Activation of p38 MAPK is a key step in tumor necrosis factor–mediated inflammatory bone destruction




To investigate whether activation of p38 MAPK is a crucial signaling factor in inflammatory bone destruction mediated by tumor necrosis factor (TNF). Mice overexpressing TNF were treated with 2 different inhibitors of p38 MAPK, and the effect of this treatment on joint inflammation and structural damage was assessed.


Human TNF-transgenic mice received systemic treatment with 2 different p38 MAPK inhibitors (RO4399247 and AVE8677). Treatment was started at the time of symptom onset and lasted for 6 weeks. Mice were assessed for clinical signs of arthritis, bone erosion, and cartilage damage. In addition, the effect of these inhibitors on osteoclast generation in vitro and in vivo was assessed.


Both p38 MAPK inhibitors significantly reduced clinical signs of TNF-mediated arthritis. This was attributable to reducing synovial inflammation by 50% without affecting the cellular composition of the infiltrate. Synovial expression of interleukin-1 and RANKL was reduced upon p38 MAPK blockade, and activation of the molecular target MAPK-activated protein kinase 2 (MAPKAP-2) was also inhibited. Proteoglycan loss of articular cartilage was reduced by 50%, although p38 MAPK inhibition did not change matrix molecule synthesis by cultivated chondrocytes. Importantly, bone loss was almost completely prevented by p38 MAPK inhibition. The numbers of synovial osteoclasts and precursors were dramatically reduced, and both p38 MAPK inhibitors also inhibited in vitro osteoclastogenesis at micromolar concentrations and blocked activation of MAPKAP-2 as well as differentiation markers in cultured osteoclast precursors.


These results suggest the major importance of p38 MAPK for TNF-mediated inflammatory bone destruction in arthritis and suggest that inhibition of p38 MAPK might be an important tool for reducing structural damage in rheumatoid arthritis.

Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by synovial inflammation, cartilage damage, and bone erosion. The hallmark of RA is the formation of tumor-like inflammatory tissue that has a propensity to invade bone. Upon initiation of disease, the synovial membrane becomes hyperplastic due to the accumulation of fibroblasts and cells of hematopoietic origin. This alteration of the synovial membrane is followed by the destruction of neighboring structures, such as bone and articular cartilage, due to invasive properties of synovial tissue. This latter property of inflamed synovial tissue is based on a tight interplay between synovial fibroblasts, lymphocytes, and osteoclasts.

This complex series of pathologic events leading to full-blown arthritis is governed by various proinflammatory cytokines that allow communication between inflammatory cells, culminating in concerted actions such as tissue invasion. Several of these proinflammatory cytokines, such as tumor necrosis factor (TNF), interleukin-1 (IL-1), and IL-6, are of particular clinical importance, because their blockade ameliorates RA (1–3). Most importantly, the role of TNF in chronic arthritis has been convincingly demonstrated in many ways, as follows: in the setting of RA, TNF is overexpressed by synovial membrane cells (4), systemic overexpression of TNF leads to an RA-like disease in rodents (5), and the 3 TNF blockers currently approved for the treatment of RA are highly effective in reducing signs and symptoms of RA and retarding joint destruction (3, 6, 7). Moreover, TNF itself induces proinflammatory cytokines, enhances the generation of matrix metalloproteinase (MMP), and supports the formation of osteoclasts, which are essential tools for generating inflammatory bone loss (8).

The proinflammatory and destructive potential of TNF is mediated through activation of multiple intracellular signal transduction pathways. Among these molecules, p38 MAPK is considered to be one of the most important signals for TNF-mediated inflammatory responses. This supposition is based on 4 important observations: 1) p38 MAPK is strongly activated in RA synovial membrane but not in the synovial membrane of patients with degenerative joint disease (9); 2) active p38 MAPK is predominantly expressed in endothelial and lining layer cells, which represent 2 regions of significant importance when considering the necessity of transendothelial migration by blood-derived cells in the course of inflammation as well as the role of the lining layer in the formation of the destructive pannus (10); 3) synthetic blockers of p38 MAPK have potent antiinflammatory properties and inhibit experimental arthritis in rodents (11); and 4) p38 MAPK is involved in regulating the expression of many proinflammatory cytokines, including TNF (12). Thus, p38 MAPK is pivotally involved in TNF expression induced by lipopolysaccharide or IL-1 and is, therefore, thought to play a major role in inducing TNF expression during inflammatory disease in humans (13). This regulatory function of p38 MAPK on TNF expression is yet another important reason for considering inhibitors of p38 MAPK as promising future antiinflammatory drugs.

So far, however, all of the evidence suggesting p38 MAPK to be importantly involved in joint destruction is circumstantial and indirect, and it is unclear whether activation of p38 MAPK downstream of TNF is a truly relevant factor for the induction of inflammatory—and especially destructive—joint disease in vivo. To investigate this issue, we tested the capacity of 2 specific inhibitors of p38 MAPK to block arthritis in human TNF–transgenic mice. This animal model of arthritis is based on stable transgenic overexpression of TNF leading to inflammatory arthritis. The effects of the two p38 MAPK inhibitors were assessed with respect to interference with synovitis as well as cartilage damage and bone erosion. Based on the important involvement of p38 MAPK in osteoclast differentiation (14), we specifically addressed the ability of this therapeutic approach to block osteoclast differentiation and protect the structural integrity of the inflamed joints.


Animals and treatments.

Heterozygous human TNF–transgenic mice (strain Tg197; genetic background C57BL/6) have been described previously (15). In these mice, a chronic inflammatory and destructive polyarthritis develops within 4–6 weeks after birth. In the present study, a total of 24 mice were examined in 2 independent experiments, each of which comprised 12 mice. The local ethics committee approved all animal procedures. Mice were divided into 3 groups of 8 mice each and were treated according to the following protocol: group 1 received vehicle only (1% hydroxypropyl methylcellulose, 0.4% polysorbate 80, 0.9% benzyl alcohol in 0.01M sodium citrate buffer), group 2 received the p38 MAPK inhibitor RO4399247 (kindly provided by Roche Biosciences, Palo Alto, CA), and group 3 received the p38 MAPK inhibitor AVE 8677 (kindly provided by Aventis Pharmaceuticals, Bridgewater, NJ). RO4399247 (MW 414.4) and AVE8677 (MW 535.6) are pyridinyl imidazole derivatives, which potently inhibit p38 MAPK at a 50% inhibition concentration of <2 nM. Both compounds were administered by daily intraperitoneal injection at a dose of 100 mg/kg in a total volume of 100 μl. Therapy was started at the onset of symptoms (week 5) and lasted for 6 weeks.

Clinical assessment.

Clinical evaluation was performed weekly, starting 4 weeks after birth. Arthritis was evaluated in a blinded manner as described previously (16). Briefly, paw swelling was examined in all 4 paws, and a clinical score of 0–3 was assigned (0 = no swelling, 1 = mild swelling, 2 = moderate swelling, and 3 = severe swelling of the toes and ankle). In addition, grip strength was examined in each paw, using a 3-mm–diameter wire, and was scored on a scale of 0 to −4 (0 = normal grip strength, −1 = mildly reduced, −2 = moderately reduced, −3 = severely reduced, and −4 = no grip strength). Groups were matched for arthritis severity at the beginning of treatment. Mice were weighed weekly beginning at age 5 weeks. At the end of the treatment period, animals were killed by cervical dislocation, blood was withdrawn by cardiac puncture, and all paws were obtained for histologic examination.

Conventional histologic analysis and assessment of osteoclasts.

Hind paws and right knees were fixed overnight in 4.0% formalin and then decalcified in 14% EDTA (Sigma, St. Louis, MO) at 4°C (the pH was adjusted to 7.2 by adding ammonium hydroxide [Sigma]) until the bones were pliable. Serial paraffin sections (2 μm) from all paws and the right knee joint were stained with hematoxylin and eosin (H&E) for assessment of synovial inflammation and bone erosion, toluidine blue for proteoglycan loss of articular cartilage, and tartrate-resistant acid phosphatase (TRAP) for detection of osteoclasts. TRAP staining was performed as previously described, by using a leukocyte acid phosphatase staining kit (Sigma) (17).

Synovial inflammation, bone erosions, osteoclast numbers, and cartilage destruction were quantified with the use of a Zeiss Axioskop 2 microscope (Zeiss, Marburg, Germany) equipped with a digital camera and image analysis system (OsteoMeasure; OsteoMetrics, Decatur, Georgia), as described previously (18). The area of inflammation was quantified on H&E-stained sections. Total scores were calculated as the sum of the areas of inflammation in all tarsal joints of each mouse. Erosions were quantified in the same H&E-stained sections and identified by the presence of inflamed synovial tissue within the outer cortical bone surface. Osteoclasts (≥3 nuclei; TRAP positive) and osteoclast precursors (<3 nuclei; TRAP positive) were assessed in TRAP-stained serial sections. The fraction of negatively stained cartilage from the area of total cartilage was measured and calculated on toluidine blue–stained sections.

Immunohistochemical analysis.

For immunohistochemical analysis, deparaffinized, ethanol-dehydrated tissue sections were used. Rat monoclonal antibodies against surface markers of neutrophil granulocytes (dilution 1:2,000) (MCA771G, clone 7-4; Serotec, Dusseldorf, Germany), macrophages (dilution 1:300) (anti-F4/80, clone Cl:A3-1; Serotec), T lymphocytes (dilution 1:50) (anti-CD3, CD3-12; Novocastra, Newcastle, UK), and B lymphocytes (dilution 1:200) (anti-B220, clone RA3-6B2; PharMingen, San Jose, CA) were used. In addition, polyclonal antibodies against phosphorylated MAPK-activated protein kinase 2 (MAPKAP-2) (dilution 1:50; BioSource, Camarillo, CA), IL-1β (dilution 1:25; R&D Systems, Minneapolis, MN), and RANKL (dilution 1:400; clone FL-317; Santa Cruz Biotechnology, Santa Cruz, CA) were used. Sections were pretreated with proteinase K (Roche Diagnostics, Mannheim, Germany) (0.05% for 5 minutes at 37°C) for detection of neutrophils, macrophages, and T lymphocytes; protease IV (Sigma) (0.05% for 5 minutes at 37°C) for detection of MAPKAP-2 and RANKL; heat (95°C for 20 minutes) for detection of IL-1; or were left without pretreatment for staining of B lymphocytes.

All of the above-mentioned procedures were followed by rinsing and blocking of endogenous peroxidase with 0.3–3% hydrogen peroxide in phosphate buffered saline (PBS) for 10 minutes and subsequent 30-minute incubation with a biotinylated species-specific anti-IgG secondary antibody (goat anti-rabbit antibody [Santa Cruz Biotechnology] for detection of RANKL and MAPKAP-2, and goat anti-rat antibody [BD Biosciences] for detection of all other primary antibodies). Sections were incubated with the appropriate avidin–biotin–peroxidase complex (Vectastain ABC reagent [Vector Laboratories, Burlingame, CA] for detection of CD3, F4/80, MAPKAP-2, RANKL, and IL-1; StreptABComplex/HRP [Dako, Glostrup, Denmark] for detection of CD45 receptor) for another 10 minutes using 3,3-diaminobenzidine (Sigma) as chromogen, resulting in brown staining of antigen-expressing cells.

Semiquantitative scoring was performed as follows. Total cell numbers and numbers of positively stained cells were counted in at least 3 different sites of the tarsal joint pannus of each mouse. A total of 9 mice from the 3 treatment groups (n = 3 per group) were evaluated in these immunohistochemical analyses.

In vitro osteoclastogenesis assay.

Murine osteoclasts were generated as previously described (19). Briefly, spleen cells from 6-week-old wild-type C57BL/6 mice were cultured overnight in α-minimum essential medium containing 10% fetal bovine serum and 30 ng/ml macrophage colony-stimulating factor (M-CSF). After 24 hours, nonadherent cells were harvested and subjected to gradient purification. Cells (1 × 106/ml) were then seeded onto 48-well plates and supplemented with 30 ng/ml M-CSF and 50 ng/ml RANKL (both from R&D Systems). Cultures were incubated in triplicate, with a complete medium change on day 3; p38 inhibitors at different concentrations were added throughout the culture. TRAP staining for the evaluation of osteoclast differentiation was performed after 5 days of culture, using a leukocyte acid phosphatase kit (Sigma). Osteoclasts were identified by the presence of 3 or more nuclei and purple staining.

Reverse transcription–polymerase chain reaction (RT-PCR).

Six hours after restimulation, total RNA was isolated from cultivated osteoclasts that were or were not treated with the p38 inhibitors, using the RNeasy Mini Kit (Qiagen, Hilden, Germany). One microgram of total RNA was used for first-strand complementary DNA (cDNA) synthesis (Amersham Biosciences, Buckinghamshire, UK), and a fraction (1/50) of cDNA from the RT reaction was then utilized for PCR, using the following primers: for cathepsin K, 5′-GGAAGAAGACTCACCAGAAGC-3′ and 5′-GTCATATAGCCGCCTCCACAG-3′; for MMP-9, 5′-CCTGTGTGTTCCCGTTCATCT-3′ and 5′-CGCTGGAATGATCTAAGCCCA-3′; for TRAP, 5′-ACAGCCCCCACTCCCACCCT-3′ and 5′-TCAGGGTCTGGGTCTCCTTGG-3′; for calcitonin receptor, 5′-CATTCCTGTACTTGGTTGGC-3′ and 5′-AGCAATCGACAAGGAGTGAC-3′; for β-actin, 5′-TGTGATGGTGGGAATGGGTCAG-3′ and 5′-TTTGATGTCACGCACGATTTCC-3′.

Analysis of kinase activity by Western blotting.

Cultured osteoclasts were starved in serum-free Dulbecco's modified Eagle's medium for 2 hours on day 3 of culture and stimulated with 50 ng/ml of RANKL and 30 ng/ml of M-CSF, with or without prior addition of the p38 MAPK inhibitors, for 30 minutes. Supernatants were removed, and cells were washed twice in PBS and subsequently lysed in buffer containing 10 mM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 0.2% sodium deoxycholate, 1% Nonidet P40, 1 mM NaF, 2 mM Na3VO4, and protease inhibitors. Protein extracts were separated on 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels and stained with a phospho-specific antibody against MAPKAP-2 (Biosource) and antibodies against total and phospho-specific p38 MAPK (both from Cell Signaling Technology, Beverly, MA). For control purposes, an antibody against actin (Sigma) was used.

Biosynthesis of macromolecules in cultured chondrocytes.

Knee joints from both treated and untreated mice were opened after the mice were killed, and the joint surface including cartilage and subchondral bone were obtained aseptically and distributed in 24-well multiwell plates (Costar, Cambridge, MA). Thereafter, the samples were washed 3 times with PBS (Life Technologies, Gaithersburg, MD). Consecutively, explants were incubated in a chemically defined serum-free basal medium (1 ml/well) and labeled with 20 μCi/ml of 35S-sulfate (carrier-free; Amersham, Buckinghamshire, UK) for 3 hours at 37°C. After radiolabeling, the medium was discarded, and explants were washed 3 times with ice-cold buffer (10 nM EDTA, 0.1M sodium phosphate [pH 6.5]) followed by overnight digestion in 1 ml sodium phosphate washing buffer containing Proteinase K (1 mg/ml) at 80°C. Unincorporated isotope was removed by using Sephadex G-25 (PD-10 columns; Pharmacia Biotech, Piscataway, NJ) gel chromatography. Values were determined by liquid scintillation counting (Wallac 1410 liquid scintillation counter; Wallac, Turku, Finland) of aliquots from void volume fractions and normalized to DNA content. DNA content was assessed by use of bisbenzimide (Hoechst 33258; Sigma) (20).

Statistical analysis.

Data are shown as the mean ± SEM. Group mean values of histologic data were compared by paired Student's t-test. Nonparametric Wilcoxon's signed rank test was used for comparison of clinical assessments.


Slowing of clinical signs of arthritis in human TNF–transgenic mice by p38 MAPK inhibition.

To investigate the influence of p38 MAPK inhibition on clinical signs of arthritis, scores for paw swelling and grip strength were longitudinally assessed in human TNF–transgenic mice treated with 2 different p38 MAPK inhibitors (RO4399247 and AVE8677) as well as vehicle-treated controls (Figures 1A and B). Whereas paw swelling increased throughout the observation period in vehicle-treated controls, to a mean ± SEM score of 0.78 ± 0.16 at week 11, both p38 MAPK inhibitors significantly slowed the progression of paw swelling (for mice receiving RO4399247, the mean ± SEM paw swelling score at week 11 was 0.41 ± 0.11; for mice receiving AVE8677, the score was 0.27 ± 0.12) (P < 0.05 for both inhibitors versus control). There was no significant difference between the 2 inhibitors. Similar results were obtained for grip strength. In untreated mice, grip strength deteriorated (the mean ± SEM score at week 11 was −1.88 ± 0.26), whereas both RO4399247 and AVE8677 slowed the decrease in grip strength scores (the mean ± SEM grip strength scores at week 11 were −1.19 ± 0.23 and −0.78 ± 0.33, respectively; P < 0.05 for both inhibitors versus control). Hence, inhibition of p38 MAPK improved clinical signs of TNF-mediated arthritis. There was no significant difference with respect to body weight among the 3 groups at either the start or the end of treatment (Figure 1C). Also, no side effects were noted.

Figure 1.

Efficacy of p38 MAPK inhibition on the clinical course of arthritis. The clinical course of arthritis, as indicated by the paw swelling score (A), the grip strength score (B), and body weight gain (C), was assessed in human tumor necrosis factor–transgenic mice (n = 24) that were receiving treatment with vehicle (control) or 100 mg/kg of p38 MAPK inhibitor RO4399247 or p38 MAPK inhibitor AVE8677. Values are the mean ± SEM.

Role of p38 MAPK inhibition in TNF-induced synovial inflammation.

To determine the effects of p38 MAPK inhibition on synovitis, histologic sections of tarsal joints from all 3 groups of mice were quantitatively analyzed by a digital image system (Figure 2A). Untreated mice showed large areas of inflammation affecting the entire tarsus, covering a mean area of 2.23 mm2. Treatment with both p38 MAPK inhibitors led to an ∼50% reduction of the synovial inflammation (P < 0.05 for both inhibitors versus control) by reducing the area covered by inflamed synovial tissue to 1.09 ± 0.27 mm2 (RO4399247) and 1.08 ± 0.28 mm2 (AVE8677). Representative H&E-stained sections depicting the effects of p38 MAPK inhibition on synovial inflammation of human TNF–transgenic mice are shown in Figures 2B–D.

Figure 2.

Effects of p38 MAPK blockade on synovial inflammation. A, Histologic signs of synovial inflammation were quantitatively assessed in human tumor necrosis factor–transgenic mice (n = 24) that were treated with vehicle (control), p38 MAPK inhibitor RO4399247, or p38 MAPK inhibitor AVE8677. The reduction of the inflamed area was statistically significant (P < 0.05) for mice treated with p38 MAPK inhibitors versus control mice. Values are the mean and SEM. B–D, Representative hematoxylin and eosin–stained sections showing synovial inflammation in mice treated with vehicle (B), RO4399247 (C), or AVE8677 (D). (Original magnification × 100.) E, Cellular composition of inflamed synovial tissue from human tumor necrosis factor–transgenic mice after blockade of p38 MAPK. Inflamed synovial tissue was stained for neutrophils (anti–7-4), macrophages (F4/80), T cells (anti-CD3), and B cells (anti-CD45R0). The mean ± SEM percentages of positive cells within inflamed synovial tissue are shown. Arrows indicate synovial inflammation.

Next, to determine whether these compounds alter the cellular composition of inflamed synovial tissue, we stained macrophages, neutrophil granulocytes, T lymphocytes, and B lymphocytes. As shown in Figure 2E, the relative distribution among the various cell types within inflamed synovial tissue remained stable. Hence, p38 MAPK inhibitors reduced synovitis in human TNF–transgenic mice without selectively affecting the cellular influx of specific cell types or the overall cellular composition of inflamed tissue.

Role of p38 MAPK inhibition in the expression of synovial cytokines and MAPKAP-2.

Next, we sought to investigate whether p38 MAPK inhibition reduces the expression of molecules downstream of TNF, such as other proinflammatory cytokines. In addition, we questioned whether p38 MAPK inhibitors are able to affect in vivo phosphorylation of MAPKAP-2, the direct molecular target of p38 MAPK. Immunohistochemical analyses of inflamed paw joints from placebo-treated human TNF–transgenic mice showed expression of IL-1 in a mean ± SEM 12.1 ± 4.4% of cells of the synovial tissue. Treatment with the p38 MAPK inhibitors RO4399247 and AVE8677 significantly reduced the expression of IL-1 (3.3 ± 1.4% and 3.8 ± 1.0%, respectively), indicating interference with TNF-mediated IL-1 expression. Furthermore, RANKL expression was also significantly reduced by p38 MAPK inhibitor treatment (19.9 ± 3.2% for control versus 4.0 ± 2.3% and 7.7 ± 3.8%, respectively, for RO4399247 and AVE8677). Finally, analysis of MAPKAP-2 activation revealed that 16.1 ± 1.8% of cells within inflamed synovial tissue were positive for MAPKAP-2. This frequency was reduced by more than two-thirds by treatment with p38 MAPK inhibitors (for RO4399247, 5.2 ± 2.8%; for AVE8677, 3.5 ± 1.5%; P < 0.05), suggesting that, in fact, p38 MAPK inhibition affects the expression of MAPKAP-2, the direct molecular target of p38 MAPK (Figure 3).

Figure 3.

Immunohistochemical analysis of MAPK-activated protein kinase 2 (MAPKAP-2), interleukin-1 (IL-1), and RANKL expression. Paw sections from human tumor necrosis factor (TNF)–transgenic mice treated with vehicle (control), p38 MAPK inhibitor RO4399247, or p38 MAPK inhibitor AVE8677 were stained for expression of IL-1, RANKL, and phosphorylated MAPKAP-2. A–C, Mean and SEM percentages of positive cells within inflamed synovial tissue. ∗ = P < 0.05 versus control. D–F, The p38 MAPK inhibitors significantly reduced activation of MAPKAP-2 and expression of IL-1 and RANKL in synovial tissue of TNF-transgenic mice. Sections are representative, and positive cells are indicated by brown staining. (Original magnification × 1,000.)

Role of p38 MAPK blockade in inflammatory proteoglycan loss and biosynthetic capacity.

When we evaluated cartilage damage in human TNF–transgenic mice by assessing proteoglycan loss through toluidine blue staining of articular cartilage (Figures 4A–D), untreated mice showed marked proteoglycan loss affecting 20.7 ± 5.2% of articular cartilage. In contrast, both p38 MAPK inhibitors reduced proteoglycan loss by >50%, showing efficacy similar to that observed for synovial inflammation (for RO4399247, 10.0 ± 1.8%; for AVE8677, 10.1 ± 2.0% [P < 0.05 for both inhibitors versus control]). When we subsequently investigated whether the level of ex vivo biosynthesis activity of articular cartilage, as determined by matrix macromolecule synthesis, was altered in treated compared with untreated human TNF–transgenic mice (Figure 4E), we made the following observations. Although wild-type mice incorporated 12.8 ± 3.1 counts per minute/μg DNA, human TNF–transgenic mice showed significantly less proteoglycan synthesis (7.2 ± 0.4 cpm/μg DNA incorporation [P < 0.05]). However, both p38 inhibitors showed results similar to those for control human TNF–transgenic mice (7.8 ± 1.1 cpm/μg DNA and 6.6 ± 1.1 cpm/μg DNA for RO4399247 and AVE8677, respectively). Hence, both p38 MAPK inhibitors reduced cartilage degradation but did not increase macromolecule synthesis.

Figure 4.

Effects of p38 MAPK blockade on cartilage damage. A, The percentage of destained cartilage area, which indicates proteoglycan loss, relative to total cartilage area was determined in toluidine blue–stained sections from human tumor necrosis factor (TNF)–transgenic mice (n = 24) that were treated with B, vehicle (control), C, p38 MAPK inhibitor RO4399247, or D, p38 MAPK inhibitor AVE8677. Preservation of cartilage proteoglycan was statistically significant (P < 0.05) for both inhibitors versus controls. (Original magnification × 100 in BD.) E, Determination of proteoglycan synthesis of articular chondrocytes from knee joints of wild-type mice and TNF-transgenic mice treated with vehicle (control), p38 MAPK inhibitor RO4399247, or p38 MAPK inhibitor AVE8677. Incorporation of 35S-sulfate into newly synthesized matrix proteoglycans present in the cell layer was then measured. Values are the mean and SEM. ∗ = P < 0.05 versus wild-type mice.

Propensity of p38 MAPK inhibition to inhibit TNF-mediated bone destruction by reducing the generation of osteoclasts in synovial tissue.

To analyze the impact of p38 MAPK inhibition on bone erosion and synovial osteoclastogenesis, we assessed the area of bone erosions and quantified the number of synovial osteoclasts adjacent to bone erosions (Figures 5A and B). Untreated human TNF–transgenic mice showed extensive bone erosions (mean ± SEM area 0.39 ± 0.11 mm2). Treatment with both p38 MAPK inhibitors effectively reduced the occurrence of bone erosions (80% reduction with RO4399247, and 82% reduction with AVE8677; P < 0.05 for both inhibitors versus control). Consistent with this finding, untreated arthritic human TNF–transgenic mice showed an abundant accumulation of synovial osteoclasts (72 ± 27 osteoclasts/section), whereas both p38 MAPK inhibitors significantly reduced synovial osteoclastogenesis (76% reduction with RO4399247, and 74% reduction with AVE8677; P < 0.05 for both inhibitors versus control). Interestingly, p38 MAPK inhibition also interfered with early osteoclast differentiation in the synovial membrane, because the number of osteoclast precursors (as defined by TRAP-positive cells with <3 nuclei) in the synovial membrane was significantly reduced (P < 0.05) (Figure 5C). Thus, vehicle-treated human TNF–transgenic mice showed a mean ± SEM of 182 ± 67 osteoclast precursors within a paw section, whereas upon treatment with the p38 MAPK inhibitors RO4399247 and AVE8677 this number was reduced by 81% and 76%, respectively. Representative TRAP-stained paw sections depicting subchondral bone erosions as well as purple-stained osteoclasts are shown in Figures 5D–H.

Figure 5.

Effects of p38 MAPK blockade on bone erosion and synovial osteoclastogenesis. The area of bone erosion (A), the number of synovial osteoclasts (B), and the number of synovial osteoclast precursors (C) were determined in tartrate-resistant acid phosphatase (TRAP)–stained sections from tumor necrosis factor–transgenic mice (n = 24) that were treated with vehicle (control), p38 MAPK inhibitor RO4399247, or p38 MAPK inhibitor AVE8677. Reductions in the area of bone erosion, the number of synovial osteoclasts, and the number of osteoclast precursors were statistically significant (P < 0.05) for both inhibitors versus untreated control mice. Values are the mean and SEM. Representative TRAP-stained sections of mice treated with vehicle (D and E), RO4399247 (F and G), or AVE8677 (H) are shown.

Impaired osteoclast differentiation and down-regulation of osteoclast-lineage markers by p38 MAPK inhibitors in vitro.

To determine the effect of both p38 MAPK inhibitors on osteoclastogenesis in vitro, mononuclear cells derived from spleen extracts were cultured with M-CSF and RANKL. A dose-dependent reduction of osteoclast numbers was observed with increasing concentrations of both RO4399247 and AVE8677 (Figure 6A). Thus, at doses of 1 × 10−6M, both p38 MAPK inhibitors reduced osteoclast formation in the presence of M-CSF and RANKL by 70%. Along with the reduction in the number of mature osteoclasts, the number of osteoclast precursors was also significantly reduced (∼70%), suggesting that the inhibitory effect of p38 MAPK blockers on osteoclast formation also affects osteoclast differentiation (Figure 6A).

Figure 6.

Ex vivo analysis of p38 MAPK inhibition on osteoclastogenesis. Murine spleen cells were treated with 30 ng/ml of macrophage colony-stimulating factor (M-CSF) and 50 ng/ml of RANKL and various concentrations of the p38 MAPK inhibitors RO4399247 and AVE8677 for 5 days. A, Osteoclast precursors (top left) and osteoclasts (top right) were identified as purple-colored cells by tartrate-resistant acid phosphatase (TRAP) staining. Values are the mean and SEM. ∗ = P < 0.05 versus control. The lower panel shows representative TRAP stainings of osteoclast cultures treated with various concentrations of the p38 MAPK inhibitors RO4399247 and AVE8677. B, Expression of phosphorylated MAPK-activated protein kinase 2 (phospho-MAPKAP-2), phospho-p38, and total p38 MAPK in osteoclast cultures after incubation for 3 days. C, Expression pattern of osteoclast-specific genes by reverse transcription–polymerase chain reaction. CTR = calcitonin receptor; MMP-9 = matrix metalloproteinase 9.

Next, we performed immunoblot analyses of these treated and untreated cultures to assess the expression of total and phosphorylated p38 MAPK as well as phosphorylated MAPKAP-2, a downstream target of p38 MAPK. Whereas the expression of total and phosphorylated p38 MAPK was not affected by the inhibitors, they both significantly reduced activation of the downstream target MAPKAP-2 at a concentration of 10−6M (Figure 6B). To further analyze the functional changes of these cells, total RNA from these cultures was collected, and RT-PCR for the expression of the osteoclast markers calcitonin receptor (CTR), TRAP, MMP-9, and cathepsin K was performed (Figure 6C). Both inhibitors significantly reduced expression of the osteoclast-specific differentiation markers CTR and TRAP at the concentrations described above, whereas the levels of activation markers such as MMP-9 and cathepsin K did not change. These data suggest that both p38 MAPK inhibitors affect signaling downstream of p38 MAPK and thereby interfere with osteoclast differentiation.


In this study, we investigated the impact of p38 MAPK inhibition on TNF-induced inflammatory arthritis. The results of this proof-of-concept study suggest that activation of p38 MAPK is of central importance in transducing the deleterious effects of TNF on cells of the synovial membrane. We showed that inhibition of p38 MAPK by 2 different specific inhibitors not only decreased the severity of synovial inflammation and cartilage damage but also led to a dramatic protection against bone destruction, when applied in the setting of chronic arthritis induced by the overexpression of TNF. These effects were associated with reduced expression of proinflammatory cytokines, such as IL-1, in the synovial membrane, a decrease of proteoglycan loss from articular cartilage, as well as an impaired differentiation of osteoclasts in inflamed synovial tissue.

The MAPK family member p38 is well known as a major signaling molecule of inflammation. Importantly, p38 MAPK plays a major role in proinflammatory cytokine production by activating transcription factors binding to the promoter regions of many proinflammatory cytokines, including TNF and IL-1 (21). For instance, endotoxin-induced production of TNF is regulated by p38 MAPK activation (22). These observations have stimulated research on inhibitors of p38 MAPK as potential antiinflammatory drug therapy. Indeed, p38 MAPK inhibitors have been successfully tested in animal models of arthritis, such as collagen-induced arthritis, and have also passed early-phase clinical trials (23, 24). Many of these studies have supported the concept that p38 MAPK is importantly involved in the production of proinflammatory cytokines upon initiation of inflammatory arthritis. In addition, however, p38 MAPK itself is also regarded as an important signaling molecule of proinflammatory cytokines. Among 4 different isoforms of p38 MAPK, the α and β isoforms are mainly involved in signaling of cytokines, whereas the δ isoform, which is also expressed in RA synovial tissue, is activated by cytokine-independent mechanisms such as retrotransposable viral sequences termed L1 elements (25).

TNF activates the p38 MAPK pathway through the type I TNF receptor, and this downstream activation of p38 MAPK allows TNF to transduce and communicate inflammatory signals to cells of the involved organ (e.g., the synovial membrane) (26). Because signaling of type I TNF receptor is rather complex and, apart from p38 MAPK, involves other MAPK families as well as NK-κB, the impact of p38 MAPK signaling on the systemic inflammatory effects of TNF is rather poorly understood. In this model, we made use of the stable overexpression of TNF due to a transgene to investigate the impact of p38 MAPK as a downstream mediator of the proinflammatory effects of TNF. Thus, this model does not address the production of TNF by p38 MAPK but rather investigates the importance of this signaling molecule as a downstream mediator of continuous high amounts of TNF. In this model, inhibition of p38 MAPK mitigated clinical signs and all major histopathologic features of chronic arthritis, such as synovial inflammation, cartilage damage, and bone loss. This finding suggests that activation of p38 MAPK is a key signaling step in TNF-mediated pathology in the joint.

Upon inhibition of p38 MAPK, TNF-induced formation of inflamed synovial tissue was generally reduced, and no qualitative effect on specific cell populations was noted; rather, the quantity of synovial cellularity was significantly reduced. This points to a general inhibition of synovial inflammation by p38 MAPK blockade rather than specific effects on single cell types, and suggests that p38 MAPK activation is a critical factor in the severity of synovial inflammation. It is compatible with our previous finding of overexpression of p38 MAPK in endothelial cells, allowing us to speculate that p38 MAPK blockade may affect transendothelial migration (27). Furthermore, the observed reduction in the expression of downstream cytokines such as IL-1 and RANKL is also in accordance with the reduction in the inflammatory response. The efficacy of p38 MAPK inhibition was demonstrated by reduced activation of the downstream target MAPKAP-2, whereas phosphorylation of p38 MAPK itself, which is accomplished by upstream kinases, was not altered.

Apart from synovial inflammation, p38 MAPK inhibition beneficially affected cartilage damage. Proteoglycan loss from articular cartilage adjacent to inflamed synovial tissue was significantly reduced, and the efficacy of p38 MAPK inhibition was similar to that observed for synovial inflammation. The protection of articular cartilage might be an indirect effect due to lower expression of proinflammatory cytokines, especially IL-1, which is a key inducer of metalloproteinases. This notion is supported by the observation that matrix synthesis did not increase upon p38 MAPK inhibition.

A reduction in TNF-mediated bone loss was the major hallmark of p38 MAPK inhibition. Tissue invasion into juxtaarticular bone was almost completely abolished with use of p38 MAPK inhibitors. In keeping with this effect, a dramatic reduction in the number of mature osteoclasts was noted. However, this was accompanied by a significant reduction in the number of in vivo osteoclast precursor cells, suggesting that impaired osteoclastogenesis is a consequence of blocking signaling through p38 MAPK. This was further confirmed in studies of in vitro osteoclastogenesis. These observations are consistent with 2 important principles, as follows. First, activation of p38 MAPK is a key signal for osteoclastogenesis. Thus, activation of p38 MAPK is important for the differentiation of mature osteoclasts from their precursors (28); this observation is also relevant for TNF-mediated osteoclastogenesis in relation to bone loss as a result of the inflammatory process involved in arthritis. The second principle is based on the role of osteoclasts in inflammatory bone destruction. Osteoclasts invade juxtaarticular bone in close contact with synovial fibroblasts and T cells and are essential players in this bone resorption. Complete absence of osteoclasts results in blockade of bone resorption but not synovial inflammation, and osteoclast-targeted therapies strongly inhibit bone loss in arthritis (29). Thus, unraveling p38 MAPK inhibition as a potent tool to reduce synovial osteoclast formation and retard inflammatory bone damage is consistent with the above-mentioned concepts.

In summary, blockade of p38 MAPK inhibits arthritis caused by overexpression of TNF. This suggests that p38 MAPK is an important signaling molecule downstream of TNF, and that its inhibition might be a potent tool to interfere with the deleterious effects of TNF on the joint. Most strikingly, inflammatory bone loss depends on activation of p38 MAPK, because formation of osteoclasts in the inflamed synovium depends on intact activation of the p38 MAPK signaling pathway. Effective blockade of p38 MAPK might therefore be regarded as an interesting therapeutic option to protect joints from the destructive attack of inflamed synovial tissue.


We thank Birgit Türk and Ivana Mikulic for excellent technical assistance and Elisabeth Allen (Aventis Pharmaceuticals, Bridgewater, NJ) and Anthony Manning (Roche Biosciences, Palo Alto, CA) for providing us with the p38 MAPK inhibitors.