Fas Signaling in Macrophages Promotes Chronicity in K/BxN Serum–Induced Arthritis




A nonapoptotic role of Fas signaling has been implicated in the regulation of inflammation and innate immunity. This study was undertaken to elucidate the contribution of Fas signaling in macrophages to the development of arthritis.


K/BxN serum–transfer arthritis was induced in a mouse line in which Fas was conditionally deleted in the myeloid lineage (CreLysMFasflox/flox mice). The arthritis was assessed clinically and histologically. Expression of interleukin-1β (IL-1β), CXCL5, IL-10, IL-6, and gp96 was determined by enzyme-linked immunosorbent assay. Bone marrow–derived macrophages were activated with IL-1β and gp96. Cell phenotype and apoptosis were analyzed by flow cytometry.


Arthritis onset in CreLysMFasflox/flox mice was comparable with that observed in control mice; however, resolution was accelerated during the chronic phase. The attenuated arthritis was associated with reduced articular expression of the endogenous Toll-like receptor 2 (TLR-2) ligand gp96 and the neutrophil chemotactic chemokine CXCL5, and enhanced expression of IL-10. Activation with IL-1β or gp96 induced increased IL-10 expression in Fas-deficient murine macrophages compared with control macrophages. IL-10 suppressed IL-6 and CXCL5 expression induced by IL-1β plus gp96. IL-1β–mediated activation of ERK, which regulates IL-10 expression, was increased in Fas-deficient mouse macrophages.


Taken together, our findings indicate that impaired Fas signaling results in enhanced expression of antiinflammatory IL-10 and reduced expression of gp96, and these effects are associated with accelerated resolution of inflammation during the chronic phase of arthritis. These observations suggest that strategies to reduce endogenous TLR ligands and increase IL-10 may be beneficial in the treatment of rheumatoid arthritis.

Myeloid cells, including macrophages and neutrophils, are critical to the pathogenesis of rheumatoid arthritis (RA) through the release of proinflammatory cytokines, chemokines, and other mediators such as prostaglandins and matrix metalloproteinases ([1-3]). Monocytes, macrophages, and granulocytes express the death receptor Fas ([4]). RA synovial tissue ([5]) and RA synovial fluid macrophages ([6]) also express both Fas and FasL. We previously demonstrated that RA synovial fluid macrophages were resistant to FasL-mediated apoptosis, due to increased expression of the antiapoptotic molecule FLIP ([6]). Further, RA peripheral blood monocytes and synovial fluid macrophages are resistant to FasL-mediated apoptosis induced by activated CD4+CD25− responder T cells ([7]).

There is mounting evidence that Fas also has a nonapoptotic role, which is dependent on cell type and context ([8]). Fas signaling has been implicated in T cell proliferation and activation, liver regeneration after partial hepatectomy, and nerve regeneration after crush injury ([8]). The severity of collagen-induced arthritis (CIA) has been shown to be reduced in DBA/J mice that expressed a mutant Fas receptor (lpr) ([9]). Interruption of Fas–FasL interactions on human macrophages resulted in reduced Toll-like receptor 4 (TLR-4) and interleukin-1 receptor (IL-1R) signaling, mediated at least in part by damping of myeloid differentiation factor 88 (MyD88) signaling due to the increased availability of FADD, which was no longer recruited to the Fas receptor ([9]). To specifically examine the role of Fas on myeloid cells in vivo, we recently generated a myeloid-lineage Fas-deficient line by crossing Fasflox/flox and CreLysM mice ([10]). These mice exhibited no phenotypic abnormalities at 2–4 months of age (young); however, when they reached 6–8 months of age (old) they developed a systemic lupus erythematosus–like disease associated with leukocytosis, splenomegaly, antinuclear antibodies, and glomerulonephritis.

In order to determine the role of myeloid-specific Fas in the effector phase of arthritis, we studied young CreLysMFasflox/flox mice, using the immune complex–mediated K/BxN serum–transfer model of arthritis. These young CreLysMFasflox/flox mice exhibited no alteration in the initial induction of arthritis; however, once the clinical arthritis peaked, amelioration of disease was more rapid in CreLysMFasflox/flox mice. This improvement was associated with reduced inflammation and neutrophil infiltration. While there was no difference in levels of proinflammatory IL-1β or IL-6 within the involved joints, levels of IL-10 were significantly increased, and levels of the neutrophil chemotactic chemokine CXCL5 and the endogenous TLR-2 ligand gp96 were significantly reduced. There was no significant difference in the ability of IL-1β to induce IL-6 or CXCL5 in CreLysMFasflox/flox versus control mouse macrophages, but the level of IL-1β- and gp96–induced IL-10 was significantly increased in CreLysMFasflox/flox macrophages. Further, IL-10 suppressed the synergistic induction of IL-6 and CXCL5 in macrophages in response to IL-1β and gp96. These observations demonstrate that intact macrophage Fas signaling promotes ongoing inflammation by lessening the expression of IL-10 and enhancing the expression of the endogenous TLR-2 ligand gp96.


Generation and characterization of the mouse line with Fas deficiency in myeloid-lineage cells

The genetically modified mouse line with myeloid linage–specific deletion of floxed Fas (Fasflox/flox) was generated by crossing Fasflox/flox mice with mice expressing Cre driven by the myeloid lineage–specific LysM promoter (CreLysM), as previously described ([10]). Fas and CreLysM were genotyped by polymerase chain reaction using genomic DNA extracted from tail biopsy specimens. Mice genotyped with CreLysmFasflox/flox have deletion of Fas in myeloid cells. Littermates or age- and sex-matched mice with Fasflox/flox or CreLysMFas+ were used as controls. All animal procedures were approved by the Office of Research Safety and the Institutional Animal Care and Use Committee of Northwestern University.

Immunophenotyping and blood cell count

Fas cell surface expression was determined with multicolor fluorochrome-conjugated antibody cocktails, using antibodies to Fas or isotype-matched control IgG. Cell types were determined by immunophenotyping using antibodies to CD11b, F4/80, CD115, CD3, CD19, CD11c, and Gr-1 (eBioscience or BD PharMingen). Cell membrane integrity was assessed by exclusion of DAPI. Apoptosis was assessed by detection of annexin V (BD PharMingen). Data were acquired with a BD LSR II flow cytometer (BD FACSDiva software) and analyzed using FlowJo (Tree Star). Complete blood cell and differential cell counts were determined using a Hemavet 950 (Drew Scientific).

Induction and analysis of K/BxN serum–transfer arthritis

K/BxN mice were generated, and anti–glucose-6-phosphate isomerase (anti-GPI)–positive serum was collected when the mice were 8–9 weeks of age, as previously described ([11]). Arthritis was induced in 7–16-week-old CreLysMFasflox/flox mice and age- and sex-matched controls, using 100 μl anti-GPI–positive serum administered intraperitoneally on day 0. The development of arthritis was assessed by measuring hind ankle swelling (width) using a caliper and grading the clinical index of all 4 paws/ankles on a scale of 0–3 (maximum possible score 12) according to previously published methods ([12, 13]). Arthritis was evaluated through day 11 postinduction.

Ankles were harvested on day 11 for histologic analysis, cytokine/chemokine quantification, and immunophenotyping by flow cytometry. For histologic assessment, ankles were fixed in 10% neutral buffered formalin and then incubated in EDTA decalcification buffer for 2 weeks, after which they were embedded in paraffin and 4-μm sections were stained with hematoxylin and eosin. The stained ankle sections were evaluated, under blinded conditions, for inflammation (scored 0–5), bone erosion (scored 0–5), pannus formation (scored 0–5), synovial lining thickness, neutrophil infiltration, and cartilage destruction, as previously described ([13-15]).

For quantification of cytokines and chemokines, ankles were homogenized in phosphate buffered saline (PBS) containing a protease inhibitor cocktail and supernatants collected by centrifugation as previously described ([12]). Levels of IL-1β, IL-10, IL-6, and CXCL5 were determined by enzyme-linked immunosorbent assay (ELISA) (DuoSets; R&D Systems) according to the manufacturer's instructions. Concentrations of gp96 were determined using a mouse ELISA as previously described ([12, 16]). The concentration of each protein was adjusted to milligrams of total ankle homogenate protein, determined with a BCA Protein Assay kit (Thermo Scientific).

To further analyze the cells infiltrating the joints, tibias and paws were dissected, carefully removing muscle without opening the bone, using an established protocol ([17]). The intact tibias and paws, with the ankles and knees opened, were incubated for 1 hour in 1 mg/ml collagenase in PBS. Cells eluted from the knee and ankle joints were immunophenotyped by flow cytometry.

Macrophage differentiation and activation

Bone marrow–derived macrophages were generated from whole bone marrow cells isolated from the femurs and tibias of control and CreLysMFasflox/flox mice, followed by in vitro differentiation in RPMI 1640 medium supplemented with 10% (volume/volume) fetal bovine serum (FBS), 20 ng/ml mouse granulocyte–macrophage colony-stimulating factor (GM-CSF; R&D Systems), and 0.1 % (v/v) 2-mercaptoethanol (Gibco Invitrogen), as previously described ([18]). On day 7–9, the in vitro–differentiated macrophages were collected by gentle pipetting, washed with PBS, and allowed to rest for 1 hour prior to activation.

Recombinant human IL-1β was purchased from R&D Systems. Recombinant N-terminal domain of gp96 was prepared and purified with extensive endotoxin removal procedures, as previously described ([16, 19]). The endotoxin level in the recombinant gp96 used was below the level of detection by Limulus amebocyte cell lysate assay ([16]). Macrophages were seeded at 1 × 105/200 μl/well in 96-well cell culture plates. Cells were activated in RPMI 1640 medium supplemented with 10% FBS and 1 μg/ml polymyxin B. Macrophages were incubated with IL-1β or gp96 for 4 hours or 20 hours, and the supernatants were harvested and examined by ELISA for IL-10, IL-6, and CXCL5. The concentration of each cytokine was adjusted for cell number using an MTT assay (optical density 490 nm) at the time activation was terminated.


IL-1β–mediated macrophage Akt, glycogen synthase kinase 3 (GSK-3), ERK, and p38 signaling was assessed by immunoblot analysis. Bone marrow–derived macrophages were incubated with IL-1β (20 ng/ml) for lengths of time ranging from 15 minutes to 20 hours. Cells were lysed in buffer supplemented with 1× protease inhibitor and 1× phosphatase inhibitor cocktails (Sigma). Whole cell protein extracts (10–20 μg) were resolved by sodium dodecyl sulfate–10% polyacrylamide gel electrophoresis and transferred to PVDF membranes (Immobilon-P; Millipore) as previously described ([20]). Individual blots were probed with phosphospecific antibodies to Akt, GSK-3, ERK, or p38 (Cell Signaling Technology). After stripping, the same blot was reprobed with antibodies to each total protein and GAPDH (Cell Signaling Technology). The immunoblots were developed using enhanced ECL Prime Western Blotting detection reagent (GE Healthcare) and an Ultralum image acquisition system. Densitometric analysis of immunoblots was performed with ImageJ software (National Institutes of Health).

Statistical analysis

Student's 2-tailed t-test was used to analyze the significance of differences between groups. For samples that failed the normality test, the Mann-Whitney rank sum test was performed. Correlations were determined by Pearson's linear correlation, or by Spearman's nonparametric correlation if the data analyzed had a non-Gaussian distribution. P values less than 0.05 were considered significant.


Characterization of the genetically modified mouse line with Fas deficiency in the myeloid lineage

In order to investigate the in vivo role of Fas on myeloid cells, we generated a genetically modified mouse line with Fas deleted in the myeloid lineage (CreLysMFasflox/flox), which has been recently characterized ([10]). The CreLysMFasflox/flox mice developed normally and exhibited no phenotypic abnormalities prior to 6–8 months of age ([10]). The mice used in the present study were examined at 2–4 months of age. These young CreLysMFasflox/flox mice expressed Fas on B and T lymphocytes isolated from peripheral blood (ref.[10] and data not shown) and the peritoneal cavity (Figure 1A). In contrast, Fas was deleted in peritoneal macrophages (Figure 1A), and its expression on the surface of peritoneal macrophages from CreLysMFasflox/flox mice was significantly reduced compared with littermate controls (P < 0.001) (data not shown). Fas on neutrophils and monocytes in peripheral blood was also reduced compared with levels in controls (ref.[10] and data not shown).

Figure 1.

Reduced severity of serum transfer–induced arthritis in mice with Fas deficiency in the myeloid lineage. A, Fas membrane expression on cells isolated from the resident peritoneal cavities of control and CreLysMFasflox/flox mice, as examined by flow cytometry. Fas expression levels on the surface of peritoneal macrophages, B cells, and T cells were assessed based on mean fluorescence intensity (MFI). Representative flow cytometric images are shown. B, Course of arthritis in the 2 groups of mice, as assessed by the clinical activity score (0–3 for each limb; maximum possible score of 12) and the change in ankle thickness (hind ankles), determined between 2 and 11 days after arthritis induction. Data were combined from 5 individual experiments, with a total of 19–36 mice at each time point for each group. Values are the mean ± SEM. C, Representative photomicrographs of hematoxylin and eosin–stained ankle specimens from control and CreLysMFasflox/flox mice and comparison of histologic scoring (mean ± SEM) between ankles from 9 control mice and 12 CreLysMFasflox/flox mice, harvested on day 11 after arthritis induction. On the photomicrographs, arrows identify bone erosion and pannus formation, and brackets identify articular and extraarticular inflammation. PMN = polymorphonuclear cell. D, Flow cytometric analysis of cells isolated from joints harvested on day 11 after arthritis induction, and total number of cells (mean ± SEM) isolated from the joints of control mice (n = 6) and CreLysMFasflox/flox mice (n = 7). Hematopoietic cells were identified by CD45 positivity, and granulocytes (Gran) were identified by CD45+, CD11c−, CD11b+, and Gr1+ staining. E, Correlation between the number of granulocytes and the degree of ankle swelling, analyzed using the same ankles as in C. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

More rapid resolution of K/BxN serum transfer–induced arthritis following deletion of Fas in myeloid cells

The effect of Fas deletion in myeloid cells on the effector phase of immune complex–mediated arthritis was examined using the K/BxN serum–transfer model ([11, 21, 22]). CreLysMFasflox/flox mice developed arthritis that was initially comparable with that observed in controls. However, after the peak of clinical disease on day 7, significantly greater improvement was seen in CreLysMFasflox/flox mice compared with controls, as evaluated by clinical score or ankle thickness (Figure 1B). Consistent with this observation, histologic examination of ankles on day 11 revealed a significant reduction of joint and extraarticular inflammation (P < 0.001 and P < 0.05, respectively), as well as a reduction of neutrophil infiltration (P < 0.001) (Figure 1C). A decrease in the number of neutrophils was also identified by flow cytometry using cells lavaged from the joints of the CreLysMFasflox/flox mice, compared with controls (P < 0.01) (Figure 1D), and the number of neutrophils in the lavage fluid correlated with the clinical examination results just before the mice were killed on day 11 (r = 0.57, P < 0.05) (Figure 1E). There was a trend toward reduction of bone erosion and cartilage destruction, but this did not reach statistical significance (Figure 1C).

Effect of Fas deletion on monocyte and neutrophil migration

A previous study demonstrated that deletion of FasL in myeloid cells resulted in reduced thioglycollate-induced recruitment of macrophages and neutrophils, which was mediated by Fas–FasL interactions ([23]). Therefore, experiments were performed to determine if there was an intrinsic defect in recruitment of myeloid cells to a site of inflammation in CreLysMFasflox/flox mice. In the absence of stimulation there was no difference in the numbers of total neutrophils or monocytes in the circulation between CreLysMFasflox/flox and control mice (Figure 2A). Furthermore, following intraperitoneal injection of thioglycollate, there was no difference in the recruitment of macrophages at 72 hours, or neutrophils at 6 or 16 hours, between CreLysMFasflox/flox and control mice (Figure 2B). The decrease in neutrophils in the joints of the CreLysMFasflox/flox mice might be explained by increased apoptotic cell death. However, there was no increase in the number of annexin V–positive apoptotic neutrophils at 6 or 16 hours following intraperitoneal injection of thioglycollate in the CreLysMFasflox/flox mice (Figures 2C and D). These observations suggest that there were no intrinsic differences in recruitment of Fas-deficient neutrophils or monocytes to an inflammatory stimulus, and no increase of apoptosis at an inflammatory site by Fas-deficient neutrophils.

Figure 2.

No effect of Fas deletion on neutrophil migration and survival. A, Number of neutrophils and monocytes in the peripheral blood of control mice (n = 37) and CreLysMFasflox/flox mice (n = 40), as determined by complete blood cell counts. B, Number of thioglycollate-induced macrophages at 72 hours and granulocytes at 6 and 16 hours, as determined by flow cytometry (n = 10 per group). C, Representative annexin V and DAPI flow cytometric gating for granulocytes harvested at 6 hours. D, Mean fluorescence intensity (MFI) of annexin V expression on thioglycollate-induced peritoneal granulocytes at 6 hours (n = 6 per group) and 16 hours (n = 7 per group). Values are the mean ± SEM.

Increased IL-10 levels and decreased CXCL5 and gp96 levels in ankles from CreLysMFasflox/flox mice

IL-1β is essential for the development of anti-GPI–positive serum transfer–induced arthritis ([24]). Therefore, ankles were examined on day 11 to determine if altered expression of cytokines might contribute to the reduced arthritis in the myeloid cell–specific Fas-deficient mice. Contrary to expectations, both IL-1β and IL-6 were expressed to a comparable degree in the ankles of CreLysMFasflox/flox mice and control mice on day 11 (Figure 3A). In contrast, the neutrophil chemotactic chemokine CXCL5 was significantly reduced in the joints of the CreLysMFasflox/flox mice (P < 0.001), and the concentration of CXCL5 correlated with the degree of ankle swelling (r = 0.77, P < 0.001) (Figure 3B). Further, IL-10 was significantly increased in the ankles of the CreLysMFasflox/flox mice (P < 0.001), and its concentration in the joints was inversely correlated with ankle thickness (r = −0.59, P < 0.01) (Figure 3C).

Figure 3.

Increased interleukin-10 (IL-10) concentrations and reduced CXCL5 and gp96 concentrations in mice with Fas deficiency in the myeloid lineage. Concentrations of the cytokines IL-1β, IL-6, and IL-10 (A and C), the chemokine CXCL5 (B), and gp96 in ankles obtained on day 11 were determined by enzyme-linked immunosorbent assay. Mean ± SEM values in 9 control mice and 12 CreLysMFasflox/flox mice are shown; correlations with ankle thickness are also shown for CXCL5, IL-10, and gp96. ∗∗∗ = P < 0.001.

We recently demonstrated that the endogenous TLR-2 ligand gp96 was strongly expressed in anti-GPI–positive serum transfer–induced arthritis at the peak of clinical disease and that neutralization promoted disease amelioration ([12]). In the present study, the expression of gp96 was significantly reduced in the ankles of CreLysMFasflox/flox mice (P < 0.001), and the concentration of gp96 was strongly correlated with joint swelling (r = 0.73, P < 0.001) (Figure 3D). These findings indicate that the increased expression of IL-10 and reduction of CXCL5 and gp96 may contribute to the enhanced resolution of arthritis observed in the CreLysMFasflox/flox mice.

Altered activation of Fas-deficient macrophages by IL-1β and the TLR-2 ligand gp96.

Since IL-1β is essential for disease induction in the K/BxN serum transfer model ([24]) while gp96 promotes disease progression ([12]), the responses of bone marrow–derived macrophages to IL-1β and very-low-endotoxin gp96 were examined. There was no difference in the induction of either IL-6 or CXCL5 by IL-1β or gp96 in CreLysMFasflox/flox mouse macrophages, except for a reduction of IL-6 levels at 4 hours in response to gp96 (Figures 4A and B). In contrast, both IL-1β and gp96 induced significantly more IL-10 from Fas-deficient murine macrophages compared with control macrophages (P < 0.01) (Figure 4C). These observations suggest that increased IL-1β–induced IL-10 by macrophages may contribute to the more rapid improvement noted in the CreLysMFasflox/flox mice. Further, since the level of gp96 was reduced in the ankles of CreLysMFasflox/flox mice, its diminished expression may also have contributed to the diminished disease activity later in the clinical course.

Figure 4.

Increased induction of interleukin-10 (IL-10) by Fas-deficient macrophages in response to IL-1β and gp96. In vitro bone marrow–derived macrophages from control mice and CreLysMFasflox/flox mice were activated with recombinant IL-1β (5 ng/ml) or gp96 (5 μg/ml) for 4 hours or 20 hours. Concentrations of IL-6 (A), CXCL5 (B), and IL-10 (C) in culture supernatants were determined by enzyme-linked immunosorbent assay. Values were normalized for viable cell numbers using MTT and are the mean ± SEM of macrophages from 4 mice per group. ∗ = P < 0.05; ∗∗ = P < 0.01.

IL-10 suppression of macrophage activation by IL-1β and gp96.

IL-1β and gp96 are each capable of promoting macrophage activation and are highly expressed in the ankles of wild-type mice following the induction of serum transfer arthritis ([12]). Therefore, the role of IL-10 in suppressing macrophage activation by these mediators was examined. The combination of IL-1β and gp96 synergistically induced the expression of IL-6 and CXCL5 (Figures 5A and B), but not IL-10 (data not shown). IL-10 significantly suppressed the expression of IL-6 and CXCL5 induced by IL-1β plus gp96 (P < 0.05–0.001) (Figures 5C and D). These results provide evidence that the increased expression of IL-10 in CreLysMFasflox/flox mice may contribute to the reduced inflammation, due in part to suppression of proinflammatory cytokines and chemokines, such as CXCL5, downstream of IL-1β and gp96.

Figure 5.

Interleukin-10 (IL-10) suppression of the synergistic effect of IL-1β and gp96 in activating bone marrow–derived macrophages. Bone marrow cells from wild-type B6 mice were differentiated to macrophages in vitro and incubated with recombinant IL-1β (5 ng/ml) or gp96 (10 μg/ml) alone or in combination, with or without preincubation with recombinant murine IL-10 at 5 or 10 ng/ml. A and B, Concentrations of IL-6 (A) and CXCL5 (B) induced without IL-10 preincubation. The results obtained with the addition of IL-1β alone and the results obtained with the addition of gp96 alone were summed, and compared statistically with the results obtained with the addition of IL-1β and gp96 together. C and D, Amount of IL-6 (C) and CXCL5 (D) induced with IL-10 preincubation, expressed as a percentage of the concentrations obtained without IL-10 preincubation. (Values obtained from the experiments shown in A and B were set at 100% and are shown for IL-10 at 0 ng/ml.) Values are the mean ± SEM of macrophages from 4 mice per group. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

Mechanism for the increased induction of IL-10.

To identify the potential mechanism for the increased IL-1β–mediated induction of IL-10 by Fas-deficient murine macrophages, we examined pathways known to regulate the expression of IL-10, including the phosphatidylinositol 3-kinase/Akt, GSK-3, ERK, and p38 pathways. The activation of Akt, GSK-3, and p38 did not differ between CreLysMFasflox/flox mice and control mice (results not shown). In contrast, IL-1β–induced ERK activation was increased in experiments using macrophages from CreLysMFasflox/flox mice compared with macrophages from control mice (Figure 6). This observation suggests that intact Fas signaling mediated through ERK modulates the IL-1β–induced expression of IL-10 by macrophages.

Figure 6.

Increased interleukin-1β (IL-1β)–mediated ERK activation in Fas-deficient macrophages. In vitro–differentiated macrophages were incubated without (No) or with IL-1β (20 ng/ml) for 15–1,200 minutes (20 hours). Macrophage cell lysates were collected at the indicated time points and analyzed by immunoblotting, using antibodies to p44 and p42 phospho-ERK, p44 and p42 total ERK, and GAPDH. The numbers below the blots for p44 and p42 phospho-ERK represent the fold change compared with time 0, after adjusted loading with GAPDH. The relative expression of each protein was calculated by densitometry using ImageJ software. Gels shown are representative of blots from 3 mice.


This study demonstrates that Fas expression on myeloid cells promotes anti-GPI–positive serum transfer–induced arthritis. The initial phase of the clinical course of arthritis in CreLysMFasflox/flox mice was comparable with that observed in control mice. However, the CreLysMFasflox/flox mice exhibited more rapid amelioration during the chronic phase of the disease, which was confirmed histologically by the findings of reduced inflammation and decreased neutrophils in the joints (Figure 1). At the time the mice were killed (day 11), levels of IL-10 were significantly increased, and levels of the neutrophil chemotactic chemokine CXCL5 and the endogenous TLR-2 ligand gp96 were significantly reduced, in the ankles of the CreLysMFasflox/flox mice (Figure 3). IL-1β is critical for the initiation of serum transfer–induced arthritis ([24]), while local expression of gp96 within the joint promotes progression of disease after peak clinical activity ([12]). Activation of Fas-deficient macrophages with IL-1β or gp96 resulted in significantly higher concentrations of IL-10 compared with the levels in control macrophages, and this was associated with increased phosphorylation of ERK. These observations suggest that the absence of intact Fas signaling in macrophages results in reduced gp96 and increased IL-10, promoting resolution of the arthritis. This may be mediated, at least in part, by reduction of chemokines such as CXCL5.

A recent study demonstrated that myeloid cell FasL promotes central nervous system inflammation following injury ([23]). Mice with deficiency of FasL in myeloid cells exhibited reduced spinal cord damage, reduced neutrophil infiltration, and reduced neutrophil and monocyte recruitment to the peritoneum following injection with thioglycollate ([23]). In contrast, in the current study, although the number of neutrophils was decreased in the inflamed joints of CreLysMFasflox/flox mice, thioglycollate-induced recruitment of Fas-deficient neutrophils was not reduced. Neutrophil apoptosis is crucial to the resolution of inflammation ([25]), and reduction of neutrophil apoptosis exacerbates inflammatory arthritis ([26]). Fas signaling may promote neutrophil apoptosis, which is inhibited by the antiapoptotic proteins Mcl-1 and Bcl-2 ([27]), while Mcl-1 is essential for neutrophil survival in the absence of a death signal ([28, 29]). We observed no difference in spontaneous apoptosis by Fas-deficient neutrophils. Therefore, although neutrophil numbers were reduced in the ankles of CreLysMFasflox/flox mice, there was no evidence of an intrinsic defect in neutrophil migration and no spontaneous increase of neutrophil cell death.

Neutrophils are critical to the pathogenesis of anti-GPI–positive serum transfer–induced arthritis ([30]). The neutrophil chemotactic chemokines CXCL1, CXCL2, and CXCL5 and the receptor for these chemokines CXCR2 are increased in the joints of mice with serum transfer–induced arthritis ([31]). Mice deficient in CXCR2 exhibit attenuated serum transfer–induced arthritis, with an onset similar to that observed in control mice ([31]), consistent with the course observed in the CreLysMFasflox/flox mice in the present study. We chose to examine CXCL5 because it was previously demonstrated that anti-CXCL5 ameliorates IL-17–induced arthritis but anti-CXCL1 does not ([32]). Reduced levels of CXCL5 were observed in the joints of the CreLysMFasflox/flox mice, and at the time of tissue acquisition the concentration of CXCL5 was highly correlated with joint swelling, which also correlated with the number of neutrophils in the joints. Since there was no intrinsic defect in the recruitment or survival of Fas-deficient neutrophils, these findings indicate that reduced CXCL5 contributed to the observed disease amelioration during the chronic phase of arthritis in the CreLysMFasflox/flox mice.

Consistent with the results of an earlier study in DBA/llpr/lpr mice with CIA ([9]), there was no reduction in levels of the proinflammatory cytokines IL-1β or IL-6 in the joints of the CreLysMFasflox/flox mice. Therefore, reduced IL-1β, which is necessary for induction of the arthritis, does not appear to be the cause of the enhanced amelioration of the disease in the CreLysMFasflox/flox mice. Further, the induction of CXCL5 expression by IL-1β was not reduced in Fas-deficient mouse macrophages, suggesting that reduced IL-1β–induced CXCL5 was not responsible for the amelioration of arthritis. However, we observed that levels of the endogenous TLR-2 ligand gp96 were significantly reduced in the joints of the CreLysMFasflox/flox mice on day 11 (Figure 3D). We previously demonstrated that gp96 concentrations in the ankles were increased during serum transfer–induced arthritis, peaking at the time of maximal joint swelling, and that neutralizing anti-gp96 antibody ameliorated the clinical course ([12]). Since gp96 induced the expression of CXCL5 comparably in Fas-deficient and control macrophages, the reduction of gp96 in the CreLysMFasflox/flox mice may have contributed to the reduced CXCL5 observed in the arthritic joints. In support of this interpretation, reduced levels of CXCL5, but not IL-10, were observed on day 4 in the ankles of the CreLysMFasflox/flox mice compared with control mice in additional experiments using 300 μl of K/BxN serum (data not shown).

The mechanism for the induction of gp96 in arthritis is unknown, although IL-2 and interferon-γ are known to promote its expression ([33, 34]). Gp96 is highly expressed in RA synovial tissue and synovial fluid, and is capable of activating macrophages through TLR-2 ([12, 16]). These features identify a pivotal role of gp96 in the persistent activation of macrophages, leading to the self-perpetuating inflammatory process observed in RA ([35]). It is possible that other endogenous TLR ligands, such as tenascin C or high mobility group box chromosomal protein 1, may also contribute to the progression of arthritis ([36-38]). These observations suggest that a reduction of the endogenous TLR-2 ligand gp96 may contribute to the amelioration of arthritis later in the clinical course in the CreLysMFasflox/flox mice.

In a previous study using human macrophages that express both Fas and FasL, interruption of Fas signaling resulted in suppressed IL-1R1 and TLR-4 signaling which was mediated by the interaction of FADD with MyD88, resulting in reduced expression of IL-6 ([9]). Further, macrophages from lpr and gld mice exhibited reduced TLR-4–induced IL-6. Consistent with these observations, in the current study Fas-deficient murine macrophages exhibited reduced levels of IL-6 at 4 hours in response to the endogenous TLR-2 ligand gp96 (Figure 4A), as well as to microbial TLR-2 and TLR-4 ligands (data not shown). However, Fas-deficient murine macrophages exhibited no reduction in the induction of IL-6 or CXCL5 in response to IL-1β. Nevertheless, in human macrophages, interruption of Fas–FasL signaling via an antagonistic anti-FasL antibody resulted in increased IL-1β–induced IL-6 expression and NF-κB activation ([9]). These differences may be due to the fact that we were not able to definitively detect FasL on the surface of murine macrophages (data not shown), consistent with the earlier observation that expression of FasL on the surface of unmanipulated macrophages is quite low ([39]). Therefore, the influence of Fas–FasL interactions between macrophages may be less dramatic in mice than in humans. Nonetheless, DBA/llpr/lpr mice exhibited no reduction of cellular or humoral immunity to collagen ([9]), consistent with the important role of myeloid-expressed Fas shown in the current study.

IL-10 is important in the pathogenesis of rheumatoid and experimental arthritis. It is expressed in the joints of patients with RA ([40]). In studies using ex vivo RA synovial tissue cultures, neutralization of IL-10 promoted, and the addition of exogenous IL-10 suppressed, the spontaneous expression of proinflammatory cytokines ([40]), supporting the notion that it plays a key role in controlling inflammation in RA. Further, the absence of IL-10 exacerbates CIA ([41, 42]), while treatment with IL-10 suppresses it ([43]). Additionally, the severity of K/BxN serum transfer–induced arthritis is increased in IL-10–deficient mice ([44]), demonstrating that IL-10 suppresses, but does not prevent, arthritis in this model. In the present study IL-10 levels were significantly higher in the joints of CreLysMFasflox/flox mice than in controls, suggesting that increased IL-10 may contribute to the amelioration of arthritis during the chronic phase of the disease in these mice.

Other mechanisms for the reduced severity of arthritis during the chronic phase, in addition to reduced levels of gp96, are also possible. Concentrations of IL-1R antagonist (IL-1Ra) are also increased in RA and, although only modestly effective therapeutically, IL-1Ra is approved for the treatment of the disease. IL-1Ra–deficient mice spontaneously develop arthritis ([45]) and develop more severe K/BxN serum transfer–induced arthritis, while IL-1Ra–transgenic mice are largely protected ([44]). Type I interferon-β is also expressed by macrophages and other cell types and is capable of suppressing anti-GPI–mediated arthritis ([44]), although it was not effective in RA ([46]). Overall, these observations indicate that the enhanced expression of IL-10, and possibly other antiinflammatory cytokines, contributed to the amelioration of arthritis observed in CreLysMFasflox/flox mice.

In experiments to examine the mechanism for the increased concentrations of IL-10 in the joints of CreLysMFasflox/flox mice, we found that incubation of Fas-deficient murine macrophages with either IL-1β or gp96 resulted in increased IL-10 expression. Activation of ERK and activation of p38 are important for the induction of IL-10 by microbial TLR ligands and immune complexes, while activation of Akt exerts a permissive effect by reducing the activity of GSK-3 ([47]). In response to IL-1β, only ERK pathway expression was differentially regulated between macrophages from CreLysMFasflox/flox mice and from control mice. There was no difference in the expression of markers of alternative macrophage differentiation, including Fizz1, Arg1, Ym1, or IL-10, in GM-CSF–differentiated bone marrow macrophages (data not shown). In addition, macrophages from aged CreLysMFasflox/flox mice exhibited no increase in the constitutive activation of ERK, p38, or Akt ([10]). These results provide evidence that increased ERK activation in response to IL-1β, rather than a difference in the pattern of macrophage differentiation, contributed to the enhanced expression of IL-10 in CreLysMFasflox/flox mice.

Taken together, the findings reported herein suggest that in the presence of intact Fas signaling in macrophages, the expression of IL-10, and possibly other suppressive cytokines, is restrained, permitting the full expression of inflammation. Further, the induction of the endogenous TLR-2 ligand gp96 is enhanced in the presence of intact macrophage Fas signaling, promoting disease progression and further joint destruction. These observations also provide a rationale for therapeutic strategies in RA that reduce the expression of endogenous TLR ligands and promote enhanced expression of IL-10.


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. Pope 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. Huang, Birkett, Koessler, Cuda, Haines, Jin, Pope.

Acquisition of data. Huang, Birkett, Koessler, Cuda, Jin.

Analysis and interpretation of data. Huang, Haines, Perlman, Pope.