|Home | About | Journals | Submit | Contact Us | Français|
c-Jun-N-terminal kinase (JNK)-mediated cell signaling plays a critical role in metalloproteinase (MMP) expression and joint destruction in rheumatoid arthritis (RA). Gadd45β (growth arrest and DNA damage inducible gene), which is an NF-κB regulated gene, was recently identified as an endogenous negative regulator of the JNK pathway by blocking the upstream kinase MKK7. We evaluated whether low Gadd45β expression in RA enhances JNK activation and overproduction of MMPs in RA and whether Gadd45β deficiency increases arthritis severity in passive K/BxN arthritis.
Activation of the NF-κB and the JNK pathway and Gadd45β expression was analyzed in human synovium and fibroblast like synoviocytes (FLS) using quantitative PCR, immunoblotting, immunohistochemistry, electrophoretic mobility shift assay (EMSA) and luciferase reporter constructs. Gadd45β null and wild type (WT) mice were evaluated in the K/BxN serum transfer model of inflammatory arthritis and clinical signs of arthritis, osteoclast formation, and bone erosion assessed.
Gadd45β gene and protein expression were unexpectedly low in human RA synovium despite abundant NF-κB activity. Forced Gadd45β expression in human FLS attenuated TNF-induced signaling through the JNK pathway, AP-1 activation, and MMP expression. Gadd45β deficiency exacerbated K/BxN serum-induced arthritis in mice, dramatically increased signaling through the JNK pathway and MMP3 and MMP13 gene expression in joints, and increased the area of inflammation and number of osteoclasts.
Deficient Gadd45β expression in RA can contribute to activation of JNK, clinical arthritis and joint destruction. This process can be mitigated by enhancing Gadd45β expression or by inhibiting JNK or its upstream regulator MKK7.
Rheumatoid arthritis is a chronic inflammatory disease marked by synovial lining hyperplasia, infiltration of synovium with immune cells, and joint destruction (1). Matrix metalloproteinases (MMP) are highly expressed in RA and involved in the joint degradation and remodeling. c-Jun N-terminal kinase (JNK) is thought to be especially important in extracellular matrix degradation, as JNK is a key regulator of MMP gene transcription. Furthermore, this signaling pathway, including its upstream kinases MKK4 and MKK7, is activated in RA synovium and JNK inhibition suppresses MMP gene expression and joint destruction in animal models of RA (1-3).
Recently Gadd45β (growth arrest and DNA damage inducible gene) was identified as a negative regulator of JNK. The Gadd45 genes, including Gadd45α, Gadd45β and Gadd45γ encode for 18 kDa evolutionarily conserved proteins. Initially, Gadd45β, also referred to as myeloid differentiation factor 118, was identified as a primary response gene activated in murine myeloid leukemia cells by interleukin-6 during terminal differentiation (4). Today Gadd45β is known to be involved in cellular stress responses, cell cycle control and cell survival (5, 6). Gadd45β is induced by NF-κB, binds directly to the JNK activating kinase MKK7 and inhibits the catalytic function of MKK7 by blocking access to ATP (6-8). Hence, Gadd45β serves as an endogenous inhibitor of MKK7 that can blunt signaling through the JNK pathway (7, 9). This is especially relevant to RA as MKK7 is expressed and activated in rheumatoid synovium (10) and MKK7, rather than the other JNK activating kinase MKK4, is required for JNK activation in cytokine-activated synoviocytes (11).
Gadd45β deficiency has been linked to increased disease severity in murine experimental allergic encephalomyelitis (12), hepatocelluar carcinoma (13, 14), hepatic regeneration (15) and diabetes (16). Based on the MKK7 inhibitory properties of Gadd45β and the “loss of protection” associated with Gadd45β deficiency in other models, we hypothesized that dysregulation of Gadd45β expression contributes to an enhanced MKK7 and JNK responses and subsequent MMP production in RA synovium. To explore this hypothesis, we examined the expression of Gadd45β in human synovial tissue from patients with RA and osteoarthritis (OA) and found no difference in Gadd45β gene or protein expression, despite markedly higher NF-κB activation in the former. The role of Gadd45β in the pathology associated with inflammatory arthritis was also assessed using the K/BxN serum transfer model in Gadd45β-/- mice. This experiment showed that deficient Gadd45β expression leads to an enhanced JNK activity and exacerbation of the disease. These studies suggest that MKK7 and JNK play a key role in synovial inflammation and joint damage. Therapeutic strategies designed to increase Gadd45β, inhibit MKK7, or block JNK may have potential utility in rheumatoid arthritis.
Synovial tissue and FLS were obtained from patients with OA or RA, at the time of total joint replacement or synovectomy, as previously described (17). The protocol was approved by the Human Research Protection Program of the University of California, San Diego. The diagnosis of RA conformed to the 1987 revised criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (18). The samples were either processed for cell culture or snap-frozen and stored at −80°C until processed for protein or gene expression analysis.
For preparation of human FLS, synovial tissues were minced and incubated for 1.5 hours at 37°C with 0.5 mg/ml of type VIII collagenase (Sigma, St. Louis, MO) in serum-free RPMI 1640 (Mediatech, Herndon, VA). Tissues were then filtered through a nylon strainer, washed, and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) (Gemini Biosciences, Calabasas, CA), L-glutamine, penicillin, streptomycin and gentamicin in a humidified chamber containing an atmosphere of 5% CO2. After overnight culture, nonadherent cells were removed and adherent cells were trypsinized and split at a 1:3 ratio when the cells were 80–90% confluent. FLS were used from passages 5–9, when cells were a homogeneous population (<1% CD11b positive, <1% phagocytic, and <1% Fcγ receptor II positive) (19). Mouse FLS were derived from Gadd45-/- mice (15) and wild-type mouse knee and ankle joints by microdissecting synovium and enzymatically dispersing the cells as previously described (2). Mouse FLS were used from passage 4-5 at 80-90% confluency.
Rabbit antibodies: MKK7, JNK1/2, P-MKK4 (Ser257/Thr261), P-MKK7 (Ser271/Thr275) (for WB), P-JNK1/2 (Thr183/Tyr185), P-c-Jun (Ser63/Ser73) (Cell Signaling Technology, Beverly, MA); Mouse-antibodies: p65 (Santa Cruz Biotechnology), CD68 (Dako), β-actin (Sigma), Gadd45β (8), for IHC); Goat-antibodies: MKK7 (for immunoprecipition, Santa Cruz Biotechnology), Gadd45β (Santa Cruz Biotechnology) (for WB), recombinant human TNF and IL-1β (R & D Systems, Minneapolis, MN) and anisomycin (Calbiochem, San Diego, CA) were used. Glutathione S-transferase (GST)-c-Jun was a gift from Roche Bioscience, Palo Alto, CA).
Serial cryosections (5 μm) of synovial tissue from RA and OA patients were used for immunohistochemistry. The tissue was fixed with 4% formalin (10 min) and endogenous peroxidase was depleted with 0.1% H2O2. Blocking was performed for 1 hour in phosphate buffered saline (PBS) containing 5% normal goat, 2.5% horse serum and 1% human serum albumin followed by overnight incubation with antibodies against Gadd45β, NF-κB (p65) and CD68 in blocking buffer (described above) at 4°C. Mouse IgG served as a negative control. The primary antibodies were detected using biotinylated horse anti-mouse secondary antibodies (Vector Laboratories, Burlingame, CA), followed by streptavidin–horseradish peroxidase (HRP) and aminoethylcarbazole (AEC) substrate (Dako, Carpinteria, CA). Nuclei were counterstained with hematoxylin.
All animal experiments were carried out according to protocols approved by the Institutional Animal Care Committee of the University of California, San Diego. To induce passive K/BxN arthritis (20, 21), sera, pooled from arthritic adult K/BxN mice, were injected i.p. to Gadd45β-/- and wild type mice. Preliminary studies were performed to determine a protocol for submaximal dosing that would permit detection of increased disease activity (data not shown). A low dose of 75 μl was injected day 0 and day 2 was selected because it consistently caused mild arthritis. Arthritis severity was assessed using a semiquantitative clinical scoring system for each paw where 0 = normal and swelling of each of the following gave a score of 1, digits, knuckles, mid-hind paw/mid-forepaw area and ankle/wrist joint. The maximum clinical score per leg = 4 and the maximum total per mouse = 16.
Paraffin sections of hind paws were stained with H&E and TRAP (leukocyte staining kit; Sigma, St. Louis, MO). Areas of synovial inflammation and number of osteoclasts were quantified by histomorphometry using an Axioskop 2 microscope (Zeiss, Oberkochen, Germany) and OsteoMeasure Analysis software (OsteoMetrics, Decatur, GA) as previously described (22).
Using the Amaxa Human Dermal Fibroblast Nucleofector kit (NHDF-adult; Amaxa, Gaithersburg, MD) with program U-23, 1 × 106 cells were transfected with 5 μg pcDNA3.1-Gadd45β or control (empty, mock) pcDNA3.1 plasmid according to the manufacturer's protocol (Amaxa). Following transfection, FLSs were seeded in 6-well dishes and cultured in DMEM with 10% FCS at 37°C for 24 h. The cells were incubated in fresh media for 24 hours and subsequently synchronized (0.1% FCS/DMEM) for 48 h and then stimulated with TNF (50 ng/ml) or anisomycin (1 μg/ml) for 15 minutes for studies of phosphorylation of MKK4, MKK7 and JNK, or with TNF (50 ng/ml) for 60 min for EMSA, 7 h for luciferase reporter assays and 24 hours for gene expression analysis.
Protein was extracted from human synovial tissue, human FLS and mouse ankle joints using lysis buffer (150 mM NaCl, 50mM Tris 0.5% Triton X-100, 3% SDS, 1 mM EDTA, protease inhibitor cocktail (Sigma), phosphatase inhibitor cocktail I and II (Sigma)) and sonication. The homogenates were centrifuged at 14,000 rpm for 15 min and the supernatant fractionated by NuPAGE 4-12% Bis-Tris gel electrophoresis (Invitorgen) and then transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). After blocking nonspecific binding sites with 5% nonfat milk in 0.1% Tween 20/Tris buffered saline (TBS-T) for 1 hour at room temperature, the membranes were incubated with antibodies in 5% bovine serum albumin in TBS-T overnight at 4°C. After washing the membranes with TBS-T, the antibody-protein complexes were probed with appropriate secondary HRP-conjugated secondary antibodies in 5% nonfat milk TBS-T for 1 hour at room temperature. The immunoreactive proteins were detected with chemiluminescent reagents (Pico and Femto SuperSignal, Pierce, Rockford, IL, USA). The nitrocellulose membranes were stripped with Re-Blot Western blot recycling kit (Chemicon, Temecula, CA, USA) and re-blotted with different antibodies. Densitometry analysis was done by ImageQuant (Molecular Dynamics, Inc, Sunnyvale, CA) (23) and immuno-positive bands normalized relative to β-actin or GAPDH.
Kinase assays were performed according to previously described methods (11), with modifications. Following transfection 3 × 106 FLSs were seeded in 10-cm dishes and cultured in DMEM with 10% FCS at 37°C for 24 h. The cells were incubated in fresh media for 24 hours, subsequently synchronized (0.1% FCS/DMEM) for 48 h and then stimulated with TNF (50 ng/ml) with PBS or TNF (50 ng/ml) for 15 minutes. Cells were washed with cold PBS, scraped directly into lysis buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 25 mM MgCl2, 1 mM EDTA, 10% glycerol, 1% Triton X-100, 20 mM β-glycerophosphate, 10 mM sodium fluoride, 1 mM Na3VO4, protease inhibitor cocktail (Complete, Mini, Roche Applied Sciences, Indianapolis, IN) 1 mM dithiothreitol and 5 mM PNPP. After 30 minutes incubation on ice the homogenate was centrifuged at 14,000g and the supernatant retained for immunoprecipitation. Samples containing the same amount of protein were incubated with specific goat anti-MKK-4, anti-MKK-7, or control IgG, for 3 hours at 4°C on a rotating wheel after which protein A-Sepharose CL-4B (Oncogene Research Products, Cambridge, MA) was added and incubation continued over night. After centrifugation the immune complexes were washed with lysis buffer and then with kinase buffer (25 mM HEPES, pH 7.4, 25 mM MgCl2, 20 mM β-glycerophosphate, 0.1 mM Na3VO4, protease inhibitor cocktail (Roche Applied Sciences), 2 mM dithiothreitol, 10 mM PNPP, 20 μM ATP). The kinase reaction was started by adding 30 μl of the kinase buffer with GST–c-Jun as substrate at 8 μg per reaction, 2 μCi of 32P-ATP) and then incubating for 30 minutes at 37°C. Samples were heated for 5 minutes at 95°C, separated on NuPAGE 4-12% Bis-Tris gels (Invitorgen) and visualized by autoradiography.
Following transfection, FLSs were seeded in 10 cm dishes and cultured in DMEM with 10% FCS at 37°C for 24 h. The cells were incubated in fresh media for 24 hours and subsequently synchronized (0.1% FCS/DMEM) for 48 h (NF-κB) or 72 h (AP-1) and then stimulated with TNF (50 ng/ml) for 60 minutes. Nuclear extracts were isolated using a nuclear protein extraction kit (Chemicon International, Temecula, CA) according to the manufacturers manual. Nuclear extracts (5 μg) were incubated with [γ-32P]-ATP labeled or unlabeled AP-1 (5′-CGCTTGATGAGTCAGCCGGAA-3′), NF-κB (5′-AGTTGAGGGGACTTTCCCAGGC-3′) oligonucleotides (Promega, Madison, WI, USA) for 30 minutes at room temperature and resolved on 6% DNA retentation gels (Invitrogen). The DNA binding was visualized by autoradiography and bands quantified using ImageQuant (Molecular Dynamics Inc.)
The Amaxa system program U-23 was used also for transfection of luciferase reporter constructs. FLS (5 × 105 cells) were transfected with 2.8 μg of Gadd45β or control (empty, mock) pcDNA3.1 plasmid, 2 μg AP-1-Luc or NF-κB-Luc and 0.2 μg SV-40-Luc (Renilla) reporter control vector (generously provided by Dr. Michael David, San Diego, CA) Following transfection, FLS were seeded in 12-well dishes and cultured in DMEM with 10% FCS at 37°C for 24 h. The cells were incubated in fresh media for 24 hours and subsequently synchronized (0.1% FCS/DMEM) for 48 h and then stimulated with TNF (50 ng/ml) for 7 hours. Luciferase activity was measured in cell lysates using a luminometer (MGM Instruments, Inc, Hamden, CT) and the Dual Luciferase Reporter Assay System according to the manufacturers manual (Promega). Relative AP-1 and NF-κB activity was determined by normalizaiton to SV-40 Renilla-Luc activity.
Messenger RNA in frozen mouse and human tissues and cultured FLS was isolated using RNA Stat (Tel-Test, Friendswood, TX), as described previously (24). Complementary DNA was prepared and quantitative real-time PCR performed with TaqMan Gene Expression Assays (both according to the manufacturer's instructions, Applied Biosystems, Foster City, CA) to determine relative mRNA levels, using the GeneAmp 7300 Sequence Detection system (Applied Biosystems). Pre-developed specific primes were used to detect Gadd45β (Assay ID Hs01063373) MMP3 (Assay ID Hs00233962, Mm00440295), MMP13 (Assay ID Mm00439490), and COX-2 (Assay ID Hs00153133, all from Applied Biosystems). Sample threshold cycle (Ct) values were used to calculate the number of cell equivalents in the test samples. The data were then normalized to GAPDH expression (human samples) (catalog no. 402869; Applied Biosystems) and HPRT1 (mouse samples) (Assay ID, Mm00446968, Applied Biosystems) to obtain relative cell equivalents and expressed as relative expression units (REU).
Data are shown as mean ± SEM. Differences were assessed by one-way and two-way repeated measures of ANOVA followed by Bonferroni post-hoc test for multiple groups, or a student t-test for two groups, with a criterion of p < 0.05 for significance.
Gadd45β is expressed at very low levels in most tissues but is induced by NF-κB (4, 6). Therefore, we expected Gadd45β to be high in RA relative to OA due to the marked activation of NK-kB. To study the link between NF-κB activation, Gadd45β expression and JNK activity in human synovial tissue, we first assessed NF-κB DNA-binding by EMSA in synovia from RA and OA patients. In agreement with previous work (3) we found higher NF-κB DNA binding in the rheumatoid synovium as compared with OA (4.7±0.6 vs. 1.4±0.3 relative expression units (REU), p=0.03, n=7 each) (Fig. 1A). Expression of an NF-κB regulated gene such COX-2 was elevated in the RA group as compared to the OA group (17.5±6.3 vs. 3.2±1.0 REU, p=0.04, n=7 each) (Fig 1B). These data confirm higher NF-κB activity and NF-κB regulated genes in RA compared with OA.
Surprisingly, when Gadd45β gene and protein expression was assessed in the same synovial tissues we found that Gadd45β levels were similar in RA and OA (p>0.05 for Gadd45β expression, n=7 each) (Fig. 1 C, D). The cellular distribution of Gadd45β in synovial tissue was then determined in serial sections from RA and OA synovia by immunohistochemistry. Gadd45β was expressed predominantly in intimal lining (Fig. 1E, G, I) and was mainly present in CD68+ macrophages rather than intimal lining fibroblast-like synoviocytes (Fig. 1I, J). The Gadd45β detected in the synovial sublining was mainly present in macrophages rather than T cells. Gadd45β and NF-κB (p65) immunoreactivity showed a similar pattern with no apparent difference in Gadd45β immunoreactivity between the OA and the RA group (Fig 1E, G), despite stronger immunoreactivity for NF-κB (p65) (Fig 1F, H).
Discoordinate expression of Gadd45β and NF-κB in RA synovium and relatively low expression in synovial lining fibroblast-like cells led us to explore the expression and regulation of Gadd45β in cultured FLS. To determine if RA FLS express Gadd45β in response to cytokines, the cells were stimulated with TNF (50 ng/ml) or IL-1β (2 ng/ml) and Gadd45β gene expression examined by quantitative PCR. TNF or IL-1β stimulation of RA FLS under conditions that optimize NF-κB activation (see below and (25)) induced only low level, transient expression of Gadd45β (Fig. 2A, B). The cells were also stimulated with anisomycin, a potent and non-selective activator of kinases, including the MKK4/MKK7/JNK pathway. Hence, anisomycin served as a positive control for comparison with cytokine receptor mediated JNK signaling. Anisomycin (1μg/ml) evoked massive and persistent Gadd45β expression (Fig. 2C) in human FLS, indicating that synoviocytes are able to express abundant Gadd45β under some circumstances. Similar results were observed in OA FLS (data not shown) and in HS68 cells, a human dermal fibroblast cell line (Fig. 2D, E), suggesting that this is a general property of cultured FLS as well as fibroblast lineage cells.
To assess the effect of Gadd45β expression on JNK mediated signaling, we used human RA FLS transfected with Gadd45β cDNA containing pcDNA3.1 plasmid or empty pcDNA3.1 control plasmid. The effect of forced Gadd45β protein expression on TNF-induced MKK4 and MKK7 phosphorylation and activity was examined. Using Western blotting, phosphorylation of MKK4 and MKK7 was readily detected in FLS stimulated with TNF (15 min, 50 ng/ml), with no difference observed between Gadd45β cDNA and empty pcDNA3.1 transfected cells (p>0.05, n=5) (Fig 3A, B). As Gadd45β is thought to inactivate MKK7 through binding within the kinase ATP pocket, rather than prevention of phosphorylation (7), MKK7 and MKK4 functional activity was measured using an in vitro protein kinase activity assay with GST-c-Jun as a substrate. JNK forms a tight complex with c-Jun and may therefore serve as a substrate in MKK4/7 kinase activity assays (26). Figure 3C shows that Gadd45β forced expression blocked TNF-induced MKK7 function without affecting MKK4.
After observing that forced Gadd45β expression in FLS attenuates MKK7 activity, we assessed the effect of Gadd45β on JNK phosphorylation. RA FLS transfected with Gadd45β or control plasmid on stimulated with TNF (50 ng/ml, 15 min). Gadd45β over expression significantly reduced JNK phosphorylation (0.31±0.05, p<0.001, n=7) compared with TNF-stimulated control (Fig. 3D). JNK-phosphorylation was normal in Gadd45β transfected cells that had been stimulated with anisomycin (1.26±0.14 fold of TNF-stimulated control, p>0.05 compared with control, n=5) (Fig 3E). This finding is in line with previous work showing that cytokine-induced JNK phosphorylation is strictly MKK7 dependent, while anisomycin-induced JNK activation can utilize either MKK4 or MKK7 (11). Thus, Gadd45β modulation of MKK7/JNK signaling suppresses cytokine-mediated FLS activation while leaving other stress responses through MKK4 intact.
The consequences of increased Gadd45β expression for JNK function was then assessed by examining AP-1 ad NF-κB transcriptional activity. Cultured RA FLS were co-transfected with Gadd45β or mock pcDNA3.1 plasmid and AP-1 or NF-κB promoter luciferase construct. TNF (50 ng/ml, 7 h) stimulation increased AP-1 and NF-κB transcriptional activity in pcDNA3.1-transfected cultured FLS. Gadd45β over-expression blocked AP-1 (2.8±0.3 vs. 0.9±3 fold change, p=0.004, n=4), but not NF-κB promoter activity (7.3±1.7 vs. 5.5±0.6 fold change, p=0.34, n=4) (Fig. 3F, G). Cultured RA FLS were also transfected with Gadd45β or mock pcDNA3.1 plasmid, stimulated with TNF (50 ng/ml, 1 h) and subjected to EMSA to assess AP-1 and NF-κB DNA binding. In agreement with the luciferase reporter construct assays described above, Gadd45β over-expression attenuated AP-1 DNA binding, but had no effect on NF-κB DNA binding (data not shown).
The MMP3 gene contains key AP-1 binding sites in its promoter. Thus, JNK-activated AP-1 is required for MMP gene transcription and increased Gadd45β expression should suppress this pathway. As shown in Figure 3H, TNF-stimulated MMP3 mRNA expression was reduced in cells transfected with the Gadd45β plasmid compared with control FLS (50 ng/ml of TNF for 24 h) (p<0.0001, n=6). Previous studies have demonstrated that MMP3 mRNA expression regulated by MKK7 and JNK in cultured FLS closely parallel protein levels in culture superntatants (11).
Initial studies were performed to determine the time course of Gadd45β expression is the passive K/BxN arthritis model (20, 21). Serum (150 μl, day 0) was injected i.p. and ankle joints harvested day 0, 1, 4, 8 and 12. Quantitative PCR showed that articular Gadd45β gene expression in naïve WT mice, can be detected, as with RA and OA synovium. The levels increased modestly after serum injection and were similar in magnitude to the changes observed in cytokine-stimulated FLS (Fig 4). The regulatory role of Gadd45β in the model was then evaluated using Gadd45β-/- mice. To permit detection of disease exacerbation, a modified protocol with a sub-maximal dose of K/BxN serum was used (see Material and Methods). Mild arthritis was observed in the wild type (WT) group, but disease severity was significantly greater in the Gadd45β-/- group (p<0.01 by 2-way ANOVA) (see Fig. 5A for a representative experiment).
In line with the increased clinical score in the Gadd45β-/- group, histologic evaluation of the joints at the peak of KB/xN serum-induced arthritis (day 6) demonstrated significantly greater synovial inflammation in Gadd45β-/- mice compared with WT (WT = 0.15 ± 0.05 and Gadd45β-/- = 0.47±0.09; p<0.002, n=10, Fig. 5B). Because NF-κB and JNK participate in osteoclast differentiation and bone erosion (27, 28), we assessed the number of osteoclasts in the mouse joints. Using TRAP staining, we found that the number of osteoclasts was greater in Gadd45β-/- mice compared with the WT group (WT = 2.10±0.46 and Gadd45β-/- = 5.00±1.09; p<0.05, n=10, Fig. 5C-E).
In separate experiments, joints were collected at the peak of KB/xN serum-induced arthritis (day 6) and evaluated by Western blot analysis to determine the phosphorylation state of JNK. While only a modest increase of phospho-JNK was observed in the WT group, a marked increase in phospho-JNK was observed in the Gadd45β-/- mice (16.7±0.4 fold increase vs. 3.6±0.6, n=7, p<0.05) (Fig. 6A, B).
The relationship of elevated JNK activity in the joints and MMP expression was then evaluated by quantitative PCR. On day 6 after serum administration, MMP3 and MMP13 gene expression was significantly greater in Gadd45β-/- mice compared with WT mice (MMP3: 3.6±0.8 vs. 1.3±0.3 REU, respectively, p=0.02, n=7; MMP13: 4.4±0.6 vs. 2.5±0.3 REU, respectively p=0.03, n=7) (Fig. 6C). These findings link Gadd45β not only to a protective role in the progression of inflammatory arthritis, but also to critical MKK7 and JNK regulatory mechanisms. Fibroblast-like synoviocytes are thought to be the primary source of MMPs in inflammatory arthritis, (29). Therefore, we determined if MMP gene expression is altered in Gadd45β-deficient murine FLS. As shown in Figure 6D, Gadd45β-/- FLS expressed significantly more MMP3 and MMP13 mRNA than WT mouse FLS after TNF stimulation (MMP3, 179±11 vs 82±4 REU p<0.001; MMP13, 39±4 vs 19±5 REU p<0.05, n=3).
Rheumatoid arthritis is a chronic autoimmune disease marked by synovial inflammation and joint destruction (1). Degradation of articular extracellular matrix in the RA joint is, in part, mediated by MMPs (29, 30). The production of these enzymes is regulated by cytokines such as TNF and interleukin-1, most notably through activation of mitogen activated protein kinases (MAPK). Activation of a key MAPK, namely JNK, is especially important in this process, as it phosphorylates transcription factors, such as AP-1, that are required for MMP transcription (2, 31). JNK activity is regulated by two upstream kinases, MKK4 and MKK7. Of these two kinases, MKK7 is particular important in RA, as only MKK7 is required for JNK activation in FLS after cytokine stimulation (10, 11). The kinases in the JNK pathway are highly activated in RA synovium and contribute to the production of cytokines and MMPs. While this is usually consider a response to pro-inflammatory cytokines, we considered whether Gadd45β deficiency might contribute to over-activation of JNK and rheumatoid synovitis.
Our initial studies showed that the Gadd45β expression in RA synovium is similar to OA synovium despite the activation of NF-κB in the intimal lining. In contrast, other NF-κB driven genes are readily detected in higher amounts in RA than in OA samples, which indicate that the NF-κB activity is sufficient for other NF-κB regulated genes in the synovial environment. Similarly, cultured synoviocytes had only minimal and transient Gadd45β induction despite stimulation with cytokines that maximally induce NF-κB translocation. The mechanism for this deficient transcriptional response to NF-κB activation has not been defined yet. However, it is stimulus-specific because anisomycin induces abundant Gadd45β expression. It appears to be a common feature of fibroblasts rather than disease specific because OA FLS and HS68 dermal fibroblasts also had limited TNF-induced Gadd45β gene expression as compared to stimulation with anisomycin. Therefore, the relative lack of Gadd45β is not specifically associated with RA. Pre-programmed deficient Gadd45β responses in the intimal lining would not cause RA but could amplify the innate immune processes that lead to joint damage.
Uncoupled NF-κB activation and Gadd45β in the presence of pro-inflammatory cytokines is a potential mechanism that explains why the JNK pathway is highly responsive in a inflammatory disease state like RA. Thus, MKK7 could be phosphorylated but would not be restrained by Gadd45β. Because MKK7 is the pivotal upstream kinase that regulates JNK in FLS, it could readily phosphorylate JNK and increase MMP production. The end result would be enhanced extracellular matrix destruction (31). This hypothesis is supported by our studies demonstrating that Gadd45β over-expression in cultured FLS reverses this process by suppressing AP-1 binding, AP-1-mediated transcription and MMP expression. Other negative regulators of JNK activity could also be important and contribute to the regulation cytokines and MMPs. The relative contribution of one such factor, x-linked inhibitory of apoptosis protein (xIAP), has not been studied in relation to Gadd45β, but it has been associated with deficient synoviocyte apoptosis in RA (32).
Injection of K/BxN serum increased synovial Gadd45β expression transiently, 2 to 3-fold, which was similar to that observed in TNF-stimulated cultured FLS. Direct comparisons between RA and normal tissue with respect to Gadd45β, and how this correlates with the mouse model, are difficult because matched normal human samples are rarely available. Thus, the passive model needs to be interpreted with some caution. However, the experiments allowed us to evaluate the role of Gadd45β in inflammation and the consequences of a deficient Gadd45β response. A sub-maximal dose of serum Gadd45β-/- mice led to rapid onset of arthritis, higher clinical scores, an increased inflammation and increased numbers of osteoclasts in the joints compared with WT mice. However, it is not known if the differences in osteoclast differentiation in vivo are due to an inherent property of Gadd45β-/- precursors or a reflection of an altered cytokine milieu. These findings demonstrate that the net effect of Gadd45β deficiency is disease exacerbation. Gadd45β deficiency was accompanied by increased JNK phosphorylation and elevated MMP3 and MMP13 gene expression in the joints of Gadd45-/- mice, as compared to the arthritic controls. Of note, the K/BxN serum transfer model is independent of the adaptive immune responses (33) and thus, our studies are the first demonstration that Gadd45β influences diseases that are strictly dependent on innate immunity.
The expression and role of Gadd45β in arthritis was not obvious from previous studies. For instance, a previous report suggested that Gadd45β expression in synovial fluid T cells in RA might suppress apoptosis (34). In that case, deficient Gadd45β could enhance lymphocyte death and ameliorate disease. However, this distribution of Gadd45β was not observed in our studies of synovial tissue, where expression was localized to lining synoviocytes and sublining macrophages rather than T cells. While we have focused on the role of Gadd45β in FLS in this study, we have also begun to assess the role of Gadd45β in murine macrophages. Preliminary studies showed that MMP13 gene expression in Gadd45β deficient peritoneal macrophages is higher than WT cells in response to TNF stimulation (data not shown). Hence, Gadd45β might be an important factor in synovial FLS and macrophages and warrants further studies to examine how Gadd45β may regulate JNK activity in RA synovial macrophages in situ.
In contrast to FLS, chondrocytes from normal individuals and patients with early OA (35) constitutively express Gadd45β mRNA. The role of Gadd45β in chondrocytes is complex and age dependent. At the embryonic stage, Gadd45β is critical for terminal differentiation of mouse chondrocytes in the growth plate. However, at this stage Gadd45β activates, rather than inhibits, JNK through interactions with MTK-1/MEKK4, upstream activators of JNK (36). As a consequence, Gadd45β-/- mice embryos display a decreased MMP13 gene expression, which leads to a defective mineralization and decreased bone growth. In adult cartilage, chondrocytes are quiescent, and cell survival is essential for maintenance of the avascular tissue. In late stage OA, a reduction in chondrocytes Gadd45β expression is associated with increased TNF-induced cell death (35, 36). Hence, endogenous Gadd45β might provide a protective function in adult chondrocytes by promoting cell survival. On the other hand, Gadd45β decreases chondrocyte collagen production, which could have a negative effect on cartilage composition and integrity.
Gadd45β might play an important role in many other diseases by mediating a delicate balance between different cell survival and cell death pathways. For example, inadequate expression of Gadd45β is thought to be a key feature in insulin-secreting beta cells undergoing apoptosis in response to interleukin-1β (16). In contrast, down-regulation of Gadd45β has been associated with cell survival as it is strongly correlated with the degree of malignancy in hepatocellular carcinoma (13, 14). Gadd45β is also thought to play an especially important role in adaptive immunity (12). Liu et al showed that Gadd45β limits the proliferation of CD4+ T cells in response to T cell receptor signaling and cytokines. T cells lacking Gadd45β proliferate faster than wild-type controls and are more resistant to activation-induced cell death (12). In accordance with our study, deletion of Gadd45β exacerbates murine experimental allergic encephalomyelitis.
While Gadd45β is a complex molecule that can potentially affect many cellular functions, its role in the JNK pathway is especially noteworthy in arthritis. Previous studies implicated JNK in RA (34), and recent in vitro studies with cultured FLS suggest that MKK7 is the dominant upstream kinase for cytokine-mediated processes (10, 11). However, MKK4 does play a role in some circumstances, such as TLR3-mediated signaling or anisomycin stimulation (37). One potential therapeutic approach in RA is to block a subset of relevant JNK functions, which might be safer than inhibiting all JNK activity. This could be accomplished by suppressing MKK7 while leaving MKK4 intact. Gadd45β accomplishes this goal by binding to MKK7 and preventing it from activating JNK (8).
While the role of JNK in RA has been well documented, our data in Gadd45β null mice suggest that MKK7 is a pivotal kinase for synovial JNK regulation. The beneficial effects of Gadd45β provide support for the a therapeutic strategy that targets the JNK pathway for diseases involving innate immunity, either by directly inhibiting JNK or MKK7 or by enhancing Gadd45β expression. This approach could potentially suppress osteoclast development, expression of MMPs, and synovial inflammation in RA.
This work was supported by the National Institutes of Health AR047825 (GSF), R21 DA021654 (CIS), NIH R01 CA084040 (GF), NIH R01 CA098583 (GF), Cancer Research UK programme grant C26587/A8839 (GF), the Arthritis Foundation (CIS) and the Interdisciplinary Center for Clinic Research Erlangen (project C6, GS).