Transcriptional Induction of ADAMTS5 Protein by Nuclear Factor-κB (NF-κB) Family Member RelA/p65 in Chondrocytes during Osteoarthritis Development*

Background: ADAMTS5 is a crucial proteinase for osteoarthritis development. Results: NF-κB family member RelA/p65 was identified as a potent transactivator of ADAMTS5 and regulated the expression and aggrecanolysis. Conclusion: RelA/p65 is a potent transcriptional activator of ADAMTS5 in chondrocytes during osteoarthritis development. Significance: The molecular network related to the RelA/p65-ADAMTS5 axis may represent a therapeutic target for osteoarthritis. Here we sought to identify transcription factors that induce ADAMTS5, a crucial proteinase for osteoarthritis development. Exhaustive comparison of the genomic sequences of human, macaque, and mouse ADAMTS5 genes revealed that the proximal 1.4 kb of the 5′-end-flanking regions containing several consensus motifs was highly conserved. Among putative transcription factors for these motifs, NF-κB family member RelA/p65 most strongly stimulated the promoter activity. In the ADAMTS5 gene, there were three NF-κB binding motifs, in which deletion, mutagenesis, and tandem repeat analyses of the luciferase assay identified the core responsive elements of RelA/p65 to be −896/−887 and −424/−415 bp with specific bindings. The endogenous Adamts5 expression in ATDC5 cells was increased by RelA/p65 overexpression and decreased by knockdown through its siRNA. The expression was also inhibited by the Rela deletion through Cre transfection in primary articular chondrocytes from Relafl/fl mice. In the ex vivo culture of femoral head cartilage from mesenchymal cell-specific Rela knock-out (Prx1-Cre;Relafl/fl) mice, aggrecanolysis was significantly lower than that in the Relafl/fl cartilage. Finally, in the experimental mouse osteoarthritis model, ADAMTS5 and RelA were co-localized in chondrocytes of degraded articular cartilage. We conclude that RelA/p65 is a potent transcriptional activator of ADAMTS5 in chondrocytes during osteoarthritis development.

Osteoarthritis (OA) 2 is a progressive disease of the joints characterized by degradation of articular cartilage, which appears to be a result of uncontrolled proteolytic extracellular matrix destruction. A major component of the cartilage extracellular matrix is aggrecan, a proteoglycan that imparts compressive resistance to the tissue. Aggrecan degradation by aggrecanases is a significant event in early stage OA (1,2). The aggrecanases are members of the ADAMTS family, which are the disintegrin and metalloprotease with thrombospondin motifs and consist of secreted zinc metalloproteinase with a precisely ordered modular organization. In the cartilage, two different aggrecanases, i.e. ADAMTS4 (aggrecanase-1) and ADAMTS5 (aggrecanase-2), have been isolated. The knock-out mice of ADAMTS5 are protected from early aggrecan loss and cartilage erosion in the surgical mouse OA models (3,4). Because this protection was not seen in the ADAMTS4 knockout mice (5), ADAMTS5 is thought to be substantially more important than ADAMTS4 for OA development (6). Although ADAMTS4 expression has been reported to be induced by transforming growth factor-␤ (TGF-␤) and proinflammatory cytokines in chondrocytes, less is known about the regulation of ADAMTS5 expression (7)(8)(9)(10)(11). Despite the substantial information about the expression profiles and the substrates of ADAMTS5 (1), little is known about the upstream signaling or the functional regulation of the ADAMTS5 promoter. To elucidate the molecular network as a therapeutic target of OA, the present study attempted to identify transcription factors to induce ADAMTS5 expression and to examine the underlying mechanism.
We initially compared the genomic sequences of proximal promoter regions among mouse, macaque, and human ADAMTS5 genes and identified highly conserved regions containing putative transcription factor binding motifs. Among the candidate transcription factors, our further analyses found that nuclear factor B (NF-B) family member RelA (NF-B p65) most strongly activated the human ADAMTS5 promoter activity. The NF-B family of transcription factors plays a crucial role in a broad range of biological processes, including immune responses, inflammation, proliferation, differentiation, and apoptosis (12)(13)(14). The family includes RelA/p65, RelB, Rel, p105/p50, and p100/p52, each of which contains a Rel homology domain that mediates DNA binding and dimerization. Numerous studies have established that IB proteins are phosphorylated and degraded by a large protein complex IB kinase (IKK) in response to several signals, thereby allowing free NF-B complexes to translocate from the cytoplasm into the nucleus, leading to target gene transactivation (15,16). The NF-B family genes are expressed in the chick limb cartilage, and blockage of the NF-B activity causes the arrest of the limb outgrowth (17). The IB kinase ␣-deficient mice also exhibit suppression of limb outgrowth (18,19). RelA/p65 is a key active subunit in NF-B transcription in several cell types (20,21), and the knockdown of RelA/p65 with specific siRNA inhibits the expression of cartilage catabolic factors such as matrix metalloproteinases-9 (MMP-9), nitric oxide synthase-2 (NOS-2), and cyclooxygenase-2 (COX-2) in chondrocytes (22). Because these lines of evidence implicate the interaction between RelA/ p65 and ADAMTS5 signals in cartilage, the present study investigated the mechanism underlying the transcriptional regulation of the ADAMTS5 promoter by RelA/p65.

EXPERIMENTAL PROCEDURES
Similarity Search of the Proximal Promoter Sequences among Species-We compared the sequences of the 5Ј-end-flanking regions relative to the transcription start site (TSS) among 2-kb human, macaque, and mouse ADAMTS5 genes, using BLASTN search (23). TSS and exon1 were determined from the ADAMTS5 gene information of the GenBank TM database. The detected sequences were aligned by the Vector-NTI software (Invitrogen), and the transcription factor-binding motifs were predicted using the TFSEARCH web site (Computational Biology Research Center).
Construction of Expression Vectors-Full-length human cDNA sequences of the transcription factors were PCR-amplified and cloned into pCMV-HA vector (Clontech Luciferase Assay-We prepared the ADAMTS5 promoter region (from Ϫ1,400 to ϩ1 bp relative to the TSS) by PCR using human genomic DNA as the template and cloned it into the pGL4.10 (luc2) vector (Promega). We created deletion and mutation constructs by PCR. Furthermore, tandem repeat constructs containing Ϫ896/Ϫ887 and Ϫ424/Ϫ415 bp were created by ligating the double-strand oligonucleotides of Ϫ903/ Ϫ869 and Ϫ442/Ϫ410 bp, respectively, into the EcoRI site of the modified pGL3 vector. Transfection of ATDC5 cells was performed in triplicate in 48-well plates using FuGENE 6 transfection reagent (Roche Diagnostics): FuGENE 6 with a total amount of 150 ng of plasmid DNA, 100 ng of pGL3 or pGL4 reporter vector, 50 ng of effector vector, and 4 ng of pRL-TK vector (Promega) for internal control per well. For the doseresponse experiment of RelA/p65, we used 25, 50, and 75 ng of the expression vector. We performed luciferase assays with the PicaGene dual sea pansy luminescence kit (Toyo Ink) using a GloMax 96 microplate luminometer (Promega) and showed the data as the ratio of the firefly activities to the Renilla activities.
Electrophoretic Mobility Shift Assay (EMSA)-We prepared nuclear extracts from COS-7 cells transfected with expression vector of RelA/p65 using FuGENE 6 transfection reagent and performed the EMSA with the digoxigenin gel shift kit (Roche Diagnostics). Regions of the oligonucleotide probes were about 20 -30 bp containing Ϫ896/Ϫ887 and Ϫ424/Ϫ415 bp. Binding reactions were incubated for 30 min at room temperature. For competition analysis, we used 50-fold excess of unlabeled competitor probe containing the binding reaction. For the supershift experiment, we added 1 l of an antibody to RelA/p65 (C-20; Santa Cruz Biotechnology) after 30 min of the binding reaction, and the reaction was incubated for an additional 30 min at room temperature. Samples were loaded onto Novex 6% Tris borate-EDTA gels (Invitrogen) and electrophoresed at 100 V for 60 min.
Chromatin Immunoprecipitation (ChIP) Assay-The ChIP assay was performed with a OneDay ChIP kit (Diagenode, Liege, Belgium) according to the manufacturer's instructions. In vivo cross-linking was performed after 1 day with and without IL-1␤ (10 ng/ml) treatment in the culture of human chondrogenic OUMS-27 cells, and then the cell lysates were sonicated to shear genomic DNA. For immunoprecipitation, an antibody to RelA/p65 and the control normal rabbit immunoglobulin G (IgG) were used. Primer sets that amplify the area including the NF-B motif (Ϫ951/Ϫ793 or Ϫ494/Ϫ326 for Ϫ896/Ϫ887 or Ϫ424/Ϫ415, respectively) or that do not include the motif (ϩ17/ϩ213) were employed. PCR was performed using KOD FX Neo (TOYOBO).
Real-time RT-PCR-Total RNA from cells was isolated with an RNeasy mini kit (Qiagen) according to the manufacturer's instructions, and an aliquot (1 g) was reverse-transcribed with QuantiTect reverse transcription (Qiagen) to make singlestranded cDNA. Real-time RT-PCR was performed with an ABI Prism 7000 sequence detection system (Applied Biosystems) using QuantiTect SYBR Green PCR master mix (Qiagen) according to the manufacturer's instructions. Standard plasmids were synthesized with a ZERO Blunt TOPO TA Cloning Kit (Invitrogen), according to the manufacturer's instructions. All reactions were run in triplicate. Copy numbers of target gene messenger RNA (mRNA) in each total RNA were calculated by reference to standard curves and were adjusted to the human or mouse standard total RNA (Applied Biosystems) with the human GAPDH or rodent GAPDH as an internal control. The primer sequences are available upon request.
Western Blotting-Cells were lysed in M-PER mammalian protein extraction reagent (Thermo Scientific). The cell lysates were fractionated by SDS-PAGE and transferred onto nitrocellulose membranes (Bio-Rad). The membranes were incubated with primary antibodies to RelA/p65 (C-20; Santa Cruz Biotechnology) and actin (Sigma). Membranes were incubated with HRP-conjugated antibody (Promega), and immunoreactive proteins were visualized with ECL Plus (Amersham Biosciences).
Gene Silencing-For siRNA experiments, ATDC5 cells were transfected with siRNA using FuGENE 6 transfection reagent according to the manufacturer's protocol. After 48 h of culture, cells were washed and placed in DMEM supplemented with or without IL-1␤ (10 ng/ml). After 24 h of culture, RNA was harvested with TRIzol reagent (Invitrogen) for gene expression analysis.
Mice-All experiments were performed according to protocols approved by the Animal Care and Use Committee of the University of Tokyo. In each experiment, we compared genotypes of littermates that were maintained in a C57BL/6 background. We purchased Prx1-Cre mice from The Jackson Laboratory (25). To generate Prx1-Cre;Rela fl/fl mice, Rela fl/fl mice were mated with Prx1-Cre mice to obtain Prx1-Cre;Rela fl/ϩ mice and then mated with Rela fl/fl mice. To harvest articular chondrocytes and femoral heads, we used two mice/group/genotype. Proteoglycan Release Assay-Proteoglycan release was assessed as reported previously (26). Briefly, femoral heads were harvested from 3-week-old Rela fl/fl and Prx1-Cre;Rela fl/fl mice. The explants were cultured for 3 days with or without IL-1␤ (10 ng/ml) in DMEM containing 1% penicillin/streptomycin. The proteoglycan content in the medium was measured as sulfated glycosaminoglycan by a colorimetric assay using dimethylmethylene blue.
Histological Analysis-Tissue was dissected and fixed in 4% paraformaldehyde buffered with PBS (pH 7.4) for 1 day at 4°C. The specimens were decalcified for 2 weeks with 10% EDTA (pH 7.4) at 4°C. After dehydration with an increasing concentration of ethanol and embedding in paraffin, 4-m sagittal sections were cut from the specimens. Safranin-O staining was performed according to standard protocols. For immunohistochemistry, we incubated the sections with antibodies to RelA/ p65 (C-20) (1:200; Santa Cruz Biotechnology) and ADAMTS5 (H-200) (1:500; Santa Cruz Biotechnology) or the control nonimmune IgG diluted in blocking reagent without antigen retrieval. For immunofluorescence, we used a secondary antibody conjugated with Alexa Fluor 568 (Invitrogen) for RelA/ p65 and a CSA II biotin-free catalyzed amplification system (DAKO) for ADAMTS5. For immunoperoxidase methods in ADAMTS5 detection, we also used the CSA II system and applied methyl green for counterstaining. These histological analyses were performed at least three times using 2-3 mice/ group/genotype for the confirmation of the results.
OA Experiment-We performed the surgical procedure to create an experimental OA model with removal of the medial collateral ligament and transection of the medial meniscus in knee joints of 8-week-old male mice, as reported previously (27,28). Briefly, the knee joint was exposed following a medial capsular incision and gentle lateral displacement of the extensor mechanism without transection of the patellar ligament. The medial collateral ligament was transected, and the medial meniscus was removed without anterior cruciate ligament transection. We analyzed various regions of the joint cartilage 8 weeks after surgery. To confirm the reproducibility, the analyses were performed at least three times using three mice.
Statistical Analyses-We performed statistical analyses of experimental data with the unpaired two-tailed Student's t test. p values less than 0.05 were considered significant.

Identification of Transactivators of the ADAMTS5 Promoter by Comparison of Mouse, Macaque, and Human Genes-To
identify transcription factors that activate the ADAMTS5 promoter, we initially performed exhaustive comparison of the sequences of about 2 kb of the 5Ј-end-flanking regions among mouse, macaque, and human genes and found that the 1.4 kb upstream from the TSS was about 80% conserved among all the species (Fig. 1A). The sequence search identified the binding motifs of HIF, CCAAT/enhancer binding protein (C/EBP), RUNX, NF-B, STAT, OCT, SOX, and SP-1 in the highly conserved region (Fig. 1B).
Identification of the Core Region Responsive to RelA/p65 in the ADAMTS5 Proximal Promoter-We initially confirmed that RelA/p65 activated the ADAMTS5 promoter activity in a FIGURE 1. Identification of transcription factors of the ADAMTS5 promoter. A, the sequence similarity of the proximal promoter (2 kb) sequences among the mouse, macaque, and human ADAMTS5 genes analyzed by BLASTN search and Vector-NTI software (Invitrogen). B, conserved nucleotides in the two or three species are denoted by white letters shaded in gray or black, respectively. An arrow shows the TSS of the human gene. Putative binding motifs of transcription factors identified in highly conserved regions are indicated below. C, luciferase assay for the human ADAMTS5 promoter activity by the putative transcription factors whose binding motifs were identified above. ATDC5 cells were co-transfected with a luciferase reporter construct containing the 5Ј-end-flanking region (Ϫ1,400/ϩ1) of the human ADAMTS5 gene and the effecter vectors or the control empty vector (EV). *, p Ͻ 0.05, **, p Ͻ 0.01 versus EV. D, luciferase assay for the human ADAMTS5 promoter activity by the co-transfection of putative binding partners with RelA/p65 in the reporter construct containing ATDC5 cells as above. For C and D, data are shown as means (bars) Ϯ S.E. (error bars) of relative luciferase activity (the ratio of the firefly activities to the Renilla activities) for three wells/group.

Induction of ADAMTS5 by RelA/p65 in Chondrocytes
dose-dependent manner of the expression vector ( Fig. 2A). There were three NF-B motifs in the 1.4-kb region of the human ADAMTS5 proximal promoter: Ϫ1,196/Ϫ1,187 bp, Ϫ896/Ϫ887 bp, and Ϫ424/Ϫ415 bp (Fig. 1B). Hence, to identify the region responsive to RelA/p65 in the human proximal promoter, we performed the deletion analysis of the luciferase assay in ATDC5 cells co-transfected with RelA/p65. A series of 5Ј-deletions of the Ϫ1,400/ϩ1 fragment identified the decreases of the transcriptional activity moderately between Ϫ1,400 and Ϫ1,022 bp, and significantly between Ϫ1,022 and Ϫ811 bp and between Ϫ506 and Ϫ405 bp, all of which contained the three motifs (Fig. 2B). Hence, we next compared the luciferase activity by a single and double mutagenesis of the respective motifs (M1, M2, and M3 in one motif) and found that the promoter activities were significantly decreased by the mutagenesis in the Ϫ896/Ϫ887 and in the Ϫ424/Ϫ415 motif, although not in the Ϫ1,196/Ϫ1,187 motif (Fig. 2C). We further compared the luciferase activity of the tandem repeats of the Ϫ896/Ϫ887 and the Ϫ424/Ϫ415 motif elements and found that both responded to RelA/p65 in a repeat number-dependent manner, indicating that the two proximal motifs are responsive regions to RelA/p65 (Fig. 2D).
EMSA revealed the complex formation of the COS-7 nuclear extracts overexpressing RelA with the oligonucleotide probes including the Ϫ896/Ϫ887or Ϫ424/Ϫ415-bp region (Fig. 2E,  lane 2, arrowheads). The complex formation was abolished by mutations inside the NF-B motif (M1) (Fig. 2E, lane 3). Cold competition with excess amount of the unlabeled wild-type probe suppressed the complex formation (Fig. 2E, lane 4), whereas that with the unlabeled mutation probes did not affect it (Fig. 2E, lane 5). Both complexes by the Ϫ896/Ϫ887 and the Ϫ424/Ϫ415 oligonucleotide probes underwent supershifts with the addition of an antibody to RelA/p65 (Fig. 2E, lane 6), although not by the nonimmune IgG (Fig. 2E, lane 7). These results demonstrate the specific binding between RelA/p65 and the two NF-B motifs in the ADAMTS5 promoter.
ChIP assay showed in vivo binding of RelA/p65 with the ADAMTS5 promoter including the Ϫ424/Ϫ415-bp region in cell lysates of human chondrogenic OUMS-27 cells stimulated by IL-1␤, a putative inducer of the NF-B signal as well as OA development, although it was not detected with that including the Ϫ896/Ϫ887-bp region, indicating that the Ϫ424/Ϫ415-bp region may be more physiologically important for the RelA binding than the Ϫ896/Ϫ887-bp region (Fig. 2F). The specificity of the binding was confirmed because it was not immunoprecipitated by the nonimmune IgG, and no amplification was seen with a primer set that does not span the NF-B motif.
Effects of Gain and Loss of Functions of RelA/p65 on ADAMTS5 Expression and Aggrecanolysis-We then looked at the endogenous ADAMTS5 expression by gain and loss of functions of RelA/p65 in chondrocytes. Adenoviral overexpression of RelA/p65 increased the endogenous Adamts5 expression in ATDC5 cells (Fig. 3A). IL-1␤ enhanced Adamts5 and Rela expressions in ATDC5 cells. The Adamts5 induction by IL-1␤ was suppressed by the knockdown of Rela through its specific siRNA transfection (Fig. 3B). Furthermore, Adamts5 expression was inhibited in a dose-dependent manner by the Rela deletion through adenoviral transfection with Cre in primary articular chondrocytes derived from mice homozygous for a floxed Rela allele (Rela fl/fl mice) (Fig. 3C).
Moreover, we used mesenchymal cell-specific Rela knockout (Prx1-Cre;Rela fl/fl ) mice. Primary chondrocytes derived from articular cartilage of the Prx1-Cre;Rela fl/fl mice were confirmed to express Rela in small amounts as compared with the control Rela fl/fl chondrocytes (Fig. 4A, upper panel). ADAMTS5 expression was stimulated by IL-1␤ treatment in the Rela fl/fl chondrocyte culture; however, the basal level was much lower in the Prx1-Cre;Rela fl/fl chondrocyte culture and did not respond to IL-1␤ treatment (Fig. 4A, lower panel). In the ex vivo culture of femoral head cartilage of the Rela fl/fl mice, the release of proteoglycan into the medium was stimulated by IL-1␤ treatment; however, this stimulation was abrogated in the Prx1-Cre;Rela fl/fl cartilage (Fig. 4B). Immunohistochemical analysis revealed that ADAMTS5 expression was lower in the Prx1-Cre;Rela fl/fl chondrocytes than in the Rela fl/fl chondrocytes in the presence and absence of IL-1␤ (Fig. 4C). These indicate that loss of function of RelA reduces ADAMTS5 expression and subsequent aggrecanolysis of the articular cartilage.
Co-localization of RelA and ADAMTS5 in Chondrocytes of the Degraded Articular Cartilage-To know the possible interaction between RelA and ADAMTS5 and in vivo, we next created an experimental mouse OA model by surgical induction of The arrowheads indicate the protein-probe binding. Cold competition with 50-fold excess of unlabeled WT or M1 probe is presented. A supershift of the binding by an antibody to RelA or nonimmune IgG is also shown. F, ChIP assay for specific binding of RelA to the NF-B motif using cell lysates of human chondrogenic OUMS-27 cells with or without IL-1␤ treatment. Primer sets that amplify the area including the NF-B motif (Ϫ896/Ϫ887 or Ϫ424/Ϫ415) or that do not included the motif (ϩ17/ ϩ213) were employed before (input) and after immunoprecipitation with an antibody to RelA or the control nonimmune IgG. instability in the knee joints of 8-week-old mice (27) and examined their localizations in various regions of the joint cartilage by immunohistochemistry. In the superficial layer with high degeneration, RelA was translocated into the nuclei, and ADAMTS5 was strongly stained pericellularly, both being colocalized in chondrocytes (Fig. 5, left column). Contrarily, in the superficial layer with low degeneration, RelA was localized in the cytoplasm and ADAMTS5 was decreased (Fig. 5, middle  column). Both were hardly detected in the deep layer (Fig. 5,  right column). These findings suggest an interaction between RelA and ADAMTS5 during OA development.

DISCUSSION
The present comparison of the 5Ј-end-flanking regions of ADAMTS5 genes among the three species found RelA/p65 to be a potent transcription factor of ADAMTS5. We speculate that RelA/p65 in chondrocytes transcriptionally induces ADAMTS5 expression at least in the process of OA development. Besides RelA/p65, a previous analysis of the 2.6-kb ADAMTS5 promoter sequence identified four RUNX motifs, to which RUNX2 exhibited strong bindings (30). In fact, OA development by hedgehog signaling is known to be mediated by ADAMTS5 transactivation by RUNX2 (31). Furthermore, C/EBP␤ is also reported to transcriptionally activate not only MMP-3, but also ADAMTS5 expression by binding to the Ϫ1,517/Ϫ1,509-bp motif (32). Although there were also C/EBP motifs in the Ϫ1,400/ϩ1 region we examined in the present study, the transactivation by C/EBP␤ was weak, at least in this region (Fig. 1C). However, a moderate decrease of the ADAMTS5 promoter activity between Ϫ1,400 and Ϫ1,022 bp containing the distal motif (Ϫ1,196/Ϫ1,187) ( Fig. 2A), despite there being no change by the mutagenesis in the motif (Fig. 2B), might be due to a deletion of binding motif of RUNX2 (Ϫ1,303/ Ϫ1,299) and/or C/EBP␤ (Ϫ1,310/Ϫ1,299) lying in this region. In addition, the thyroid hormone tri-iodothyronine is also known to up-regulate ADAMTS5 mRNA expression with subsequent aggrecan degradation in growth plate cartilage during endochondral ossification (33). In other tissues, a study using the computational tool PRI (Pattern-defined Regulatory Islands) identified transcription factors MyoD, myogenin, and the E-protein HEB as very strong myogenic transcriptional enhancers of ADAMTS5 expression through binding to the "coding" region (34). Hence, ADAMTS5 expression may be regulated by several factors distinctly depending on its functions and tissues.
During OA development, it is noteworthy that the activity of ADAMTS5 is regulated not only at the transcriptional level, but also by post-translational modifications starting with the processing of the proprotein form by proprotein convertases or proteases, such as furin, proprotein convertase 7 (PC7), paired basic amino acid cleaving enzyme 4 (PACE4), and syndecan-4 (35)(36)(37). ADAMTS5 is synthesized as latent, inactive zymogens and requires removal of the prodomain before the catalytic activity is obtained. Furin and a furin-like enzyme PC7 have major roles in activating ADAMTS enzymes by recognizing the furin-like recognition sequence. PACE4, which is known to bind extracellular matrix, also contributes to the processing/ activation of proADAMTS5. Syndecan-4 modulates the activa- (error bars) of three cultures/group using two mice/ genotype. C, representative images of immunohistochemical staining by an antibody to ADAMTS5 or the control nonimmune IgG in the cultured femoral head cartilages above (images with high magnification are inserted). Positive stainings are brown by diaminobenzidine with methyl green counterstaining. Scale bars, 20 m. Immunohistochemistry was performed at least three times using two mice/group/genotype and obtained reproducible results. tion of ADAMTS5 by controlling the release of MMP-3 rather than by transcriptionally regulating the expression. As RelA/ p65 is known to directly induce MMP-3 (38,39), RelA/p65 may also regulate the activity of ADAMTS5 at this level. MicroRNA and epigenetic factors are also involved in the ADAMTS5-related OA. OA-like changes in microRNA-140-knock-out mice can be attributed to elevated ADAMTS5 expression (40). In addition, in vitro treatment of chondrocytes with IL-1␤ suppresses microRNA-140 expression (41), suggesting the possibility of the regulation of microRNA-140 by RelA/p65. Histone deacetylase (HDAC) also potently inhibits IL-1-or oncostatin M-induced cartilage degradation mediated by ADAMTS5 (42). After ADAMTS enzymes have been synthesized, secreted from the cell, and activated by prodomain removal, their activity can be further modified by endogenous inhibitors such as the tissue inhibitor of metalloproteinases (TIMP). Of the four TIMP family members, TIMP-3 is the most potent inhibitor of ADAMTS5 (43). Hence, we surmise that RelA/ p65 is not the principal regulator of ADAMTS5, but is a member of a complicated molecular network for its expression and activation. Cooperation of other signals with RelA/ p65 will be needed to achieve strong ADAMTS5 function including OA development.
The present study showed that RelA/p65 is a potent transcriptional activator of a catabolic factor ADAMTS5 in chondrocytes. Similarly, we previously reported that RelA/p65 is a transactivator of early stage or anabolic factors for chondrocyte differentiation: SOX9, SOX6, and type II collagen ␣1 chain (COL2A1) (44,45). We also previously identified RelA/p65 as the most potent transcriptional factor of HIF-2␣ that controls later stages of chondrocyte differentiation and cartilage degradation (46). Furthermore, a recent study demonstrated that Nkx3.2 supports chondrocyte survival by activating RelA/p65 via a ligand-independent mechanism (47). Hence, RelA/p65 may extensively control anabolic and catabolic phases of cartilage during skeletal growth, cartilage homeostasis, and OA development with distinct upstream and downstream signals depending on the stages. Taken together, RelA/p65 and ADAMTS5 do not function in a one-to-one correspondence fashion. In fact, the global knock-out mice of Rela are embryonically lethal due to liver dysfunction (48), whereas those of Adamts5 exhibit minor abnormality with digit fusion and valve anomalies under physiological conditions (49,50).
Proinflammatory cytokines such as IL-1 and TNF-␣ are known to be representative ligands for the NF-B signal (51)(52)(53)(54). In fact, IL-1␤ enhanced Adamts5 and Rela expressions in ATDC5 cells (Figs. 3B and 4). Regarding therapeutic intervention, however, ADAMTS5 expression at least by human synoviocytes is not affected by neutralization of either IL-1 or TNF-␣ alone or in combination, whereas ADAMTS4 expression by the same cells is severely inhibited by the TNF-␣ blocker etanercept and an anti-IL-1 neutralizing antibody (9). Although it may not be simple to apply the present finding to its therapeutic application, the molecular network related to the RelA/ p65-ADAMTS5 axis may represent a therapeutic target for OA.