Inhibitors of p38 demonstrate limited benefit in rheumatoid arthritis (RA), perhaps due to the antiinflammatory functions of p38α. This study was performed to determine if selective deletion of p38α in macrophages affects the severity of arthritis and whether blocking upstream kinases in the p38 pathway, such as MKK-3 or MKK-6, avoids some of the limitations of p38 blockade.
Wild-type (WT) mice and mice with selective deletion of p38α in macrophages (p38αΔLysM) were injected with K/BxN sera. Antigen-induced arthritis was also induced in p38αΔLysM mice. Mouse joint extracts were evaluated by enzyme-linked immunosorbent assay, quantitative polymerase chain reaction (qPCR), and Western blot analysis. Bone marrow–derived macrophages (BMMs) were stimulated with lipopolysaccharide (LPS) and were evaluated by qPCR and Western blotting. Bone marrow chimeras were generated using MKK-3−/− and MKK-6−/− mice, and K/BxN serum was administered to induce arthritis.
Compared to WT mice, p38αΔLysM mice had increased disease severity and delayed resolution of arthritis, which correlated with higher synovial inflammatory mediator expression and ERK phosphorylation. In contrast to WT BMMs cultured in the presence of a p38α/β inhibitor, LPS-stimulated MKK-6– and MKK-3–deficient BMMs had suppressed LPS-mediated interleukin-6 (IL-6) expression but had normal IL-10 production, dual-specificity phosphatase 1 expression, and MAPK phosphorylation. WT chimeric mice with MKK-6– and MKK-3–deficient bone marrow had markedly decreased passive K/BxN arthritis severity.
Inhibiting p38α in a disease that is dominated by macrophage cytokines, such as RA, could paradoxically suppress antiinflammatory functions and interfere with clinical efficacy. Targeting an upstream kinase that regulates p38 could be more effective by suppressing proinflammatory cytokines while preventing decreased IL-10 expression and increased MAPK activation.
Rheumatoid arthritis (RA) is a chronic inflammatory disease marked by synovial hyperplasia and invasion into cartilage and bone (1). This process is mediated, in part, by cytokines such as interleukin-1 (IL-1) and tumor necrosis factor (TNF) that activate a broad array of cell signaling mechanisms, leading to the release of destructive proteases. The MAPK family regulates cytokines and matrix metalloproteinase (MMP) production that perpetuate inflammation and tissue injury in RA (2). Several MAPK family members, including p38, JNK, and ERK, are expressed in the rheumatoid synovium and have been implicated in the pathogenesis of RA (3). Of these signaling pathways, p38 MAPK is especially relevant to human inflammatory disease and is activated in the rheumatoid synovium. Phospho-p38α is localized to the RA synovial intimal lining, which includes fibroblast-like synoviocytes (FLS) and monocytes that produce IL-6 and a variety of other proinflammatory mediators (4). The p38 inhibitors are effective in many animal models of arthritis and decrease TNF production by cultured synovial tissue cells, providing ample rationale for testing these compounds in RA (5–7).
Despite data supporting the use of p38 inhibitors in RA, these compounds demonstrate modest or no efficacy (8–10). The reasons for the limited benefit are uncertain, and alternative approaches that target this pathway are needed (5–7, 11). Recent data from studies of acute skin inflammation suggested that p38α exhibits antiinflammatory activity, indicating that traditional inhibitors could paradoxically increase synovitis (12, 13). In the present study, we expand on this concept and show that chronic arthritis severity is significantly increased in mice with selective p38α deficiency in macrophages. The absence of p38α in macrophages leads to suppressed dual-specificity phosphatase 1 (DUSP-1) expression, increased activation of other MAPKs, such as ERK and JNK, and decreased expression of the antiinflammatory cytokine IL-10. We also show that targeting upstream kinases that regulate p38, namely, MKK-3 and MKK-6, circumvents some of these issues presented by p38 blockade. Therefore, targeting p38α in a macrophage-dominant disease such as RA might have limited benefit, while targeting upstream kinases may have antiinflammatory effects and avoid the unanticipated proinflammatory effects of p38 blockade.
MATERIALS AND METHODS
KRN T cell receptor–transgenic mice were a gift from Drs. D. Mathis and C. Benoist (Harvard Medical School, Boston, MA) and Institut de Génétique et de Biologie Moléculaire et Cellulaire (Strasbourg, France) (14). Mice with loxP-flanked Mapk14 alleles and mice expressing Cre under control of the lysozyme M (LysM) have been described previously (15, 16). LysM-Cre mice were purchased from The Jackson Laboratory. The p38αΔLysM mice were generated by crossing p38αF/F and LysM-Cre mice. MKK-3– and MKK-6–deficient mice have been described previously (17, 18). Mice were bred on the C57BL/6 background. Mice were 8–12 weeks old at the time of the experiments. The mice were bred and maintained under standard conditions in the UC San Diego Animal Facility, which is accredited by the American Association for Accreditation of Laboratory Animal Care. All animal protocols were approved by the institutional review board prior to the beginning of the study.
Lipopolysaccharide (LPS) and methylated bovine serum albumin (mBSA) were obtained from Sigma. The p38α/β kinase inhibitor SB203580 was purchased from Calbiochem.
Serum transfer and arthritis scoring.
Sera from arthritic adult K/BxN mice were pooled, and recipient mice were injected intraperitoneally with 150 μl of K/BxN serum on day 0. In the model of chronic disease, mice were injected with 150 μl of K/BxN serum on day 0, followed by 100 μl per week. Clinical arthritis scores were determined as previously described (19). Bone marrow chimeras were generated as previously described (20). Adult mice were irradiated with 5.6 Gy twice, 2 hours apart, using a Mark 1 Cs-137 irradiator (J. L. Shepherd and Associates). Twenty-four hours later, bone marrow cells were flushed from the femurs and tibias of donor mice with 10K media (RPMI 1640 with L-glutamine [2.05 mM], fetal bovine serum [FBS; 10%], penicillin [100 units/ml], streptomycin [100 μg/ml], and 2-mercaptoethanol [0.05 mM]). The red blood cells were lysed, and the remaining bone marrow cells were washed twice in sterile phosphate buffered saline (PBS). The cells were counted, and the irradiated recipient mice were injected intravenously with 5 × 106 cells in sterile PBS. The recipient mice were maintained on trimethoprim/sulfamethoxazole-supplemented water for 2 weeks postirradiation. Confirmation of engraftment was performed 8 weeks after bone marrow transplantation by quantitative polymerase chain reaction (qPCR) analysis of peripheral blood samples (21). Mice were considered chimeric if they had ≥95% donor cell genotype. The K/BxN passive transfer model was induced in chimeric mice 8 weeks after bone marrow transplantation, by intraperitoneal injection of 100 μl of K/BxN serum on day 0 and day 2. Mice were sacrificed on day 12 of the model. The experiment was considered successful if the reconstituted wild-type (WT), MKK-3−/−, and MKK-6−/− control groups performed the same as nonirradiated animals with serum-induced arthritis, as previously described (22, 23).
Antigen-induced arthritis (AIA).
Experimental AIA was induced by subcutaneous injection of 100 μg of mBSA emulsified in 100 μl of Freund's complete adjuvant (CFA) into the flank followed 1 week later by intradermal injection of 100 μg of mBSA/CFA into the tail base. Two weeks after these injections, arthritis was induced by intraarticular injection of 60 μg of mBSA in 10 μl of saline into the right knee joint. The left knee was injected with PBS to serve as a control. Ten days after intraarticular injection, disease was assessed by histologic analysis as described below.
Histologic analysis and cytokine protein analysis.
Mouse joints were fixed in 10% formalin, decalcified in 10% EDTA for 2–3 weeks, and paraffin embedded. Sections were prepared from the tissue blocks and stained with hematoxylin and eosin. A blinded semiquantitative scoring system was used to assess synovial inflammation, extraarticular inflammation, bone erosion, and cartilage damage (0–5 scale), as previously described (19). For tissue cytokine assays, snap-frozen joints were homogenized in lysis solution as previously described. Protein concentration was measured by bicinchoninic acid assay (Pierce), and IL-1β and IL-6 were measured by enzyme-linked immunosorbent assay (ELISA) according to the recommendations of the manufacturer (R&D Systems) (24).
Determination of interferon-γ (IFNγ) secretion by splenocytes.
Mouse spleen cells were isolated and washed in RPMI 1640 supplemented with 10% FBS, 10 mM HEPES, 1 mM sodium pyruvate, 50 mM 2-mercaptoethanol, 1% L-glutamine, and 100 units/ml of penicillin/streptomycin. Erythrocytes were lysed. After washing, cells were counted, and 2 × 105 cells were placed in each well of a sterile, U-bottomed microculture plate in medium with 12.5 μg/ml mBSA. Cultures were maintained at 37°C for 2 days. IFNγ levels were measured by DuoSet ELISA (R&D Systems).
Determination of serum antibodies.
Methylated BSA–specific antibodies of various isotypes (IgG1, IgG2a) in the sera of individual mice were measured by ELISA. Antigen was coated on microtiter plates at a concentration of 10 μg/ml. Antibody titers were assessed using 2-fold serial dilutions of sera, followed by detection of bound mouse Ig with a 1:500 dilution of peroxidase-conjugated rabbit anti-mouse Ig. In the peroxidase reactions, o-phenylenediamine was used as a substrate.
Bone marrow–derived macrophages (BMMs).
To generate BMMs, mouse bone marrow cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) with 10% FBS and 20% L929 supernatant containing macrophage-stimulating factor for 6 days and were replated for the assays as indicated. BMMs were stimulated with LPS (100 ng/ml) and analyzed by qPCR and Western blotting.
Mouse ankle joints and paws were collected, dissected to remove extraarticular tissue, and snap-frozen in liquid nitrogen. For macrophages, cells were collected after BMM stimulation. Total RNA was extracted with TRIzol (Invitrogen) and reverse-transcribed with random hexamers and a SuperScript II kit (Invitrogen). Quantitative PCR was performed with a SYBR Green PCR Master Mix kit (Applied Biosystems). The relative amounts of transcripts were compared to those of 18S mRNA and normalized to untreated samples by the ΔΔCt method (25).
Western blot analysis.
Mouse BMMs or joints were disrupted in lysis buffer (PhosphoSafe; Novagen) containing a protease inhibitor cocktail. Proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto a nitrocellulose membrane. Blots were probed with antibodies against phospho–ERK-1/2, phospho-p38, phospho-JNK, phospho–STAT-3, ERK-1/2 (all from Cell Signaling Technology), JNK (BD Biosciences), STAT-3, p38, MKK-3, and MKK-6 (all from Santa Cruz Biotechnology). Horseradish peroxidase–conjugated anti-IgG (Santa Cruz Biotechnology) was used as a secondary antibody. Membranes were developed using a chemiluminescence system (ECL detection reagent; Amersham Life Science). Densitometry analysis was performed using Quantity One 1-D analysis software (Bio-Rad).
Human RA synovial cell cultures.
Human RA synovial tissue was digested with 0.5 mg/ml collagenase A in RPMI for 2 hours at 37°C. The cells were washed twice with 10% FBS/DMEM and filtered using a 0.70-μm cell strainer (Falcon). The cells were washed and counted, and 3 × 106 cells were plated in each well of a 6-well plate. After overnight incubation, the cells were treated with 3 μM of the p38 inhibitor SB203580 for 48 hours, and cytokines in the cell supernatants were quantified by multiplex analysis (Bio-Rad). The data are presented as the average percent of the values in untreated controls.
Data are expressed as the mean ± SEM. Student's unpaired t-test was used for comparing 2 groups, and analysis of variance was used for multiple group comparisons. P values less than 0.05 were considered significant.
Characterization of p38α-deficient mice.
To evaluate the contribution of p38α in macrophages in chronic inflammatory arthritis, mutant mice lacking p38α in macrophages were generated. Deletion of loxP-flanked alleles of the gene encoding p38α was mediated by Cre recombinase expressed under the control of the promoter of the myeloid-specific gene encoding lysozyme M (p38αΔLysM). The mutant mice were born alive and grew to adulthood without showing discernible anomalies or developing spontaneous disease. In these mice, p38α was not detectable in macrophages but was expressed by other myeloid-lineage cells, including neutrophils and mast cells (Figure 1A).
Increased severity of subacute inflammatory arthritis in mice after selective deletion of p38α in macrophages.
Passive K/BxN arthritis was studied in p38αΔLysM and p38αF/F mice. The p38αΔLysM mice had higher clinical scores from day 5 and a delay in the resolution phase of arthritis compared with WT mice. The mean ± SEM day-8 scores were 7.1 ± 1.2 and 9.7 ± 0.1 (P = 0.05), and the day-14 scores were 1.1 ± 0.3 and 3.0 ± 0.3 (P = 0.01) for p38αF/F and p38αΔLysM mice, respectively (Figure 1B). Histopathologic analysis showed a trend toward increased inflammatory cell infiltration, joint destruction, and cartilage damage in p38αΔLysM mice (Figures 1C and D). The overall histologic damage score was significantly greater in the p38αΔLysM mice than in the p38αF/F mice (mean ± SEM 1.85 ± 0.28 versus 1.12 ± 0.23, respectively; P < 0.05). Similar results were obtained in p38αF/F and p38α▵LysM mice in the AIA model, where the overall histology score was modestly increased in the p38-deficient mice (mean ± SEM 2.6 ± 0.1 versus 3.1 ± 0.2, respectively; P < 0.05). Of interest, the p38αF/F mice had normal adaptive immune responses, as measured by IFNγ production in vitro and antibody production in vivo (mean ± SEM IgG1 antibody titer 0.64 ± 0.05 in p38αF/F mice and 0.62 ± 0.05 in p38αΔLysM mice; similar results were obtained for IgG2a). These data suggest that the proinflammatory effect in macrophages is due to an effect on innate, not adaptive, responses.
To evaluate the influence of p38α deficiency in p38αΔLysM mice on inflammatory mediators in the passive K/BxN model, joints from a second group of mice were analyzed for expression of MMP-3, determined by qPCR, and expression of IL-1β and IL-6, determined by ELISA, on day 5 after injection of K/BxN sera. (The mean ± SEM day-5 arthritis scores were 6.8 ± 1.2 and 9.7 ± 1.3 in p38αF/F and p38αΔLysM mice, respectively.) MMP-3 relative mRNA was higher, and DUSP-1 mRNA expression was lower, in paw extracts from p38αΔLysM mice than in those from p38αF/F mice (P < 0.05 for both) (Figure 2A). However, IL-6 and IL-1β protein levels in p38αΔLysM mice compared to p38αF/F controls were similar at that time point (Figure 2B), possibly because they are produced by cells of other lineages in the joint or because genetic deficiency of p38α has less effect on IL-6 than does chemical inhibition (see below). IL-10 and TNF protein expression levels were at the limit of assay detection and could not be compared. Western blot analysis of joint extracts showed increased phosphorylation of ERK in the p38αΔLysM mice (Figure 2C).
Increased severity of chronic inflammatory arthritis in mice after selective deletion of p38α in macrophages.
To assess the functional role of p38α-deficient macrophages in a long-term model of arthritis, we induced chronic disease by injecting mice with K/BxN serum every 7 days for 4 weeks. As shown in Figure 3A, arthritis scores were significantly higher in the p38αΔLysM mice (mean ± SEM 5.2 ± 1.6 in WT mice and 10.2 ± 1.1 in p38αΔLysM mice on day 28; P = 0.05). In contrast with findings in the shorter-term model, IL-6 and IL-1β protein levels were higher in inflamed p38αΔLysM mouse joints than in WT mouse joints (P < 0.05) (Figure 3B). Western blot analyses also showed increased phosphorylation of STAT-3 (Figures 3C and D), consistent with the increased IL-6 levels.
Regulation of cytokines and MAPKs by p38 in cultured mouse macrophages.
To determine the mechanism of increased arthritis severity, WT mouse BMMs were cultured in the presence or absence of the p38α/β inhibitor SB203580 and stimulated with LPS. LPS-induced IL-10 and IL-6 mRNA levels were significantly decreased by the inhibitor (Figure 4A). Of interest, DUSP-1 expression was also decreased (P < 0.01) (Figure 4A). Phosphorylation of ERK and JNK was greater and more prolonged in the presence of the p38 inhibitor (Figure 4B), which correlated with the decrease in DUSP-1 expression. Similar results were observed in p38-deficient macrophages except for IL-6 expression, which was not affected.
Inhibition of p38α/β significantly decreases IL-10 expression in human RA synovial tissue cells.
Our data indicate that p38α inhibition in macrophages is proinflammatory in arthritis and decreases IL-10 production in BMMs. To assess whether a similar phenomenon is observed in RA synovium, human rheumatoid synovial tissue cells were cultured in the presence or absence of the p38 inhibitor SB203580. Supernatants were then assayed for TNF, IL-6, and IL-10. The compound substantially decreased TNF and IL-10 production. IL-6 production was only modestly decreased, consistent with findings previously described by others (26) (Figure 4C).
Regulation of cytokines and MAPKs in MKK-deficient mouse macrophages.
Previous studies suggested that targeting of upstream kinases in the p38 pathway might eliminate some of the problems associated with a direct p38 inhibitor (22). To evaluate the effect of MKK-3 or MKK-6 deficiency on macrophage function, we cultured MKK-3−/− and MKK-6−/− mouse BMMs in the presence or absence of LPS. MKK-6– and MKK-3–deficient mouse BMMs had decreased LPS-mediated IL-6 expression compared with WT cells (P < 0.001) (Figure 4A). Surprisingly, MKK-deficient mouse macrophages had normal IL-10 production, DUSP-1 expression, and JNK and ERK phosphorylation (see Figure 5 compared with Figure 4B). As previously described (23), MKK-3 deficiency decreased phospho-p38 levels, whereas MKK-6 deficiency did not. The lack of effect on phospho-p38 in the MKK-6−/− mouse cells is most likely due to the role of this MKK as a structural protein in the p38 complexes rather than as an active kinase.
MKK-6– and MKK-3–deficient marrow is protective in passive K/BxN arthritis.
To assess whether MKK-6 or MKK-3 deletion in bone marrow cells enhances or protects against passive K/BxN arthritis, bone marrow chimeras were generated by irradiating C57BL/6 and MKK-3−/− or MKK-6−/− recipient mice and reconstituting them with MKK-3−/− or MKK-6−/− and C57BL/6 donor bone marrow. After 8 weeks, passive K/BxN arthritis was induced. Arthritis severity was dependent on MKK-3 or MKK-6 expression in bone marrow–derived cells. Thus, WT chimeric mice with MKK-6– and MKK-3–deficient marrow exhibited markedly decreased arthritis severity, while WT bone marrow restored arthritis severity in MKK-3– and MKK-6–deficient mice (Figure 6).
Inhibitors of p38α demonstrate limited utility in the treatment of RA, despite abundant preclinical evidence predicting efficacy. One possible explanation is that p38α has antiinflammatory functions in addition to its well-defined proinflammatory actions. For instance, p38α in macrophages regulates MAPK phosphatases and immunosuppressive cytokines such as IL-10. Genetic deletion of the p38α gene in macrophages increased acute skin edema after toxic exposures such as ultraviolet light (12). These data raise the possibility that p38 deficiency in macrophages could increase the severity of chronic inflammation mediated by innate immunity. If that is the case, then it might explain why inhibition of this kinase results in only a limited benefit in diseases that are dominated by macrophage cytokines, such as RA.
Macrophages participate in the evolution of passive K/BxN arthritis (27) and in RA, where anti-TNF and IL-6 therapy demonstrate clinical efficacy (28). The number of macrophages in rheumatoid synovial biopsy specimens also correlates with joint damage (29), and depletion of macrophages by some therapeutic agents is associated with improvement in RA (30). Therefore, a therapy that interferes with macrophage deactivation could increase disease severity even though production of some pathogenic cytokines is suppressed.
Previous studies suggested that p38 blockade or deficiency in macrophages suppresses production of the antiinflammatory cytokine IL-10 and enhances activation of ERK and JNK (12). The present study confirmed this observation and led us to explore complex immune-mediated models of inflammation. In vitro experiments were focused on chemical p38 inhibition to mimic the situation in human clinical trials, although the results were similar to those for p38α-deficient macrophages. The only difference observed between genetic and small-molecule inhibitors was related to IL-6 expression, where genetic deletion appeared to have less effect.
The p38αΔLysM mice demonstrated increased disease severity in a transient arthritis model as well as a more persistent, month-long chronic arthritis model. Cytokine and MMP expression, presumably by other cell types, such as FLS or mast cells, was increased in the joints of p38-deficient mice, as was activation of downstream cytokine signaling molecules such as STAT-3. Of particular interest, phospho-ERK levels were also higher in the inflamed joints of p38αΔLysM mice and correlated with our findings in cultured macrophages. Activation of mast cells and activation of neutrophils are critical initiating events in this model, but p38αΔLysM mice have normal p38 expression in these lineages. Thus, the effects observed can be ascribed to macrophages. Similar results in the AIA model in the absence of altered adaptive responses indicate that the proinflammatory effect is due to an effect on innate immunity.
The murine data are also consistent with the role of p38 in cultured human RA synovial tissue cells. We confirmed that a p38α inhibitor substantially decreased TNF production from rheumatoid synovial tissue cells, as previously described (26). We also noted that IL-6 production was only modestly affected by the inhibitor, while IL-10 production was markedly decreased. These data support the notion that p38 regulates both proinflammatory and antiinflammatory cytokines in RA cells, which could interfere with clinical efficacy.
As an alternative strategy to targeting downstream kinases such as p38, we have advocated shifting emphasis to upstream signaling molecules (6). This approach has met with success, and it is especially noteworthy that JAK and Syk inhibitors demonstrate efficacy in RA (31, 32). Our previous studies in the p38 pathway suggest that inhibiting either of its two upstream regulators, namely, MKK-3 and MKK-6, might be more effective than a traditional direct p38 inhibitor. For instance, mice deficient in either MKK have decreased joint inflammation and destruction in the passive K/BxN model and in collagen-induced arthritis (22, 33, 34). MKK-3 deficiency also mimics p38 inhibitors in a murine model of allodynia (35).
Based on these studies, we evaluated the profile of MKK-deficient macrophages in vitro. The results demonstrated a notable dissociation of IL-6 and IL-10 regulation that distinguishes MKK and p38 function. Differential regulation can therefore provide antiinflammatory benefit by modulating, rather than blocking, the p38 pathway. We previously showed that p38 and MAPKAPK-2 (MK2) activities do not necessarily correlate in a linear manner. A threshold level of p38 activation might be necessary for efficient MK2 activation, which is not reached in either MKK-3−/− or MKK-6−/− cells (23). Increased activation of other MAPKs such as ERK and JNK was also substantially less in the MKK-deficient cells.
The potential benefit of targeting MKK-3 or MKK-6 was supported by the results of experiments using bone marrow chimeras. MKK-deficient marrow protected against passive K/BxN arthritis in normal mice, while normal marrow failed to correct the defect in MKK-3−/− or MKK-6−/− mice. These data suggest that murine macrophages lacking the MKKs are protective, in contrast to the p38α-deficient cells. The chimera study design does not precisely mimic the situation in p38αΔLysM mice because MKK-3 or MKK-6 is deleted in all bone marrow cell lineages. However, it provides evidence that targeting upstream kinases to cause p38 pathway deficiency in myeloid cells has the opposite effect of blocking p38 in macrophages.
Taken together, the present in vitro and in vivo results indicate that selectively blocking an MKK might be beneficial in inflammatory arthritis by sparing p38-regulated functions such as IL-10 and DUSP-1 expression. This approach could maintain negative feedback loops that are blocked by p38 inhibitors. Therefore, the data provide a rationale for strategies that inhibit upstream MKKs as a therapeutic approach in RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Firestein had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Guma, Corr, Karin, Firestein.
Acquisition of data. Guma, Topolewski.
Analysis and interpretation of data. Guma, Hammaker, Corr, Boyle, Karin, Firestein.