To elucidate the role of tumor necrosis factor α–induced adipose-related protein (TIARP; or tumor necrosis factor α–induced protein 9 [TNFAIP-9]) in the development and pathogenesis of arthritis.
To elucidate the role of tumor necrosis factor α–induced adipose-related protein (TIARP; or tumor necrosis factor α–induced protein 9 [TNFAIP-9]) in the development and pathogenesis of arthritis.
We generated TIARP-deficient (TIARP−/−) mice and investigated several organs in aged mice. Peritoneal macrophages were collected and cultured with lipopolysaccharide (LPS) and TNFα, and then the production of cytokines and subsequent NF-κB signal transduction were analyzed. We also examined the susceptibility of young TIARP−/− mice to collagen-induced arthritis (CIA). Draining lymph nodes and splenocytes were isolated and cultured, and serum levels of anti–type II collagen (anti-CII) antibodies, interleukin-6 (IL-6), and TNFα on day 60 were measured. We further investigated the effects of anti–IL-6 receptor monoclonal antibody (mAb) on the development of arthritis in TIARP−/− mice. IL-6/STAT-3 signaling was also analyzed using TIARP−/− macrophages.
TIARP−/− mice developed spontaneous enthesitis and synovitis, had high serum levels of IL-6, had increased CD11b+ cell counts in the spleen, and showed enhanced LPS- and TNFα-induced IL-6 expression in macrophages. Sustained degradation of IκBα with dysregulated apoptosis was also noted in TIARP−/− macrophages. CIA was clearly exacerbated in TIARP−/− mice, accompanied by marked neutrophil and macrophage infiltration in joints. The levels of anti-CII antibodies in serum were unchanged, whereas autoreactive Th1 cell and Th17 cell responses were higher in TIARP−/− mice. Treatment with anti–IL-6 receptor mAb prevented the development of CIA in TIARP−/− mice, and TIARP−/− macrophages showed increased IL-6–induced STAT-3 phosphorylation.
These findings suggest that TIARP acts as a negative regulator of arthritis by suppressing IL-6 production, its signaling and TNFα-induced NF-κB signaling, resulting in enhanced apoptosis in macrophages.
The prognosis in patients with rheumatoid arthritis (RA) has improved significantly with the recent availability of biologic agents that target tumor necrosis factor α (TNFα) and interleukin-6 (IL-6) (1, 2). However, the exact mechanisms of action of these agents remain largely unknown. Glucose-6-phosphate isomerase (GPI) was first identified as an arthritogenic target in K/BxN mice (3), and GPI immunization of DBA/1 mice was shown to lead to arthritis (GPI-induced arthritis) (4). Studies in GPI-induced arthritis showed clear therapeutic benefits of treatment with TNFα and IL-6 antagonists (5, 6), suggesting its suitability for analyzing the mechanisms of action of these agents in RA.
Using GeneChip analysis, we previously demonstrated the up-regulation of TNFα-induced adipose-related protein (TIARP) in GPI-induced arthritis (7), as well as in CD11b+ splenocytes and joints of mice. Human TIARP counterparts, such as 6-transmembrane epithelial antigen of prostate 4 (STEAP-4), were also found to be highly expressed in peripheral blood monocytes, neutrophils, and synovial CD68+ cells from patients with RA (7, 8). However, the role of TIARP in the pathogenesis of autoimmune arthritis remains to be confirmed.
In this study, we found that TIARP-deficient (TIARP−/−) mice develop spontaneous enthesitis and synovitis, with high numbers of macrophages and elevated IL-6 expression. TIARP−/− macrophages showed enhanced NF-κB signaling, with dysregulated TNFα-induced apoptosis, and showed increased IL-6–induced STAT-3 phosphorylation. Moreover, the development of collagen-induced arthritis (CIA) was markedly exacerbated in TIARP−/− mice, suggesting that TIARP may provide new insights into the pathogenesis of arthritis as a negative regulator.
The TIARP−/− mouse line was generated by homologous recombination in embryonic stem cells from mice of the C57BL/6 (B6) background. A conditional targeting vector for the TIARP gene was designed to delete exon 2, with a DNA fragment containing a loxP-flanked neomycin resistance gene and herpes simplex virus thymidine kinase gene (illustration available upon request from the corresponding author). All mice were maintained under specific pathogen–free conditions in an environmentally controlled clean room at the University of Tsukuba. All animal experiments were approved by the institutional committee and were conducted in accordance with the institutional ethics guidelines.
When the mice were 12 months of age, we removed the primary organs and tissues, including the ankle joints, lungs, salivary glands, white adipose tissue, liver, thymus, spleen, inguinal lymph nodes, stomach, small intestine, large intestine, kidneys, and reproductive organs, for histopathologic examination. Tissues were fixed in neutralized 10% formalin, and the joints of the right hind paw were decalcified in 10% formic acid and embedded in paraffin. We stained 4-mm serial sections with hematoxylin and eosin or toluidine blue and scored the tissues for histologic changes. Sections were graded on a scale of 0–5, where 0 = normal and 5 = severe. One point was given for the presence of each of the following features of enthesitis: inflammatory cell infiltration, entheseal fibroblast-like cell proliferation, cartilage formation, bone formation, and ankylosis. The same system was used to grade synovitis, with 1 point given for each of the following features: inflammatory cell infiltration, synovial hyperplasia, pannus formation, cartilage degradation, and bone degradation.
For immunohistochemical staining, the sections were incubated overnight at 4°C with rat anti–Gr-1 or anti-F4/80. The signals were detected with Alexa 488–conjugated anti-rat IgG. Nuclei were counterstained with DAPI. The sections were examined under fluorescence microscopy (Keyence).
For flow cytometry, cells were stained with fluorescein isothiocyanate (FITC)–, phycoerythrin-, PerCP-, or allophycocyanin-conjugated monoclonal antibodies (mAb). Rat mAb to mouse CD3, CD4, CD8, CD11b, Gr-1, F4/80, CD11c, IL-17, and interferon-γ (IFNγ) were purchased from BioLegend, and rat mAb to mouse FoxP3 was purchased from eBioscience. Apoptosis was detected by staining with propidium iodide and FITC-conjugated annexin V (BioLegend). We performed cell surface staining according to standard techniques. Stained cells were identified with a FACSCalibur cytometer (Becton Dickinson) using FlowJo software (Tree Star).
We isolated total RNA from splenocytes and ankle joints by the Isogen (Nippon Gene) extraction method according to the instructions provided by the manufacturer. We performed real-time quantitative PCR as described previously (7) using a TaqMan gene expression assay (Applied Biosystems) and the following probes: TIARP (Mm00475402_m1), TNFα (Mm00443258_m1), IL-6 (Mm00446190_m1), CXCL2 (Mm00436450_m1), matrix metalloproteinase 3 (MMP-3) (Mm00440295_m1), RANKL (Mm00441906_m1), and GAPDH (Mm99999915_g1). Real-time quantitative PCR was carried out using an ABI 7500 analyzer (Applied Biosystems). Analysis of post-PCR melting curves confirmed the specificity of the single-target amplification. We determined the expression of each gene relative to GAPDH.
Mice were injected intraperitoneally with 2 ml of 3% thioglycolate. After 3 days, mice were euthanized and peritoneal macrophages were collected by phosphate buffered saline (PBS) lavage. Cells were incubated for the indicated durations in the presence or absence of 100 ng/ml of TNFα, 100 ng/ml of lipopolysaccharide (LPS), or 10 ng/ml of IL-6.
Cells were lysed in 0.5% Nonidet P40, 5 mM MgCl2, 50 mM Tris HCl (pH 7.4), and 2 mM phenylmethylsulfonyl fluoride. Equal amounts of protein were subjected to immunoblotting using antibodies to IκBα, phospho-STAT-3, STAT-3, and suppressor of cytokine signaling 3 (SOCS-3), as well as phospho-ERK-1/2, ERK-1/2, caspase 3, cleaved caspase 3 antibodies (Cell Signaling Technology), and anti-actin antibodies (Bio-Rad). Densitometric analysis was carried out using an ImageQuant LAS 4000 densitometer (GE Healthcare).
CIA was induced in mice of the B6 background. We immunized 8–12-week-old mice intradermally at several sites at the base of the tail with 100 μl of 200 μg chicken type II collagen (CII; Sigma) emulsified in Freund's incomplete adjuvant (Difco) and containing 5 mg/ml of heat-killed Mycobacterium tuberculosis (H37Ra; Difco). According to the usual immunization schedule, mice were rechallenged with CII and Freund's complete adjuvant on day 21 after the primary immunization. Arthritis was assessed every other day by examining each joint for swelling and redness. Arthritis severity was graded on a scale of 0–3 in each paw. The clinical score for each mouse was the sum of the scores for the 4 paws (maximum score 12).
Single- cell suspensions were prepared from the spleens and lymph nodes of wild-type (WT) and TIARP−/− mice. Cells were cultured for 72 hours at 37°C in an atmosphere containing 5% CO2, and supernatants were collected. Levels of IL-6 and TNFα in the cell culture supernatants were measured using ELISA kits (R&D Systems). Serum samples were collected on days 30 and 60 from arthritic WT and TIARP−/− mice, and serum levels of TNFα, IL-6, IL-1β, and granulocyte–macrophage colony-stimulating factor (GM-CSF) were measured with ELISA kits (R&D Systems). Serum levels of anti-CII antibodies were also analyzed for CII-specific IgG antibodies by ELISA. All serum samples were diluted 1:3,000 in PBS prior to ELISA.
For IL-6 neutralization, mice were injected intraperitoneally with 2 mg of MR16-1 (a rat IgG1 mAb against murine IL-6 receptor [IL-6R]) or control IgG (purified from the serum of nonimmunized rats) on day 21 after induction of CIA. MR16-1 was a gift from Chugai Pharmaceutical, and control IgG was purchased from Jackson ImmunoResearch. To neutralize TNFα, mice were injected intraperitoneally with 100 μg of neutralizing antibody or isotype control on day 21. Anti-TNFα mAb MP6-XT3 (IgG1, rat) and IgG1 isotype control (rat) were purchased from eBioscience.
In the CIA experiments, disease incidence was evaluated with the chi-square test and the severity score by the Mann-Whitney U test. Student's t-test was used for the evaluation of all other results. P values less than 0.05 were considered significant.
TIARP−/− mice were generated to further investigate the function of TIARP in arthritis. To do this, we used homologous recombination in embryonic stem cells derived from B6 mice to produce a line of mice carrying a defective TIARP gene. In the homozygous state, these mice carry a mutant allele with exon 2 deleted (data available upon request from the corresponding author). The expression of the mature TIARP gene and protein in splenocytes, peritoneal macrophages, and synoviocytes (data not shown) could not be detected by real-time quantitative PCR and Western blotting.
We then tested whether deletion of the TIARP gene could directly induce organ autoimmunity. We found that 76.9% of the deficient mice developed joint abnormalities by 12 months of age (Figure 1A and Table 1). Furthermore, marked enthesopathy and weak synovitis with joint destruction were observed in these TIARP−/− mice (Figure 1A). These findings were confirmed by histologic scoring of inflammation and erosion (Figure 1A). We also screened several other organs isolated from 20-week-old and 12-month-old mice. The murine white adipose tissues showed weak cell infiltration, confirming the findings of Wellen et al (9), although there was no significant cell infiltration or damage in the liver, kidney, or small intestine (Figure 1B) or in the white adipose tissue, large intestine, or other organs (data available upon request from the corresponding author).
|Characteristic||Wild-type mice (n = 13)||Knockout mice (n = 13)|
|Incidence of arthritis, no. (%)||0 (0.0)||10 (76.9)*|
|Mean ± SEM clinical score||0.00 ± 0.00||2.25 ± 0.62*|
|Inflammation, mean ± SEM||1.25 ± 0.25||2.75 ± 0.48†|
|Erosions, mean ± SEM||0.25 ± 0.25||2.25 ± 0.63†|
|Cartilage damage, mean ± SEM||0.50 ± 0.29||1.75 ± 0.75†|
Immune cell development and function in TIARP−/− mice were investigated next. The involvement of acquired immune cells in the spleen (Figure 1C) and in the thymus and mesenteric and inguinal lymph nodes (LNs) (data not shown) in TIARP−/− mice was almost comparable to WT mice. Because TIARP is highly expressed in CD11b+ cells during GPI-induced arthritis (7), we screened the spleens of TIARP−/− mice and found significantly high numbers of CD11b+Gr-1low/intermediate cells (confirmed by microscopy to be macrophages) (Figure 1C).
We next examined inflammatory cytokines in the TIARP−/− mice, since they are also key players in RA and are functionally linked to TIARP (10–12). Serum levels of IL-6 were higher in TIARP−/− mice than in WT mice irrespective of age, whereas no TNFα or GM-CSF was detected, and IL-1β levels were similar in WT and TIARP−/− mice (Figure 1D).
Taken together, these findings suggested that TIARP−/− mice develop spontaneous enthesopathy and synovitis, with elevated numbers of macrophages and elevated levels of IL-6, which suggest a pivotal role of macrophages.
Next, we focused on CD11b+ cells, based on the dominant expression of TIARP in these cells in GPI-induced arthritis (7) and on the spontaneous aberrant accumulation of macrophages in TIARP−/− mice. Thioglycolate-activated peritoneal macrophages from TIARP−/− mice produced higher amounts of IL-6, TNFα, and IL-1β upon LPS stimulation (Figure 2A). Macrophages from TIARP−/− mice also produced higher amounts of IL-6 with TNFα stimulation (Figure 2A). Levels of IL-6Rα, TNF receptor type I (TNFRI), and TNFRII expression were similar in macrophages from TIARP−/− and WT mice (data not shown).
We then examined the role of TIARP in the NF-κB pathway and in apoptosis. Macrophages obtained from TIARP−/− mice and cultured with TNFα showed sustained degradation of the NF-κB inhibitory molecule IκBα as compared to macrophages from WT mice (Figure 2B). Moreover, TNFα-induced apoptosis was increased in macrophages from WT mice, but was not significantly different in those from TIARP−/− mice (Figure 2C). Cleaved caspase 3 levels in the presence of TNFα were clearly diminished in macrophages from TIARP−/− mice as compared to WT mice (Figure 2D). Adding both TNFα and IL-6 induced proliferation in macrophages from TIARP−/− mice as compared to WT mice (data not shown).
Collectively, these results suggested that macrophages from TIARP−/− mice have a high level of response to TNFα due to a weak NF-κB-negative regulation and subsequent proliferation, with dysregulated apoptosis.
The direct pathogenic effect of TIARP in arthritis was explored using a model of CIA on a B6 background. We first examined TIARP fluctuations in CIA by screening for changes in TIARP expression in WT B6 mice. Real-time quantitative PCR showed up-regulated TIARP mRNA expression on day 23 in splenocytes from mice with CIA and on day 28 in the joints of the same mice, when the expression correlated with joint swelling (Figure 3A). These fluctuations resembled those seen in GPI-induced arthritis (7) and suggest that systemic up-regulation of TNFα and TIARP is involved in the early phases of the disease.
Next, we induced CIA in 8–12-week- old TIARP−/− mice and found exacerbated disease incidence and severity in these mice as compared to the WT mice (Figure 3B). Histologic analysis of TIARP−/− mice with CIA showed marked cell infiltration, pannus formation, and bone erosion, as well as increased histology scores (Figure 3C). Immunohistochemical analysis of cells isolated from the mouse joints confirmed the dominance of neutrophil and macrophage infiltration in the joints of TIARP−/− mice (Figure 3D).
On day 10 after immunization with CII, the TIARP−/− mice had splenomegaly, and their CD4 and CD8 T cell counts were higher than those in WT mice (Figure 4A). In contrast, CD19+ cell counts were not different between WT mice and TIARP−/− mice (data not shown). We then screened T and B cell responses during CIA to investigate antigen-specific responses. The production of antigen-specific cytokines, such as IFNγ and IL-17, was significantly higher in draining LNs from TIARP−/− mice than in those from WT mice (Figure 4B), whereas the numbers of antigen-specific Treg cells (FoxP3+) were not significantly different (Figure 4C). Unexpectedly, the levels of antigen- specific anti-CII antibodies in TIARP−/− mice and WT mice were comparable on day 30 and on day 60 (Figure 4D).
Serum IL-6 levels on day 60 after immunization were also markedly increased in TIARP−/− mice compared to WT mice, whereas TNFα was not detected (Figure 5A). Gene expression analysis in mouse joints on day 60 showed higher levels of IL-6, CXCL2, MMP-3, and RANKL in TIARP−/− mice as compared to WT mice (Figure 5B).
To further confirm the role of IL-6 in arthritis, IL-6 was inhibited after the onset of CIA (on day 21; this timing was generally not effective in WT mice ). As anticipated, injection of anti–IL-6R mAb significantly suppressed the incidence and progression of arthritis (Figure 5C, top), and histopathologic analysis on day 60 confirmed this effect (Figure 5C). In contrast, treatment with TNFα-neutralizing antibodies in TIARP−/− mice did not suppress the progression of arthritis (Figure 5C, bottom).
To further investigate the role of IL-6 signal transduction in TIARP−/− mice, we also measured the expression of pSTAT-3, STAT-3, and SOCS-3 in macrophages. IL-6 stimulation induced STAT-3 and SOCS-3 expression in both groups of mice. However, only pSTAT-3 expression was enhanced in macrophages from TIARP−/− mice as compared to those from WT mice (Figure 5D). In addition, pERK-1/2 production was not changed.
In the present study, we demonstrated that TIARP−/− mice spontaneously develop enthesitis with synovitis and become susceptible to CIA. TIARP is detected during the course of adipocyte differentiation (11) and is also induced by IL-6 (12). TIARP-like proteins such as STEAP-4 are highly expressed in the bone marrow, placenta, and fetal liver (13). In murine hepatocytes, TNFα and IL-17 induce synergistic up-regulation of TIARP (14), and the TIARP gene is a direct target of phosphorylated STAT-3 (15). Recent reports suggest that the expression of 6-transmembrane protein of prostate 2 (STAMP-2)/STEAP-4 in CD14+ macrophages was significantly decreased in women with metabolic syndrome and correlated with cardiovascular risk (16). The metabolic impact of TIARP has been confirmed in adipocytes from STAMP-2−/− mice (9), but its role in immunity and inflammation remains obscure (17).
Another TNF-induced protein, tumor necrosis factor α–induced protein 3 (TNFAIP-3), has become a particular focus because of its association with RA, as shown in genome-wide association studies (18, 19). Matmati et al (20) recently revealed that mice deficient in myeloid-specific TNFAIP-3 spontaneously develop polyarthritis with increased numbers of CD11b+Gr-1+ cells and high levels of inflammatory cytokines (20). This arthritis is clearly dependent on IL-6 (and Toll-like receptor 4), but is independent of TNFα, T cells, and B cells. TIARP (TNFAIP-9) is expressed mainly in macrophages and neutrophils in mice (7) and in humans (8, 16) with arthritic conditions, and TIARP−/− mice posses increased numbers of myeloid cells and cytokine dependency similar to that in the conditional TNFAIP-3–deficient mouse model. Thus, the expression of TNF-induced proteins of myeloid origin may be important for the regulation of arthritis via similar pathways.
We found significantly increased numbers of CD11b+Gr-1low/intermediate cells in mouse spleens; thus, we focused mainly on TIARP-dominant cells, such as macrophages, in this study. Stimulation with TNFα or with LPS enhanced the expression of IL-6 in macrophages from TIARP−/− mice. Following TNFα stimulation, sustained degradation of IκBα was noted, and subsequent proliferation with dysregulated apoptosis was seen in macrophages from TIARP−/− mice. Moreover, levels of IL-6–induced pSTAT-3 were also increased in TIARP−/− macrophages, as previously demonstrated in hepatocytes (15).
In addition to macrophages, CD11b+ neutrophils are another cellular source of TIARP/STEAP-4 (7, 8). In TIARP−/− mice with CIA, overexpression of CXCL2 (the ligand of CXCR2) in joints was noted, with abundant recruitment of neutrophils. The numbers of CXCR2+ neutrophils were increased in TIARP−/− mice (data not shown), which probably enhanced the severity of the arthritis. In our previous study (8), we confirmed that STEAP-4 transfection of human neutrophils down-regulated their migration. It is possible that overproduction of IL-6 can also augment the adhesion of neutrophils (21).
CIA in the TIARP−/− mice was clearly exacerbated, and IL-6 overproduction was seen. Neutralization of IL-6 during the arthritis induction phase clearly suppressed the arthritis, suggesting a pivotal role of IL-6. The IL-6R-gp130F759 mutant mouse is a well-known model of spontaneous arthritis that develops around 12 months of age, with elevated numbers of Gr-1+CD11b+ cells in LNs (22). In addition to Th17 cells, the importance of IL-6/STAT-3 signals in type I collagen fibroblasts was recently proven in this model (23). TIARP is weakly induced in T cells under arthritic conditions; however, their expression in joint fibroblasts has been confirmed in GPI-induced arthritis (7). STEAP-4 protein was also induced by TNFα in fibroblast-like synoviocytes (FLS), and transfection of STEAP-4 into FLS blocks inflammatory cytokines, such as IL-6 and IL-8, and induces apoptosis (24). Thus, TIARP can regulate innate immune cells and fibroblasts in joints to suppress inflammation and the proliferation that results in suppression of arthritis in an IL-6–related manner.
Taken together, the findings of this study implicate TIARP as a negative regulator of the arthritogenic process through the suppression of IL-6 production, NF-κB, STAT-3 signaling, and the induction of apoptosis. Moreover, systemic deletion of TIARP results in the specific development of RA-like pathology, suggesting that treatment with TIARP or with agents that stimulate TIARP may become an important option for the treatment of 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. Matsumoto 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. Inoue, Matsumoto, Sumida.
Acquisition of data. Inoue, Matsumoto, Yoko Tanaka, Umeda, Yuki Tanaka.
Analysis and interpretation of data. Inoue, Matsumoto, Mihara, Takahashi.
Author Mihara is an employee of Chugai Pharmaceutical Company.