To define the role of Bcl-3, a member of the inhibitor of nuclear factor κB (NF-κB) family and a known regulator of NF-κB, in interleukin-1 (IL-1)–induced matrix metalloproteinase 1 (MMP-1) transcription in chondrocytes and synovial fibroblasts.
SW-1353 cells, a human chondrosarcoma cell line, were stimulated with IL-1β, and the harvested RNA was subjected to microarray analysis and quantitative real-time reverse transcription–polymerase chain reaction (RT-PCR). The SW-1353 cells were stimulated with IL-1 or transfected with a plasmid that constitutively expressed Bcl-3, and then MMP-1 messenger RNA (mRNA) expression was assayed by quantitative real-time RT-PCR. SW-1353 cells were transfected with antisense oligonucleotides to Bcl-3, and IL-1–induced MMP-1 mRNA expression was assayed by quantitative RT-PCR. SW-1353 cells and rabbit synovial fibroblasts were transfected with a 4.3-kb human MMP-1 promoter construct along with Bcl-3 and NF-κB1 expression constructs, and MMP-1 transcription was assayed.
Microarray analysis and real-time RT-PCR showed Bcl-3 to be an IL-1β–responsive gene in SW-1353 cells. Exogenous expression of Bcl-3 in SW-1353 cells activated MMP-1 transcription. Endogenous Bcl-3 expression was required for IL-1β induction of MMP-1 gene expression. Bcl-3 also activated MMP-1 transcription in primary synovial fibroblasts. We showed previously that NF-κB1 contributes to IL-1β induction of MMP-1 transcription in stromal cells. We showed here that Bcl-3 can cooperate with NF-κB1 to activate MMP-1 transcription in SW-1353 cells.
These data define a new role for Bcl-3 in joint cells as an IL-1β–responsive early gene involved in cell-mediated cartilage remodeling. Our findings implicate Bcl-3 as an important contributor to chronic inflammatory disease states, such as osteoarthritis and rheumatoid arthritis.
In osteoarthritis (OA) and rheumatoid arthritis (RA), patients experience erosion of articular cartilage, a process that involves degradation of type II collagen and proteoglycans (1, 2). Although the loss of proteoglycan reduces the shock-absorbing properties of articular cartilage, it is the collagen loss that correlates positively with the severity of arthritic lesions (3). This has led to the suggestion that degradation of type II collagen disrupts the architecture of articular cartilage and is, therefore, a committed step in the pathogenesis of OA (4) and RA (5).
The degradation of type II collagen in articular cartilage is facilitated by enzymes that are members of a family of neutral proteases named the matrix metalloproteinases (MMPs). The MMPs are zinc- and calcium-dependent enzymes responsible for degradation of the protein components of the extracellular matrix (6). Three members of this group of enzymes, the interstitial collagenases (MMP-1, MMP-8, and MMP-13), effectively digest type II collagen, and all 3 enzymes have been localized to arthritic lesions (4). Of these 3 enzymes, MMP-1 is the most ubiquitously expressed in connective tissue cells (7). Immunolocalization studies demonstrate that MMP-1 is the predominant collagenase produced by rheumatoid synovial fibroblasts and chondrocytes (8). While MMP-13 contributes primarily to the initial degradation of collagen in OA (9), inhibition of MMP-1, but not MMP-13, enhances new collagen synthesis in OA chondrocytes (10). Thus, MMP-1 plays a central role in the cartilage destruction observed in both RA and OA.
The nuclear factor κB (NF-κB) pathway plays a central role in inflammatory responses that are pertinent to arthritis (11), including activation of the MMP-1 gene (12–15). Heterodimers of NF-κB1/p50 and RelA/p65 are sequestered in the cytoplasm of resting cells by inhibitor of NF-κB (IκB). When cells are stimulated with inflammatory signals such as interleukin-1β (IL-1β), IκB becomes phosphorylated by a protein complex known as IκB kinase (IKK). IKK-dependent phosphorylation targets IκB for ubiquitination and proteosome-dependent degradation. With IκB degraded, the NF-κB1–RelA complex can translocate to the nucleus and bind to the promoters of responsive genes (16). We have previously demonstrated that the NF-κB pathway activates the MMP-1 promoter in IL-1β–stimulated synovial fibroblasts (SF) (14). Interestingly, we found that NF-κB1 bound to the MMP-1 IL-1β–responsive element, but that RelA was not detected in this complex. This result was unexpected because RelA contains a potent transactivation domain, but NF-κB1 does not (17).
NF-κB1 homodimers also bind to promoter sequences, and like heterodimers, nuclear translocation of these complexes is regulated in an IKK-dependent manner. The precursor form of NF-κB1, p105, associates with mature NF-κB1, and the p50–p105 complex is sequestered in the cytoplasm through the carboxy-terminal ankyrin repeats of p105 (18). In stimulated cells, IKK phosphorylates p105, and like IκB, this results in ubiquitination and degradation of p105 (19, 20). Degradation of p105 liberates p50, which forms homodimers that translocate to the nucleus. An IκB-related protein, Bcl-3, has recently been shown to complex with NF-κB1 dimers and assist with nuclear translocation and transcription activity (20–22). Since Bcl-3 contributes to the transcriptional competence of NF-κB1, we have examined the role of Bcl-3 in IL-1β–induced MMP-1 gene expression.
Here, we describe Bcl-3 as an IL-1β–responsive gene in chondrocytic cells, and we demonstrate that overexpression of Bcl-3 in these cells and in SF enhances basal and IL-1β–induced MMP-1 transcription. Furthermore, inhibition of Bcl-3 with antisense oligonucleotides in chondrocytes represses IL-1β–induced MMP-1 gene expression. This inhibition of MMP-1 in Bcl-3–repressed cells is not due to enhanced apoptosis. These studies demonstrate a novel role for Bcl-3 in inflammatory responses and implicate this gene in MMP-1–dependent cartilage degradation in arthritis.
MATERIALS AND METHODS
Cell culture, reagents, and plasmids.
Rabbit synovial fibroblasts (RSF) were prepared as previously described (23), and SW-1353 chondrosarcoma cells were grown to confluence in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS; Hyclone, Logan, UT), penicillin/streptomycin, and L-glutamine (Cellgro; Mediatech, Herndon, VA). At the beginning of each experiment, cells were washed 3 times with Hanks' buffered sterile saline (Cellgro; Mediatech) to remove traces of serum and placed in DMEM containing 0.2% lactalbumin hydrolysate (Gibco BRL, Grand Island, NY), with or without 10 ng/ml recombinant IL-1β (Promega, Madison, WI). The kappaB-luciferase reporter construct was graciously provided by Dr. Takashi Fujita (Institute of Medical Science, Tokyo, Japan). The 4,372-bp human MMP-1 promoter-luciferase reporter construct has been described previously (24). To create the Bcl-3 expression construct, a Bcl-3 complementary DNA (cDNA) (graciously provided by Dr. Edward Schwarz, University of Rochester, Rochester, NY) was inserted into the Eco RI site of the pBkRSV expression plasmid (Stratagene, La Jolla, CA). The NF-κB1 cDNA was isolated from SW-1353 messenger RNA (mRNA) by reverse transcription–polymerase chain reaction (RT-PCR) using the forward primer 5′-CTCGCCACCCGGCTTCAGAATG-3′ and the reverse primer 5′-TGAACCAAGAAAGGAAGCCAAG-3′. The resulting 3,093-bp product, which encompasses the entire coding sequence of NF-κB1, was ligated into the Sca I site of pBkRSV and sequenced.
Transfection of RSF and SW-1353 cells and promoter assay.
Transient transfections of RSF and SW-1353 cells were performed as described previously (14, 25). Luciferase assay of MMP-1 promoter activity was measured as described previously (15), and statistical significance was determined through the use of Student's t-test. To create stable lines, SW-1353 cells were transfected with pBkRSV or pBkRSVBcl-3 using GenePorter 2 reagent (Gene Therapy Systems, San Diego, CA). Transfectants were then pooled and passed at a dilution of 1:15 into 10% FCS–DMEM supplemented with G418 (800 μg/ml, Stratagene). Following 3 rounds of selection, cells were seeded to 6-well plates in 10% FCS–DMEM plus G418. After 24 hours, media were replaced with serum-free DMEM (without G418), with or without IL-1β (10 ng/ml). After a further 24 hours, RNA was harvested into TRIzol (Invitrogen, Carlsbad, CA) and assayed for Bcl-3 and MMP-1 expression, as described below.
Measurement of mRNA expression.
The levels of MMP-1, Bcl-3, and GAPDH mRNA were measured by RT of SW-1353 RNA, followed by quantitative PCR. Briefly, 2 μg of total RNA was DNase-treated (DNA-free; Ambion, Austin, TX), then reverse-transcribed with Moloney murine leukemia virus reverse transcriptase. One-tenth of the RT product (equivalent to 200 ng of input RNA) was assayed first by radioactive PCR, using Platinum Taq DNA polymerase (Invitrogen) and α33P-dATP. The resulting PCR products were separated on 10% nondenaturing polyacrylamide gels and visualized by autoradiography. The primers used to detect MMP-1 were 5′-GGGAGATCATCGGGACAACTC-3′ and 5′-GGGCCTGGTTGAAAAGCAT-3′, and these yielded a 72-bp product. The primers used to detect Bcl-3 were 5′-CCACAGACGGTAATGTGGTG-3′ and 5′-TATTGCTGTGGTGCAGGGTA-3′, and these yielded a 143-bp product. The primers used to detect GAPDH were 5′-TCCCCGGCGGCCCCATGCC-3′ and 5′-GCAGGTCCACGGGCCCCT-3′, and these yielded a 63-bp product. These primers were also used in real-time PCR reactions on a DNA Engine Opticon Continuous Fluorescence Detection System (MJ Research, Waltham, MA) using 2X SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA).
Standard curves were generated (CT versus log pg DNA) using plasmids containing the MMP-1, Bcl-3, and GAPDH cDNA, and these curves were then used to calculate the amount of cDNA generated from each RT reaction. The number of molecules of Bcl-3 and MMP-1 transcript was calculated and reported as the number of molecules per 1,000 molecules of GAPDH. The fidelity of the real-time PCR reactions was confirmed by the melting curves of the products, which yielded distinct Tm. For real-time PCR detection of Bcl-3 mRNA in the antisense assay, primers and probes were made by Applied Biosystems. The upper primer sequence was 5′-ACGCCGTGGAAAACAACA-3′, the lower primer sequence was 5′-GTACATTTGCGCGTTCACGT-3′, and the probe sequence was prepared as 6-FAM-CTGCAGCACGGCGCCAA.
Treatment of SW-1353 cells with antisense oligonucleotides.
Phosphorothioate-linked sense (5′-/5Me-dC/GCCATGGACGAACTGTTCCCC-3′) and antisense (5′-/5Me-dC/GGGGAACAGTTCGTCCATGGC-3′) oligonucleotides to the Bcl-3 start site were prepared by Integrated DNA Technologies (Coralville, IA). SW-1353 cells (2 × 105/well) were seeded in 10% FCS–DMEM to 6-well plates (Costar; Corning, Corning, NY). At 60–90% confluence, cells were transfected with sense or antisense oligonucleotides (5 μM final concentration) using GenePorter transfection reagent (Gene Therapy Systems) according to the manufacturer's directions. After 5 hours, media were supplemented with an equal volume of 20% FCS–DMEM. After 24 hours, cells were placed in serum-free media with or without IL-1β (10 ng/ml). Following 18 hours of culture, total RNA was harvested, and gene expression was quantified by real-time PCR as described above.
Measurement of programmed cell death by Hoechst 33342 staining.
SW-1353 cells (5 × 104/well) were seeded in 24-well plates (Becton Dickinson, Franklin Lakes, NJ), and at 60–90% confluence, cells were treated with sense or antisense oligonucleotides as described above. Following recovery, cells were placed in serum-free media with or without IL-1β (10 ng/ml) for 24 hours. Hoechst 33342 (1 μg/ml; Sigma, St. Louis, MO) was then added to each well to detect cells with condensed chromatin, a hallmark of programmed cell death. After 30 minutes of incubation at 37°C, cells were examined under a fluorescence microscope. Photographs of each condition were developed. Each photograph was overlaid with a 5 × 5–unit grid. The same 8 squares of the grid were counted for each photograph. The number of cells with condensed chromatin was calculated as a percentage of the total cell number in each square. The mean and SD percentage of apoptosis was then calculated for each condition.
Bcl-3, an IL-1β–inducible gene in chondrocytes.
Investigators in our group have recently used gene profiling to identify IL-1β–inducible early response genes that may contribute to inflammation and degradation in arthritis (26). In this study, the Clontech Atlas Human 1.2 Array II (Becton Dickinson) was used to identify novel IL-1β–responsive genes in SW-1353 chondrocytic cells (Table 1). This analysis demonstrated that mRNA for Bcl-3 was increased 6-fold by treatment of SW-1353 cells with IL-1β (10 ng/ml). To confirm this finding, we analyzed total RNA from SW-1353 cells that had been treated for 2 hours with IL-1β by RT-PCR. By performing linear PCR reactions containing α33P-dATP, we found that Bcl-3 mRNA levels substantially increased within 2 hours of IL-1β treatment (Figure 1A). This was confirmed by real-time PCR, with a 4.5-fold increase measured at the 2-hour time point (Figure 1B). Interestingly, 2-hour IL-1β stimulation resulted in comparable fold increases for Bcl-3 and MMP-1 mRNA (Figure 1B). After 8 hours and 24 hours of IL-1β treatment, the fold induction of Bcl-3 was less than that observed at 2 hours (Figure 1C). The fold induction of MMP-1 was higher at 24 hours (7.1×), suggesting that additional pathways may contribute to IL-1–induced MMP-1 gene expression at later time points. These data demonstrate that Bcl-3 is an IL-1β–inducible gene in chondrocytic cells, and that its inducibility is temporally related to IL-1β induction of MMP-1.
Table 1. Interleukin-1β (IL-1β)–responsive genes in SW-1353 chondrocytic cells*
GenBank accession no.
SW-1353 cells were stimulated for 2 hours with 10 ng/ml IL-1β, and total RNA was harvested. This RNA was then labeled by Clontech and hybridized with the Atlas Human cDNA Array 1.2kII. Genes that showed at least a 2-fold increase with IL-1β treatment are presented.
TGF-β inducible early protein (TIEG)
Monocyte chemotactic protein 3 (MCP-3) precursor; NC28; small inducible cytokine A7 (SCYA7)
Basic helix-loop-helix protein (DEC1)
B cell lymphoma 3–encoded protein (Bcl-3)
CD83 antigen precursor; cell surface protein HB15; B cell activation protein
Astrocyte glial fibrillary acidic protein (GFAP)
Dual-specificity protein phosphatase 5; dual-specificity protein phosphatase HVH3
Interferon-γ–inducible protein IFI16; interferon-inducible myeloid differentiation transcription activator
M63838; S75433; S75417
Tumor necrosis factor α–inducible protein 3 (putative DNA binding protein A20); (zinc-finger protein A20)
G protein–coupled receptor RDC1 homolog
Eyes absent homolog 3 (EYA3)
Protein-tyrosine phosphatase MEG2 (PTPase MEG2)
B cell lymphoma 6 protein (Bcl-6); zinc-finger protein 51 (ZNF51); LAZ-3 protein
MMP-1 gene expression activatable by Bcl-3.
The coordinate activation of Bcl-3 and MMP-1 by IL-1β raised the possibility that the expression of these two genes is linked in cells of the joint. To address this possibility, we created a line of SW-1353 cells stably expressing Bcl-3 under the control of the RSV constitutive promoter. Real-time RT-PCR confirmed that after 24 hours of culture in serum-free media, basal Bcl-3 mRNA levels were ∼2-fold higher in the Bcl-3–expressing line than in cells transfected with the empty expression vector (Figure 2A). When IL-1β was added to the culture media, Bcl-3 mRNA expression was elevated in vector control cells and was higher still in the Bcl-3–expressing cells (Figure 2A). Importantly, elevated constitutive expression of Bcl-3 resulted in elevated basal and IL-1β–induced MMP-1 expression in SW-1353 cells after 24 hours of culture (Figure 2B). The fold induction by IL-1 in the empty vector–expressing cells was similar to that in the Bcl-3–expressing cells (7.7-fold and 8.5-fold, respectively), suggesting that additional pathways are involved in MMP-1 induction. These data demonstrate that forced expression of Bcl-3 can enhance MMP-1 gene expression in chondrocytes.
IL-1β induction of MMP-1 reduced by inhibition of Bcl-3 gene expression.
While exogenous Bcl-3 expression can induce MMP-1, we wanted to determine whether endogenous Bcl-3 is required for MMP-1 expression in connective tissue cells. To test this, we transfected SW-1353 cells with phosphorothioate-linked antisense oligonucleotides complementary to the Bcl-3 translational start site. The antisense oligonucleotides effectively reduced IL-1β–induced Bcl-3 mRNA expression in SW-1353 cells compared with expression in control cells containing sense oligonucleotides (Figure 3A). In addition, compared with sense oligonucleotides, antisense Bcl-3 treatment reduced IL-1β–induced MMP-1 gene expression (Figure 3B). These data are the first demonstration that Bcl-3 is a required gene for IL-1β–induced MMP-1 gene expression in connective tissue cells.
Recent evidence suggests that Bcl-3 regulates apoptosis in lymphocytes (27). Therefore, we wanted to assess the effect of Bcl-3 repression on SW-1353 cell survival to ensure that the observed effects on MMP-1 gene expression were not due to increased cell death. To this end, SW-1353 cells were treated with Bcl-3 sense and antisense oligonucleotides, and apoptosis was then quantified by Hoechst staining and counting of condensed nuclei. We observed 20–25% cell death that was associated with the oligonucleotide transfection procedure (Figure 4). However, 5 μM antisense oligonucleotide, which reduces Bcl-3 and MMP-1 gene expression (Figure 3B), did not significantly (P = 0.33) increase the number of Hoechst-positive SW-1353 cells (Figure 4). These data demonstrate that Bcl-3 is not an antiapoptotic gene in SW-1353 cells. Therefore, the inhibition of MMP-1 gene expression observed in Bcl-3–repressed cells (Figure 3) is not a function of increased programmed cell death.
MMP-1 promoter activated by Bcl-3 via NF-κB pathway.
To further define the mechanism of Bcl-3–induced MMP-1 expression, we transiently cotransfected SW-1353 cells with the Bcl-3 expression plasmid along with a human MMP-1 promoter construct. Bcl-3 at 50 ng and 100 ng significantly increased expression of the MMP-1 promoter in these cells (P = 0.05 and P = 0.003, respectively) in a dose-dependent manner (Figure 5). This effect was specific to MMP-1 promoter sequences, since Bcl-3 expression did not induce the promoterless luciferase plasmid, pGL3Basic (data not shown). Similarly, Bcl-3 at 250 ng increased basal expression of the human MMP-1 promoter in primary RSF (P = 0.001) and enhanced IL-1β–induced MMP-1 transcription in these cells (P = 0.01) (Figure 6A). These data suggest that Bcl-3 is a general activator of the MMP-1 promoter in SF and chondrocytes. Bcl-3 at 250 ng also enhanced transcription from a luciferase construct under the transcriptional control of 3 NF-κB binding sites (P = 0.01), suggesting that Bcl-3 is working through an NF-κB–dependent mechanism (Figure 6B). At low concentrations of expression vector, NF-κB activity was enhanced by the addition of IL-1β, but at higher concentrations, no additional increase was observed. This suggests that high levels of Bcl-3 expression are sufficient to maximally induce the NF-κB construct. Taken together, these data demonstrate that Bcl-3 can both activate the NF-κB pathway in connective tissue cells and drive transcription of MMP-1.
Bcl-3 contributes to NF-κB–dependent transcription by cooperating with NF-κB1.
It has been established that Bcl-3 can enhance NF-κB–dependent transcription by acting as a transcriptional coactivator for NF-κB1 homodimers (22). To determine whether this cooperation contributes to transcriptional activation of the MMP-1 promoter, RSF were transfected with the human MMP-1 promoter construct along with expression constructs for NF-κB1 and Bcl-3. Transfection of RSF with 200 ng of an NF-κB1 expression construct resulted in maximal induction of the human MMP-1 promoter (Figure 7A). We then tested the effects of low doses (100 ng) of the NF-κB1 and Bcl-3 expression constructs, alone and in combination. Importantly, concurrent expression of Bcl-3 and NF-κB1 in RSF resulted in greater stimulation of the MMP-1 promoter than did either NF-κB1 or Bcl-3 alone (Figure 7B). Furthermore, IL-1–induced MMP-1 transcription was greatest in cells expressing both factors. Thus, Bcl-3 can activate MMP-1 transcription, in part through cooperation with NF-κB1.
In this study, we showed that Bcl-3 is an IL-1β–inducible gene in chondrocytes and that this IκB-related factor can contribute to IL-1β–induced expression of MMP-1 in chondrocytes and synovial cells. Bcl-3 activates MMP-1 at the transcriptional level, in part, through cooperation with NF-κB1. These findings present a novel role for Bcl-3 as a mediator of IL-1β–dependent tissue remodeling, such as that occurring in OA and RA.
Studies from several laboratories, including our own, have shown that the NF-κB pathway is a critical component of IL-1β–induced MMP-1 gene expression in connective tissue cells (12–15). Specifically, we have identified an NF-κB1–binding site in the distal MMP-1 promoter (14). Others have shown that NF-κB1 homodimers physically interact with Bcl-3 and require Bcl-3 in order to be transcriptionally active (21, 22, 28, 29). Importantly, IL-1β induction of MMP-1 gene expression in stromal cells requires de novo gene expression (30). Thus, Bcl-3 may be an early IL-1β–responsive gene that is required for maximal MMP-1 transcription. This model is supported by our findings, that exogenous expression of Bcl-3 enhances MMP-1 transcription, and that inhibition of Bcl-3 expression reduces IL-1β–induced MMP-1 gene activation.
In contrast to reports from other laboratories (27, 31, 32), we did not find that inhibition of Bcl-3 expression caused increased programmed cell death in SW-1353 cells. This may be due to the fact that we were able to see substantial MMP-1 inhibition without complete ablation of Bcl-3 gene expression (Figure 3). However, our data do not preclude the possibility that apoptosis would occur in SW-1353 cells completely devoid of Bcl-3. Perhaps the concentration of Bcl-3 required to mediate IL-1–induced MMP-1 gene expression is higher than that required to maintain cell viability.
Our data demonstrate that Bcl-3 can enhance transactivation of the MMP-1 promoter by NF-κB1. The cooperative induction of transcription was somewhat modest, suggesting that additional factors are required to activate the promoter. Indeed, we have reported previously that NF-κB1 transactivation of MMP-1 requires an intact activator protein 1 (AP-1) site (14). Therefore, the interaction of Bcl-3 and NF-κB1 may only be fully functional in the context of other IL-1–induced transcription factors, such as AP-1 family members.
The interaction between Bcl-3 and NF-κB1 may explain how IL-1β stimulation of the NF-κB pathway can have specific effects in different cell types. While the NF-κB1/RelA heterodimer complex is essential for activation of cytokine gene expression in immune cells, NF-κB1 homodimers associated with Bcl-3 may be more critical for the activation of matrix remodeling genes in SF and chondrocytes. Indeed, investigators in our group have reported previously that IL-1β stimulation of SW-1353 cells results in a profound induction of NF-κB1 gene expression, while RelA was expressed at lower levels (26). Concurrent expression of Bcl-3 and NF-κB1 in connective tissue cells would elevate the levels of NF-κB1–Bcl-3 complexes, and this might skew transcription toward genes involved in matrix degradation, such as MMP-1. In future studies, it will be important to determine whether other genes known to contribute to OA and RA pathophysiology are targets of Bcl-3.
The authors would like to acknowledge Dr. Constance Brinckerhoff for critical reading of the manuscript.