhibited the binding of the protein complexes to the labeled probe (Fig. 2C). Taken together these results indicate that C/EBP proteins interact in a sequence-specific manner with their cognate element in the proximal promoter of the bone-specific rat OC gene.
The presence of C/EBP in the protein complex binding to the OC promoter element suggests transcriptional regulation of the OC gene by C/EBP factors. This possibility was experimentally addressed by assessing the effect of forced expression of C/EBP␤ and -␦ on activity of full-length (Ϫ1.1 kb) and proximal (Ϫ208 bp) OC promoter-CAT reporter gene constructs in the osteoblastic ROS 17/2.8 cell line (Fig. 3). The results indicate that C/EBP␤ and -␦ significantly enhance OC promoter activity, 4 -5-fold on the full-length promoter (Fig. 3A, left panel) and 8-fold on the proximal promoter segments (Fig. 3A, right  panel). Western blot analysis shows that C/EBP␤ and C/EBP␦ proteins of the expected sizes are expressed (Fig. 3B). A representative CAT autoradiogram showing basal activity of the Ϫ1.1 kb and Ϫ208 OC-CAT and the extent to which these promoters are activated by C/EBP␤ and -␦ is presented in Fig.   3C. Thus, both C/EBP␤ and -␦ are potent activators of osteocalcin gene transcription.
To establish that the C/EBP element in the proximal region directly mediates the enhancement of promoter activity by C/EBPs, site-directed mutagenesis was performed. The C/EBP element in Ϫ208 OC-CAT was mutated (see Fig. 2A) using the same oligonucleotides used in the gel shift assays. The results show that mutation of the C/EBP element blocks the C/EBPmediated stimulation (6 -8-fold) observed with the wild type promoter (Fig. 4). Therefore, responsiveness of the OC promoter to C/EBP transcription factors is dependent on the integrity of the C/EBP element in the basal promoter region.
Functional Synergism between Runx and C/EBP Proteins Is Mediated through the C/EBP Element-Runx2 is a well characterized regulator of OC gene transcription and a Runx-responsive motif is located in close proximity to the C/EBP element in the proximal OC promoter. Therefore, functional interaction between these proteins in transcriptional regulation of the OC gene was tested. For these studies we selected HeLa cells, which lack Runx2 (Fig. 5A) and have no Runx DNA binding activity (Ref. 55 and data not shown). Consistent with these observations, we find low basal activity of the Ϫ208 OC promoter in HeLa cells compared with osteoblastic ROS 17/2.8 cells, which express both C/EBP␤ and Runx2 (Fig. 5). Interestingly, HeLa cells have significant levels of C/EBP␤ (Fig. 5A). Based upon preliminary examination of dose-dependent effects of Runx2 and C/EBP␤ individually on activity of the Ϫ208 OC promoter (data not shown), we used a suboptimal concentration of each expression plasmid (0.4 g) for these studies. Expression of either C/EBP␤ or Runx2 in HeLa cells stimulates OC promoter activity 2-4-fold. However, co-expression of Runx2 and C/EBP␤ results in a massive activation (30 -40-fold) of the OC promoter (Fig. 6), demonstrating a functional synergism between these two proteins. We also observed a synergistic interaction of Runx2 and C/EBP␤ on the OC promoter in osteoblastic ROS 17/2.8 cells (20 -25-fold stimulation compared with 3-5-fold for each protein alone; data not shown). This functional synergy between two positive regulators of OC transcription is consistent with increased expression of C/EBP␤ and -␦, Runx2, and OC during late stages of osteoblast differentiation.
To investigate the specific contribution of the Runx and C/EBP elements and their cognate factors to synergistic activation of the OC promoter, a series of promoter constructs bearing mutations in either Runx or C/EBP binding sites were generated (Fig. 7A). Each of these constructs was tested for responsiveness to C/EBP and/or Runx2. Data pooled from four independent experiments show a consistent 30 -40-fold synergistic enhancement of the wild type OC promoter by C/EBP␤ and Runx2 (Fig. 7B). Mutation of the Runx site did not affect the synergistic response (23-26-fold enhancement). In contrast, we observed a loss of this functional synergism upon mutation of the C/EBP motif. A similar loss of synergistic activity was observed when both C/EBP and Runx sites were mutated (Fig. 7B). Co-expression of C/EBP␦ and Runx2 resulted in a similar pattern of activation of these OC promoter constructs (data not shown). These findings suggest that the C/EBP regulatory element is required for the synergistic enhancer activity involving Runx2.
For further insight into the mechanisms involved in Runx2-C/EBP functional synergism, we tested a series of carboxylterminal deletion mutants of Runx2 (Fig. 8A). Both mutant Runx2 proteins (1-361 and 1-230) are expressed (Fig. 8B), enter the nucleus and retain DNA binding activity (data not shown). Fig. 8C shows that both mutants also retain functional activity on the OC promoter but at a lower level than wild type Runx2. When the mutant Runx2-(1-361) was co-expressed with either C/EBP␤ or -␦, synergistic activation (40 -80-fold) of the OC promoter was observed. However, this synergism did not occur with the mutant Runx2-(1-230). These Runx mutational studies clearly demonstrate that a Runx2-C/EBP interaction is required to support synergistic activation of the OC promoter.
To determine whether functional synergism between Runx2 and C/EBP requires a physical interaction, we performed coimmunoprecipitation studies. Wild type Runx2 and Runx2-(1-361), which lacks the carboxyl terminus, each form a complex with C/EBP␤, whereas Runx2-(1-230) fails to interact (Fig.  8D). The absence of C/EBP␤ in the immunoprecipitated Runx2-(1-230) complex indicates that amino acids 230 -361 of Runx2 are required for this interaction, consistent with the functional activity data (Fig. 8C). It was shown previously that the runt homology domain (RHD) of Runx1 supports interaction with C/EBP␣ (13,23). However, our Runx2 deletion analysis reveals that the RHD of Runx2 is not sufficient for C/EBP␤ interaction (Fig. 8D). Our findings are consistent with the recent co-crystal structure of the RHD with C/EBP␤ (bZIP), which demonstrates a lack of interaction of these two domains (56). Taken together our results demonstrate that the synergism observed on the OC promoter requires an interaction between Runx2 and C/EBP that involves a region of Runx2 outside the DNA binding domain.
To establish that the Runx2-C/EBP protein-protein interaction can occur when C/EBP proteins are bound to its regulatory element, we performed gel mobility shift assays with the C/EBP motif of the OC promoter. We compared nuclear extracts from HeLa cells lacking Runx2 and mature osteoblasts, which contain both Runx and C/EBP factors (see Fig. 5A). We find similar C/EBP complexes formed with both nuclear extracts; however, addition of antibody against Runx2 resulted in a supershift only with the bone cell extracts. These results provide evidence for a Runx2-C/EBP interaction that is independent of a Runx DNA binding site. DISCUSSION Our studies demonstrate that the C/EBP transcription factors support osteoblast-specific gene expression and may play an important regulatory role during osteoblast differentiation. C/EBP␤ and -␦, but not C/EBP␣, are expressed in skeletal tissues and are developmentally regulated during osteoblast maturation. Vitamin D 3 , a positive regulator of osteoblast differentiation and of the bone-specific osteocalcin gene, also increases expression of C/EBP factors. We find that osteocalcin is a downstream C/EBP target gene, strongly up-regulated in response to forced expression of C/EBP family members. The level of enhancement is equivalent to that observed for the bone-related Runx2 transcription factor. More importantly, we also demonstrate that Runx2 and C/EBP␤ and -␦ functionally cooperate for positive regulation of the OC gene and that the synergism is mediated through a physical interaction between Runx and C/EBP at the C/EBP element. We propose that C/EBP activity may be physiologically relevant to the spatiotemporal regulation of Runx2-dependent genes in mature osteoblasts.
Transcription of the OC gene is stringently regulated during osteoblast differentiation (12), but the mechanisms involved have not been completely elucidated. OC gene induction is FIG. 7. Functional synergism between Runx2 and C/EBP factors is mediated by a C/EBP-responsive motif in the rat OC promoter. A, illustration of mutations in the proximal OC promoter. Top line diagram shows the relative positions and sequences of Runxand C/EBP-responsive elements with mutated sequences shown in lowercase. Constructs carrying mutation in either Runx or C/EBP elements alone or combined are indicated. B, HeLa cells were transiently cotransfected with 1 g of either wild type or mutated OC promoter constructs (carrying mutation in either C/EBP or Runx sites individually or together) and 0.4 g of C/EBP␤ or Runx2 expression constructs as indicated. Reporter activities were determined 24 h after transfection and normalized to luciferase values used for transfection control. WT, wild type; mt, mutant.
FIG. 8. Functional synergy between C/EBP and Runx2 requires C/EBP interacting motif. A, deletion mutants of Runx2 are shown diagrammatically. The RHD, nuclear localization (NLS), and putative C/EBP interacting region are indicated. Panel B shows Western blot analysis of Xpress-tagged Runx2 protein expressed in HeLa cells. Tubulin is shown as a loading control. C, functional synergism is demonstrated in HeLa cells, which were transiently co-transfected with 0.4 g each of cytomegalovirus empty vector, C/EBP␤, C/EBP␦, Runx2 full-length and deletion mutant expression plasmid, and 1 g of the Ϫ208 OC-CAT constructs, as well as 100 ng of a RSV-luciferase plasmid. Cells were harvested 24 h after transfection and assayed as described under "Materials and Methods." Data are presented as -fold induction (expression constructs/empty vector control, n ϭ 12 from four independent experiments). D, direct interaction between Runx2 and C/EBP␤ is shown by co-immunoprecipitation. HeLa cells were co-transfected with C/EBP␤ and the Xpress-tagged deletion constructs of Runx2 (as indicated) lysed 24 h later and immunoprecipitated (Ip) with ␣Xpress antibody or mouse IgG as described under "Materials and Methods." Immunoprecipitated complexes were resolved on 12% SDS-PAGE followed by Western blotting using either ␣XpressHRP or ␣C/ EBP␤. E, electrophoretic mobility shift analysis shows a Runx2-C/EBP interaction at the C/EBP element. A C/EBP oligonucleotide (see Fig. 2 legend) was incubated with 10 g of nuclear extracts from HeLa cells (lanes 1 and 2) or day 20 rat osteoblasts (lanes 3 and 4). Nuclear extracts (N. extract) were preincubated with 1 l of Runx2 antibody (Ab) for 20 min at 37°C (lanes 2, 4, and 5). Bracket shows C/EBP complexes; arrowhead indicates supershifted band. coupled to a post-proliferative increase in Runx2 DNA binding activity (29,30). However, Runx proteins are present in skeletal progenitor cells and immature proliferating osteoblasts, in which OC gene expression is not activated (29). Thus, other factors are contributing to strong suppression of OC transcription in such cells, as well as to maximal levels of OC expression in mature osteoblasts. Our studies indicate that the activities of Runx proteins and transcriptional activation of Runx-responsive genes during skeletal development may be regulated in part by controlling cellular levels of Runx or C/EBP proteins and/or functional cooperation between Runx and C/EBP transcription factors. In support of this mechanism are the relative expression levels of each factor during osteoblast differentiation. Although C/EBP␤ and -␦ are present in proliferating osteoblasts (consistent with their role in cell growth), the low Runx levels may be insufficient to produce a synergistic effect. In contrast, when the cellular levels of both factors are elevated during the mineralization stage, synergy may account for the massive and tissue-specific induction of OC gene expression. Notably, C/EBP␦ expression is positively regulated by Runx2 (46), which may provide the biological assurance of high C/EBP levels during osteoblast differentiation when bone-specific gene expression is required.
For definitive assessment of the contribution of Runx2 and C/EBP proteins and their synergistic effect on the OC promoter, we carried out studies not only in bone cells, but also in HeLa cells, which have a zero background for Runx factors (55). Synergistic activation of the OC promoter occurred only when both factors were expressed in HeLa cells. This situation is analogous to mature osteoblasts, where both proteins are maximally present. Based upon previous observations of the critical role of the Runx site in maintaining active chromatin conformation of the OC promoter (36), we anticipated that synergism would involve the Runx element located 30 bp upstream of the C/EBP site. Our transient transfection studies of the OC promoter with mutated elements indicate a role for the C/EBP element in mediating synergistic activation with Runx2. However, these studies do not exclude the contribution of the Runx2 binding site to regulation of OC transcription in osteoblasts.
The mechanisms by which C/EBP and Runx factors together result in a synergistic activation of tissue-specific genes may involve modifications in chromatin structure. Several studies have shown that the Runx and C/EBP classes of transcription factors can each form regulatory complexes with proteins that influence chromatin remodeling. Runx factors facilitate the formation and maintenance of transcriptionally active chromatin (36,(57)(58)(59)(60) and are known to interact with co-regulatory proteins possessing histone acetyltransferase (40) or histone deacetylase activity (57,61). Recent studies from our laboratories indicate that Runx factors regulate tissue-specific and vitamin D-mediated gene expression of osteocalcin by modifying chromatin organization (36). More importantly, C/EBP␤ has recently been shown to recruit the SWI/SNF complex to modify chromatin and regulate transcription of myeloid genes (41). Numerous studies have established that chromatin remodeling and modifications in nucleosomal organization are necessary for bone-specific activation of the osteocalcin gene (36,(62)(63)(64). Hence, C/EBP-dependent Runx synergism during development of the osteoblast phenotype may result from temporal modifications in the chromatin-related interactions among C/EBP, Runx, SWI/SNF, and/or other components of the nuclear architecture.
In summary, the data presented here address the mechanisms critical for the robust activation of the osteocalcin gene during osteoblast differentiation. Our studies indicate that these mechanisms require a C/EBP response element and syn-ergism of C/EBP and Runx factors that may be facilitated by their regulated levels of expression in mature osteoblasts. The concept that lineage-specific gene expression depends on the combination of factors, rather than being controlled by a single master regulator, is reinforced by these studies.