NF-κB Specifically ActivatesBMP-2Gene Expression in Growth Plate Chondrocytesin Vivoand in a Chondrocyte Cell Linein Vitro

Bone morphogenetic protein-2 (BMP-2) regulates growth plate chondrogenesis during development and postnatal bone growth, but the control mechanisms of BMP-2 expression in growth plate chondrocytes are unknown. Here we have used both in vitro and in vivo approaches to demonstrate that transcription factor, NF-κB, regulates BMP-2 gene expression in chondrocytes. Two putative NF-κB response elements were found in the –2712/+165 region of the BMP-2 gene. Cotransfection of mutant I-κBα expression plasmids with BMP-2 promoter-luciferase reporters into TMC-23 chondrocyte cell line suppressed BMP-2 transcription. Mutations in NF-κB response elements in the BMP-2 gene lead to decreases in BMP-2 promoter activity. Electrophoretic mobility shift assay using nuclear extracts from TMC-23 chondrocytic cells revealed that the NF-κB subunits p50 and p65 bound to the NF-κB response elements of the BMP-2 gene. Thus, NF-κB may positively regulate BMP-2 gene transcription. Consistent with these findings, expression of BMP-2 mRNA was significantly reduced in growth plate chondrocytes in NF-κB p50/p52 dKO mice, which associated with decreased numbers of 5-bromo-2′-deoxyuridine (BrdUrd)-positive cells in the proliferating zone of growth plate in these mice. Therefore, in postnatal growth plate chondrocytes, expression of BMP-2 is regulated by NF-κB, which may play an important role in chondrogenesis.

In long bones, morphogenesis is followed by an extended period of growth. Longitudinal growth occurs at the growth plate by endochondral ossification, a process in which cartilage is formed and then remodeled into bone (1,2). The multistep process of chondrogenesis involves: 1) the aggregation of mesenchymal cells into prechondrogenic condensations, 2) differentiation of the condensed prechondrogenic mesenchyme into small immature chondrocytes, and 3) chondrocyte maturation followed by ossification, starting in the center of the element.
Bone morphogenetic proteins (BMPs) 1 play an important role in skeletal development (3)(4)(5). At the early stage of bone development, BMP-2 and BMP-4 are expressed in specific areas of the limb bud 3 and misexpression of noggin, a glycoprotein that blocks actions of BMP-2 and BMP-4, prior to the onset of chondrogenesis leads to the total absence of skeletal elements (6), suggesting that BMP signaling is absolutely required for prechondrogenic condensations. At the late stage, BMP-2, expressed in the growth plate of long bone, accelerates longitudinal bone growth by stimulating growth plate chondrocyte proliferation and hypertrophy (7,8).
To understand the regulatory mechanism of BMP-2 gene, we have previously characterized the BMP-2 promoter and identified control elements in osteoblasts and chondrocytes in BMP-2 gene (9). BMP-2 autoregulates its own expression through its proximal promoter (10) in a similar fashion that the decapentaplegic (dpp) gene is regulated in drosophila. dpp is a drosophila homologue of BMP-2. In drosophila, expression of dpp is regulated by dorsal, a homologue of mammalian NF-B (11,12). In chick embryo, overexpression of a mutated form of I-B␣ protein, which blocks NF-B activation, leads to abnormality in limb bud development, and changes in downstream signals include BMP-4 (13). These findings suggest that NF-B may participate in the regulatory process of the BMP-2 gene.
NF-B is a group of transcription factors including five members, p65, c-Rel, RelB, p50, and p52. They form homodimers or heterodimers to regulate expression of a variety of genes and are involved in many biological functions (14). It has been demonstrated previously that NF-B family members p50 and p52 are absolutely required for osteoclast development and p50/p52 double knock-out (dKO) mice failed to form osteoclasts and developed severe osteopetrosis (15).
The role of NF-B in BMP-2 gene expression and functions of NF-B in other bone cells, such as chondrocytes, have not been reported. In the present studies, we have examined the regulatory mechanism of BMP-2 by NF-B in chondrocytes in vitro and in vivo. We searched the Ϫ2712/ϩ165 region of BMP-2 gene and found two putative NF-B response elements. Using different experimental approaches, we demonstrate that these NF-B response elements are functional in chondrocytes and NF-B positively regulates BMP-2 gene transcription through these response elements. We also show that in growth plate chondrocytes of 2-week-old NF-B p50/p52 dKO mice, BMP-2 mRNA expression is significantly decreased. Corresponding to a decreased expression of BMP-2, chondrocyte proliferation is also significantly reduced in growth plate chondrocytes of p50/p52 dKO mice. These results suggest that expression of BMP-2 may be controlled, at least in part, by NF-B in chondrocytes, which may have important impacts in late stage chondrogenesis.

EXPERIMENTAL PROCEDURES
BMP-2 Promoter-LacZ Transgenic Mice-The BMP-2 promoter fragment (Ϫ2712 to ϩ165) linked to AUG-LacZ plasmid (gift of Dr. Eric Olson, University of Texas Southwest Medical Center, Dallas, TX) was removed and purified from vector, and the transgene was injected into eggs from CB6F1 females. After fertilization the eggs were implanted in pseudopregnant CB6F1 females and lines were screened for the presence of the transgene by Southern hybridization. Transgenic lines had 2-20 copies per genome, and LacZ staining patterns were independent of copy numbers. Three lines of transgenic mice with the Ϫ2712/ϩ165 bp BMP-2 promoter construct were evaluated and all gave similar patterns of expressions at 13.5 to 15.5 days postcoitum (dpc).
LacZ Staining-Male transgenic animals containing the BMP-2 promoter-LacZ transgene were crossed to CB6F1 females and embryos were collected at appropriate times, fixed, and stained overnight as described by Lazik et al. (16). Embryos were embedded in plastic and sectioned at 15 m and viewed using Zeiss Axioplot microscope under dark-field illumination (16).
Site-specific mutations were created in the intact promoter by using PCR-directed mutagenesis (Stratagene, La Jolla, CA) on a construct containing a 2.7-kb BMP-2 promoter (Ϫ2712/ϩ165) or BMP-2 minimal promoter (Ϫ150/ϩ165) upstream of a luciferase reporter gene. The designed mutations were verified by sequencing.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extract was prepared from TMC-23 cells. Double-stranded oligonucleotides corresponding to consensus NF-B response element were prepared and used in competition assays. NRE-1 and NRE-2 binding sites (Table I) were end-labeled and purified as described previously (17). Five fmol of labeled oligonucleotide was incubated with 10 g TMC-23 cell nuclear extract. The incubation mix for nuclear extract binding assay consisted of binding buffer and 2 g/ml poly(dI-dC). Incubation took place at room temperature for 5 min. Supershift experiments were carried out as described above, except that the TMC-23 nuclear extract was preincubated for 10 min at room temperature with antibodies against p50, p52, and p65 in binding buffer prior to their incubation with the labeled oligonucleotide.
In Situ Hybridization-Bone samples were fixed overnight in fresh 10% paraformaldehyde in RNase-free phosphate-buffered saline, transferred to ethanol series to 100%, followed by methanol and xylene, and finally embedded in wax. 10 m sections were placed on microscope slides, dewaxed, refixed, and utilized for in situ hybridization. A 1.0-kb mouse BMP-2 RNA probe was prepared and hybridized to appropriate sections as described by Wilkinson and Nieto (18) and Shamim et al. (19). To analyze changes in message levels of BMP-2 and BMP-6 in NF-B p50/p52 dKO mice, in situ hybridization slides were viewed by the Nikon E400 microscope that is linked to a color video monitor, and images were captured. The message levels of BMP-2 and BMP-6 were quantitated using a software Image-Pro Plus, version 3.0 (Media Cybernetics, Silver Spring, MD).
BrdUrd Labeling-BrdUrd labaling was performed using Zymed Laboratories Inc. (South San Francisco, CA) BrdUrd labeling reagent and BrdUrd staining kit according to manufacturer's protocols (Zymed Laboratories Inc.). Briefly, BrdUrd was injected intraperitonally into mice 24 h before the animals were sacrificed. Bone tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. 3-4-m thick sections were cut and placed on polylysine-coated slides. Slides were dried in a 60°C oven for overnight and deparaffinized in 2 changes of xylene for 5 min each and then rehydrated in a series of graded alcohol. Slides were stained for BrdUrd according to manufacturer's protocols. Streptavidin-peroxidase was used as a signal generator for the BrdUrd system, and deaminobenzidine in the presence of hydrogen peroxide was used as a chromogen, staining BrdUrd-incorporated nuclei dark brown.

RESULTS
To understand the regulatory mechanism of BMP-2 gene in cartilage and bone tissues, previously we have isolated the BMP-2 gene and characterized the 2.7-kb BMP-2 promoter in vitro (9,10). To further determine whether this 2.7-kb BMP-2 promoter fragment controls BMP-2 gene expression in cartilage and bone in vivo, we generated transgenic mice which express lacZ reporter gene driven by the 2.7-kb BMP-2 promoter (Ϫ2712/ϩ165-LacZ). We analyzed the LacZ expression pattern of these mice and found strong expression of lacZ reporter gene in cartilage perichondrium and proliferating chondrocytes in 15.5 dpc mouse embryos (Fig. 1). As shown in Fig. 1A, a whole mount staining of LacZ showed predominate expression in almost all developing bones, mainly at the perichondrium and periosteum in this 15.5-dpc embryo. In Fig. 1B, an example of LacZ staining is shown, using darkfield microscopy, where the LacZ staining is red, in the perichondrium and  growing chondrocytes of the manubrium of a 15.5-dpc embryo. Fig. 1C showed the in situ hybridization of BMP-2 32 P-cRNA probe to a similar embryo in the manubrium and ventricle of the heart. We also detected strong staining of LacZ in the hypertrophic regions of several endochondral bones, indicating that this promoter fragment of the BMP-2 gene can direct BMP-2 expression in this region of the growth plate (data not shown).
To determine whether NF-B regulates BMP-2 gene transcription in chondrocytes, we analyzed the sequence of the Ϫ2712/ϩ165 region of BMP-2 gene. Two putative NF-B response elements were found and are designated as NRE (NF-B response element)-1 and NRE-2. The NRE-1 is located in the 5Ј-flanking region and the NRE-2 is located in Exon 1 of BMP-2 gene (Table I and Fig. 2a). To determine whether these NF-B response elements are functional in chondrocytes, we co-transfected two different mutant mouse I-B␣ expression plasmids with BMP-2 promoter (Ϫ2712/ϩ165)-luciferase reporter construct into TMC-23 cells, a clonal chondrocyte cell line (20). In one mutant I-B construct, the inducible phosphorylation sites (Ser 32 and Ser 36 ) and constitutive phosphorylation site at its carboxyl terminus were substituted with alanine residues (mI-B␣), and the other mutant was generated by substituting three lysine residues at the ubiquitination sites (I-B⌬Ub). These mutant I-B␣ proteins retain their ability to bind with NF-B proteins and cannot be degraded properly and therefore block NF-B activity in vitro and in vivo (13). Transfection of mutant I-B␣ expression plasmids inhibited BMP-2 promoter activity ϳ50% (Fig. 2b), indicating that NF-B response elements in BMP-2 gene serve as positive regulatory elements for BMP-2 gene transcription. To confirm that these mutant I-B␣ expression plasmids do inhibit NF-B activity in chondrocytes, we co-transfected them with pNF-B-luc reporter construct (Promega), which contains six copies of NF-B consensus response elements, into TMC-23 cells. Transfection of mutant I-B␣ expression plasmids significantly inhibited luciferase activity of the NF-B reporter construct (data not shown).
To confirm that NF-B transcription factors are transcriptional activators for BMP-2 gene, we transfected BMP-2 promoter luciferase construct into TMC-23 cells and treated cells with NF-B inducer TNF-␣ and NF-B inhibitor pyrrolidine dithiocarbamate (PDTC). TNF-␣ induces phosphorylation of I-B␣ and causes dissociation of I-B␣⅐NF-B complex, resulting in nuclear translocation of NF-B. Treatment with TNF-␣ (50 -200 ng/ml) for 24 h caused a dose-dependent increase in BMP-2 promoter activity in TMC-23 cells (Fig. 2c). In contrast, treatment with PDTC (0.25-1 M) for 24 h significantly inhibited BMP-2 promoter activity in a dose-dependent manner in the same cells (Fig. 2c).
To determine the function of NRE-1 in BMP-2 gene transcription, we mutated 3 nucleotides in this response element ( Table I) and found that BMP-2 promoter activity was decreased over 30% when NRE-1 site is mutated (Fig. 2b). Cotransfection of mutant I-B␣ expression plasmids with the 2.7-kb BMP-2 promoter, in which three nucleotides at NRE-1 site were mutated, caused reduction in BMP-2 promoter activity in TMC-23 cells (Fig. 2b), suggesting that NRE-2 site located at Exon 1 may also serve as a positive regulatory element. We then examined the effects of mutant I-B␣ on BMP-2 minimal promoter (Ϫ150/ϩ165). The luciferase activity of BMP-2 minimal promoter is about 30% lower than that of larger BMP-2 promoter (Ϫ2712/ϩ165) in TMC-23 cells (data not shown). Transfection of mutant I-B␣ expression plasmids with BMP-2 minimal promoter (Ϫ150/ϩ165) suppressed the activity of BMP-2 minimal promoter in TMC-23 cells (Fig. 2d). Mutation of NF-B response element NRE-2 in this region completely abolished the inhibitory effect of mutant I-B␣ (Fig.  2d). When both NRE-1 and NRE-2 sites of BMP-2 promoter (Ϫ2712/ϩ165) were mutated, the promoter activity was reduced about 50%. Mutant I-B␣ had no significant effects on this mutant BMP-2 promoter (Fig. 2e). Taken together, these results suggest that both NRE-1 and NRE-2 sites in BMP-2 gene are functional and NF-B transcription factors positively regulate BMP-2 gene transcription.
To examine the binding of NF-B to its response element in BMP-2 gene, we performed EMSA. The nuclear proteins were extracted from TMC-23 cell nuclear preparations. Labeled oligonucleotides containing NRE-1 and NRE-2 sequences of the BMP-2 gene (Table I) were incubated with TMC-23 cell nuclear extract, leading to the formation of a protein-DNA complex that migrated at the same location as that formed upon incubation of TMC-23 cell nuclear extract with labeled oligonucleotides containing NF-B consensus sequence (Table I and Fig. 3a). Addition of different amounts of unlabeled oligonucleotides containing NF-B consensus element inhibited formation of this protein-DNA complex in a dose-dependent manner (Fig.  2a). Addition of unlabeled oligonucleotides containing Sp1 element had no significant effects on NF-B binding to NRE-1 response element (data not shown). To show that NF-B was part of this protein-DNA complex, supershift experiments were performed using antibodies against different NF-B subunits, p50, p52, and p65. The incubation of TMC-23 nuclear extract with antibodies against p50 and p65 prior to the addition of labeled oligonucleotides led to the formation of a slower mobility complex. This complex was specific, as it was not observed when using an antibody against p52 (Fig. 3b).
To determine whether NF-B regulates BMP-2 gene expression in chondrocytes in vivo, we examined BMP-2 mRNA expression in 2-week-old NF-B knock-out (KO) mice and their wild-type littermates by in situ hybridization (n ϭ 3 in each group and three non-consecutive sections per mouse were analyzed). In wild-type mice, BMP-2 mRNA was predominantly expressed in the lower portion of proliferating zone, the junction between the proliferating and hypertrophic zone and the upper layer of hypertrophic zone of growth plate. BMP-2 was also highly expressed in articular chondrocytes (Fig. 4a, upper  panel). In NF-B p50 KO or p52 KO mice, no significant changes in BMP-2 mRNA expression were observed (data not shown). In contrast, in NF-B p50/p52 dKO mice, BMP-2 mRNA expression was significantly reduced over 80% in growth plate chondrocytes (Fig. 4, a and c). In articular chondrocytes, BMP-2 expression remains unchanged (Fig. 4a, lower  panel). The effect of NF-B on BMP-2 expression seems specific, since expression of BMP-6 mRNA in growth plate chondrocytes was not significantly changed in p50/p52 dKO mice (Fig 4, b and c).
It has been reported that BMP-2 stimulates chondrocyte proliferation (7,8). To investigate whether decreased BMP-2 expression in NF-B p50/p52 dKO mice causes reduction in chondrocyte proliferation, we monitored BrdUrd incorporation in growth plate chondrocytes in these mice. The numbers of BrdUrd-positive cells were decreased over 65% in proliferating chondrocytes in dKO mice (Fig. 5, a and b). These results suggest that reduced BMP-2 expression in growth plate chondrocytes in dKO mice has important biological impacts on growth plate chondrogenesis. DISCUSSION The major finding of this study was that NF-B selectively modulates BMP-2 mRNA expression in growth plate chondrocytes in vivo and directly regulates BMP-2 gene transcription growth plate is dramatically reduced and indicated by red arrows (lower panel) (a). In contrast, no significant changes in BMP-6 mRNA was observed in p50/p52 dKO mice (b). c, the expression levels of BMP-2 and BMP-6 in growth plate chondrocytes in wild-type and p50/p52 dKO mice were quantitated. The BMP-2 message was reduced about 80%, while BMP-6 expression remained unchanged. *, p Ͻ 0.01, unpaired t test, compared with BMP-2 expression in wild-type mice. in a chondrocyte cell line in vitro. In the Ϫ2712/ϩ165 region of the BMP-2 gene, two NF-B response elements were identified. Mutations of the NRE-1 and NRE-2 sites decreased BMP-2 promoter activity and responses to expression of mutant I-B␣ proteins in chondrocytes. In p50/p52 dKO mice, BMP-2 mRNA expression and cell proliferation were dramatically reduced in growth plate chondrocytes. These findings demonstrate that NF-B is a transcriptional activator for BMP-2 gene transcription in chondrocytes. Of note, it has been previously reported that the NF-B response element serves as a positive regulatory element when it is located adjacent to a Sp1 site (21). Indeed, in BMP-2 promoter, next to NRE-1 site, two overlapped Sp1 sites were found (Table I). Therefore, it is likely that NF-B and Sp1 transcription factors interact to coordinate BMP-2 gene regulation.
NF-B is a family of transcription factors, which are composed of five subunits, p65, c-Rel, RelB, p50, and p52, among which p50 and p52 are two highly homologous members and are the most frequent partners among family members (14). In our gel shift assay, we only detect binding of p50 and p65 to NRE-1 site of BMP-2 gene but not p52 in chondrocyte TMC-23 cells. Consistent with these observations, expression of p50 and p65, but not p52, proteins was detected by Western blot in TMC-23 cells (data not shown). Previously it has been shown that p50/p52 dKO mice are impaired in the development of osteoclasts and B cells, suggesting that these two subunits of NF-B play indispensable role in osteoclast and B cell development (15). In the present studies, defects in BMP-2 gene expression at growth plate chondrocyte area were also observed in p50/p52 dKO mice, suggesting that these two factors, which may be produced by different subpopulations of chondrocytes, play a redundant role in regulating BMP-2 gene expression.
The major function of BMP-2 in the growth plat of postnatal mice is to control longitudinal bone growth by stimulating chondrocyte proliferation and hypertrophy (7,8). In the present studies, we have demonstrated that NF-B positively regulates BMP-2 gene transcription, and in NF-B p50/p52 dKO mice, expression of the BMP-2 message in the growth plate chondrocytes is dramatically reduced, which leads to a significant decrease in chondrocyte proliferation (67%). Consistent with these findings, NF-B p50/p52 dKO mice showed retarded growth and shortened long bones (15). Decreases in BMP-2 expression and chondrocyte proliferation in NF-B p50/p52 dKO mice may at least partially contribute to this phenotype.
We have found that in p50/p52 dKO mice, the hypertrophic zone of the growth plate was expanded, indicating the normal process of chondrocyte hypertrophy was disrupted. Since MMP-9 produced by osteoclasts plays a key role in regulation of cell apoptosis of hypertrophic chondrocytes (22), and dKO mice have no osteoclasts, it is difficult to evaluate the contribution of BMP-2 in chondrocyte hypertrophy under this circumstance. To dissect the direct effect of NF-B on chondrocyte hypertrophy from its indirect effect through osteoclasts, tissue-or cellspecific approaches are required, such as generating transgenic mice in which NF-B signaling is disrupted specifically in chondrocytes. Nevertheless, reduction in cell proliferation in dKO mice suggests that NF-B-regulated BMP-2 expression does play an important role in chondrocyte function.
In addition to direct regulation of BMP-2 gene transcription, NF-B may also regulate BMP-2 gene expression through indirect mechanisms and other transcription factors may also play a role in BMP-2 gene transcription in growth plate chondrocytes. The summation of these transcriptional inputs determines the spatial and temporal expression pattern of BMP-2 gene in growth plate chondrocytes in vivo.