disease-modifying anti-rheumatic drugs
- IFN-γ+ T cells:
IFN-γ-producing human T cells
- IFN-γ– T cells:
IFN-γ-non-producing human T cells
receptor activator of NF-κB ligand
rheumatoid arthritis particle agglutination
TNF receptor-associated factor 6
Tartrate-resistant acid phosphatase
The current study explored our hypothesis that IFN-γ-producing human T cells inhibit human osteoclast formation. Activated T cells derived from human PBMC were divided into IFN-γ-producing T cells (IFN-γ+ T cells) and IFN-γ-non-producing T cells (IFN-γ– T cells). IFN-γ+ T cells were cultured with human monocytes in the presence of macrophage-CSF alone. The concentration of soluble receptor activator of NF-κB ligand (RANKL) and IFN-γ, and the amount of membrane type RANKL expressed on T cells, were measured by ELISA. In the patients with early rheumatoid arthritis (RA) treated with non-steroidal anti-inflammatory drugs alone, CD4+ T cells expressing both IFN-γ and RANKL were detected by flow cytometry. Surprisingly, IFN-γ+ T cells, but not IFN-γ– T cells, induced osteoclastogenesis from monocytes, which was completely inhibited by adding osteoprotegerin and increased by adding anti-IFN-γ antibodies. The levels of both soluble and membrane type RANKL were elevated in IFN-γ+ T cells. The ratio of CD4+ T cells expressing both IFN-γ and RANKL in total CD4+ T cells from PBMC was elevated in RA patients. Contrary to our hypothesis, IFN-γ+ human T cells induced osteoclastogenesis through the expression of RANKL, suggesting that Th1 cells play a direct role in bone resorption in Th1 dominant diseases such as RA.
We previously reported that IL-17 from activated human T cells in the synovial tissues of patients with rheumatoid arthritis (RA) is a potent stimulator of osteoclast formation via the synthesis of receptor activator of NF-κB ligand (RANKL) in osteoblasts 1. We also recently demonstrated that activated human T cells expressing RANKL directly induce osteoclastogenesis from human monocytes 2. These findings suggested that excess production of RANKL by activated T cells increases the level of soluble RANKL (sRANKL) in the synovial fluid and may contribute to osteoclastic bone resorption in RA patients. Other groups also reported that activated T cells induce osteoclastogenesis through the same mechanisms, based on results involving murine cell cultures 3, 4. On the other hand, it has been reported that interferon-γ (IFN-γ) inhibits osteoclastogenesis 5, 6. Recently, it was reported that this inhibition of osteoclastogenesis is induced by the degradation of the RANK adapter protein, TNF receptor-associated factor 6 (TRAF6) 7.
RA is a chronic inflammatory disease characterized by the destruction of articular cartilage and bone 8. The levels of several monocyte/macrophage-derived cytokines (TNF-α, IL-1, and IL-6) and T cell-derived IL-17, all of which induce osteoclast formation, are elevated in the synovial fluid of RA patients, suggesting that cytokine-mediated osteoclastogenesis occurs in the joints 1, 8–10. However, the role of T cells in the pathogenesis of RA remains controversial 9, 11–15. In addition, it has been difficult to detect T cell-derived cytokines (IL-2, IFN-γ, or IL-4) apart from IL-17 in synovial fluid by standard ELISA methods 1, 9, although RA has been shown to be a Th1 dominant disease by studies using T cell clones 16, 17.
Furthermore, several clinical studies have failed to demonstrate the efficacy of IFN-γ administration as an anti-osteoclastogenic agent to prevent bone loss 18–21. In addition, the role of IFN-γ on erosive resorption in an arthritic mouse model was controversial using anti-IFN- γ antibodies or gene depletion 22, 23. However, since the severity of inflammation may play a role in bone resorption in these patients with RA or murine models with arthritis, it seems difficult to analyze the direct effect of IFN- γ on bone resorption. On the other hand, it has been reported that long-term therapy with IFN- γ increases bone resorption in patients with osteopetrosis, a congenital osteosclerotic bone disease characterized by a defect in osteoclastic function 24, 25. In addition, IFN- γ in combination with M-CSF ameliorates the osteopetrotic condition in microphthalmic (mi/mi) mice 26. Thus, there are some discrepancies between the inhibition of osteoclastogenesis by IFN−γ in vitro and the effect of clinical administration of IFN-γ in animal models or patients with arthritis or osteopetrosis.
Although there are a series of controversial reports on the role of IFN-γ, based on the in vitro study of osteoclastogenesis, we hypothesized that IFN-γ-producing human T cells inhibit human osteoclast formation. To clarify the characteristics of T cells that induce or inhibit osteoclastogenesis, IFN-γ-producing human T cells (IFN-γ+ T cells) or IFN-γ-non-producing human T cells (IFN-γ– T cells) were cultured with human monocytes in the presence of M-CSF alone. Surprisingly, contrary to our hypothesis, IFN- γ+ T cells, but not IFN- γ– T cells, induced human osteoclastogenesis through the expression of RANKL.
Characteristics of human osteoclasts formed by sRANKL and M-CSF
All the adherent monocytes derived from PBMC were stained by anti-CD11b antibodies (Fig. 1A) and showed phagocytosing activity (Fig. 1B). Multinuclear cells formed by sRANKL and M-CSF showed tartrate-resistant acid phosphatase (TRAP) activity (Fig. 1C), ability to form resorption pits on dentine slices (Fig. 1D), CD51 expression (Fig. 1E), absence of phagocytosing activity (Fig. 1E, F), and no expression of CD11b (Fig. 1G). The multinuclear cells formed actin rings in the peripheral margin of the cells (Fig. 1H). The formation of multinuclear cells was completely inhibited by adding osteoprotegerin (OPG; 1000 ng/mL) (Fig. 2B). Therefore, the multinuclear cells showed the functions and properties of authentic osteoclasts.
IFN-γ+ human T cells induce human osteoclast formation in cultures of peripheral blood monocytes
IFN-γ+ T cells were cultured with human peripheral blood monocytes in the presence of M-CSF (100 ng/mL) alone to examine whether these T cells induce osteoclast formation. IFN-γ+ T cells (5 × 105 /well) induced a significantly larger number of osteoclasts than the medium with M-CSF alone (Fig. 2A, B). However, the number of osteoclasts formed by IFN-γ+ T cells was smaller than that in the presence of sRANKL (100 ng/mL) (Fig. 2A, B). This osteoclast formation induced by IFN-γ+ T cells was completely blocked by adding OPG (1000 ng/mL) (Fig. 2B). On the other hand, IFN-γ– T cells did not induce osteoclast formation in the presence of M-CSF (100 ng/mL) alone (data not shown).
Anti-IFN-γ antibodies increased the human osteoclastogenesis induced by IFN-γ+ human T cells
Next, we examined the effect of IFN-γ and anti-IFN-γ antibodies on osteoclastogenesis in our culture system. IFN-γ dose dependently (100–1000 pg/mL) inhibited osteoclastogenesis induced by sRANKL and M-CSF (Fig. 3A). On the other hand, 1.0 μg/mL anti-IFN-γ antibodies increased the number of osteoclasts inhibited by adding IFN-γ (100 pg/mL) (Fig. 3A). In addition, anti-IFN-γ antibodies dose dependently (1.0–10.0 μg/mL) increased the number of osteoclasts induced by IFN-γ+ T cells (Fig. 3B).
Semi-quantitative RT-PCR of IFN-γ and RANKL
We performed semi-quantitative RT-PCR to measure the levels of mRNA of β-actin, IFN-γ, and RANKL. The levels of β-actin mRNA in IFN-γ– T cells were the same as those in IFN-γ+ T cells (Fig. 4A). As expected, the levels of IFN-γ mRNA were higher in IFN-γ+ T cells than in IFN-γ– T cells (Fig. 4A). In addition, RANKL mRNA also showed higher levels in IFN-γ+ T cells than in IFN-γ– T cells (Fig. 4A).
IFN-γ was detected in the cultured supernatants of human IFN-γ+ human T cells
Around 100 pg/mL of IFN-γ was detected in the cultured supernatants of IFN-γ+ human T cells (Fig. 4B). On the other hand, IFN-γ was hardly detected in the cultured supernatants of IFN-γ– human T cells (Fig. 4B). These differences were not statistically significant due to the small number of human IFN-γ+ T cells and IFN-γ– T cells stimulated by phytohemagglutinin (PHA) and IL-2, usually, two wells of IFN-γ+ T cells and one well of IFN-γ+ T cells from 60 mL of peripheral blood of a single normal volunteer. Nevertheless, repeated experiments showed similar results.
sRANKL and membrane-type RANKL in IFN-γ+ human T cells or IFN-γ– human T cells
We then measured the concentration of sRANKL in the culture supernatants of human T cells stimulated by PHA and IL-2 for 24 h to investigate whether IFN-γ+ human T cells actually produce RANKL. The concentrations of sRANKL were higher in the culture supernatants of IFN-γ+ T cells than in the culture supernatants of IFN-γ– T cells (Fig. 4C). The levels of membrane-type RANKL were also higher in IFN-γ+ T cells than in IFN-γ– T cells (Fig. 4C). However, these differences were not statistically significant due to the small number of human IFN-γ– T cells stimulated by PHA and IL-2. Nevertheless, repeated experiments showed similar results.
Flow cytometry for RANKL, intracellular IFN-γ and IL-4 in CD3+ or CD4+ T cells from normal volunteers
We performed flow cytometry using antibodies against intracellular IFN-γ and IL-4. A significant number of CD3+ T cells or CD4+ T cells (46.1% or 38.0%, respectively) expressed IFN-γ, but not IL-4 (Fig. 4D). We then performed flow cytometry using antibodies against RANKL and intracellular IFN-γ. A fraction of CD3+ T cells or CD4+ T cells (93.6% or 96.6%, respectively) expressed RANKL (Fig. 4D). In addition, in CD3+ T cells or CD4+ T cells, almost all IFN-γ+ T cells (96.0% or 97.1%, respectively) expressed RANKL and a large fraction of RANKL+ T cells (48.8% or 39.6%, respectively) expressed IFN-γ (Fig. 4D).
Flow cytometry for RANKL and intracellular IFN-γ in CD4+ T cells from patients with RA or osteoarthritis
We performed flow cytometry for CD4+ T cells using antibodies against intracellular IFN-γ and RANKL. Representative flow cytometry results from patients with RA or osteoarthritis (OA) are shown in Fig. 5A. The ratios of CD4+ T cells expressing both IFN-γ and RANKL in a patient with RA and a patient with OA were 49.6% or 15.0%, respectively. In the total CD4+ T cells from PBMC, the ratio of CD4+ T cells expressing both IFN-γ and RANKL was significantly higher in patients with RA than those with OA (28.4 ± 9.3% vs. 21.1 ± 7.6%, p=0.0064) (Fig. 5B). The ratio of IFN-γ+RANKL+ cells in CD4+ T cells was not significantly correlated to serum C-reactive protein (CRP) levels in patients with RA (p=0.44) (Fig. 5C). The ratio of IFN-γ+IL-4– cells, but not the ratio of IFN-γ–IL-4+ cells, in CD3+ T cells or CD4+ T cells measured by two-color flow cytometry were significantly higher in patients with RA than in patients with OA (p=0.0055 and 0.0034, respectively) (Fig. 5D). There was positive correlation between the ratio of IFN-γ+RANKL+ cells in CD4+ T cells and the ratio of IFN-γ+IL-4– cells in CD4+ T cells measured by two-color flow cytometry in patients with RA (p=0.0004) (Fig. 5E).
Clinical course of a representative patient with elevated ratio of IFN-γ+RANKL+ cells in CD4+ T cells who was successfully treated by methotrexate
A 54-year-old female patient with early RA in the current study was treated with 4–6 mg/week of methotrexate (MTX) in 2003. MTX rapidly decreased the levels of CRP and RAPA (rheumatoid arthritis particle agglutination), but not the ratio of IFN-γ+RANKL+ cells in CD4+ T cells (Fig. 6).
In the present study, we demonstrated that IFN-γ+ human T cells induced osteoclast formation in the presence of M-CSF alone (Fig. 2A, B). Both sRANKL secretion by IFN-γ+ human T cells and the expression of membrane-type RANKL by IFN-γ+ human T cells were elevated (Fig. 4C). The expression of RANKL mRNA was also elevated in IFN-γ+ human T cells (Fig. 4B). Furthermore, the osteoclast formation induced by IFN-γ+ human T cells was completely blocked by adding OPG, a decoy receptor for RANKL (Fig. 2B). On the other hand, osteoclast formation induced by IFN-γ+ human T cells increased by adding anti-IFN-γ antibodies (Fig. 3B). Therefore, we concluded that IFN-γ+ human T cells directly induce osteoclast formation via the expression of RANKL. To our knowledge, the present findings represent the first demonstration that human T cells producing IFN-γ, but not IL-4, i.e., Th1 cells, directly induce human osteoclastogenesis.
Several authors have reported that IFN-γ inhibits murine osteoclast formation 5–7. In addition, it was recently reported that IFN-γ blocks mouse osteoclast formation by the degradation of TRAF6 7. In our culture system, IFN-γ (100–1000 pg/mL) also dose-dependently blocked human osteoclast formation (Fig. 3A). We demonstrated around 100 pg/mL IFN-γ in the cultured media of IFN-γ+ human T cells using ELISA (Fig. 4B) after isolating IFN-γ+ human T cells by a magnetic cell sorting method. These findings confirmed that the IFN-γ+ T cells added to co-cultures with human monocytes every 2 or 3 days produced IFN-γ. Recently, Chen et al. 27 also reported that Th1 cells preferentially express RANKL. On the other hand, in the present study we found that IFN-γ+ human T cells induced human osteoclast formation via the expression of RANKL. Thus, we speculate that IFN-γ+ human T cells induced human osteoclast formation, since the inductive effects of RANKL exceeded the inhibitory effects of IFN-γ. In other words, IFN-γ+ human T cells induce human osteoclast formation at the cellular level, although IFN-γ inhibits osteoclast formation through the degradation of TRAF6 at the molecular level.
In the present study, we demonstrated that IFN-γ+ human T cells expressed RANKL. We have previously demonstrated that activated CD4+ T cells in the synovial tissue of RA patients express RANKL 2. It was recently reported that immunostimulatory DNA sequences induce both Th1 response and RANKL production in rats with adjuvant arthritis 28. It was also reported that murine Th1 cells preferentially express RANKL 27. In addition, in studies using clonal cells, several authors reported that Th1 cells are dominant in the pathogenesis of RA 16, 17. Thus, in RA patients, Th1 cells may directly play roles not only in the immune system but also in inducing osteoclastogenesis. It is, however, difficult to detect IFN-γ in synovial fluids or synovial tissues by standard ELISA or immunohistological staining methods 9. Thus, Th1 cells expressing RANKL may play a role in the peripheral blood. In fact, we demonstrated that the ratio of CD4+ T cells expressing both IFN-γ and RANKL in the total CD4+ T cells from PBMC was significantly elevated in patients with early RA treated with non-steroidal anti-inflammatory drugs alone and were not receiving any disease-modifying anti-rheumatic drugs (DMARDs) or prednisone (Fig. 5A, B).
In the current study, we demonstrated that the ratio of IFN-γ+RANKL+ cells in CD4+ T cells was not positively correlated to serum CRP levels in patients with RA (p=0.44) (Fig. 5C). In some patients with RA, joint destruction could progress even without inflammation in joints and with normal level of serum CRP. Recently, it has been reported that small amount of RANKL synergistically induce osteoclastogenesis with small amount of TNF-α 29, 30. Thus, in such RA patients, small amount of sRANKL derived from IFN-γ+RANKL+ CD4+ T cells may play an important role in the osteoclastogenesis in affected joints that seem to be without inflammation.
In addition, we demonstrated the ratio of IFN-γ+IL-4– cells in CD4+ T cells (Th1 cells) was elevated in patients with RA and that the ratio of IFN-γ+RANKL+ cells in CD4+ T cells was positively correlated to the ratio of IFN-γ+IL-4– cells in CD4+ T cells (Th1 cells) measured by two-color flow cytometry in patients with RA (Fig. 5E). It has been reported that MTX decreases Th1 cytokines (IL-2 and IFN-γ) gene expression in PBMC from untreated active RA patients 31. Thus, to prevent bone resorption, it may be necessary to reduce the ratio of Th1 cells, including IFN-γ+RANKL+ CD4+ T cells in peripheral blood in patients with RA. In fact, more recently three groups demonstrated the role of T cells in bone resorption in animal models or healthy volunteers; Ogawa et al. 32 demonstrated that induction of apoptosis in activated CD4+ T cells suppressed osteoclastogenesis in RA using their SCID mice model. Valverde et al. 33 reported that a potassium channel blocker reduces inflammatory bone resorption in rats with T cell-mediated periodontal disease, decreasing Th1 clone activation. Moreover, Di Monaco et al. 34 showed that a positive association between total lymphocyte count and bone mineral density in 114 healthy postmenopausal women. In our representative RA patient treated with 6 mg/week MTX, the ratio of IFN-γ+RANKL+ cells in CD4+ T cells did not decrease even after CRP levels and RAPA titer decreased (Fig. 6). Thus, to prevent joint destruction, we should use a dose of MTX sufficient to reduce the ratio of IFN-γ+RANKL+ CD4+ T cells in peripheral blood, since incomplete reduction of amount of IFN-γ+RANKL+ CD4+ T cells may explain insufficient effect of MTX on joint destruction in some patients with RA. We are further investigating the effects of anti-TNF-α therapy and DMARDs including MTX on the ratio of IFN-γ+RANKL+ cells in CD4+ T cells in patients with RA.
More recently, Huang et al. 35 have reported that exposure to RANKL renders preosteoclasts resistant to IFN-γ by inducing terminal differentiation. In addition, these findings suggest that in inflammatory bone loss early exposure to RANKL primes osteoclast precursors to form in the presence of a high level of IFN-γ using mechanisms independent of the signal molecules STAT1 and TRAF6 35. In the present study, we demonstrated that human T cells producing both IFN-γ and RANKL induce human osteoclast formation from monocytes as osteoclast precursors. Thus, it is speculated that the T cells producing both IFN-γ and RANKL may stimulate mononuclear preosteoclasts to differentiate to mature multinuclear osteoclasts or may activate mature osteoclasts to resorb bone tissue, since mononuclear preosteoclasts or mature osteoclasts are resistant to IFN-γ.
We detected lower levels of RANKL protein and mRNA in IFN-γ– T cells compared with those in IFN-γ+ T cells in in vitro (Fig. 4A, C). However, we detected a significant number of RANKL+ T cells in IFN-γ– T cells from PBMC of normal volunteers (Fig. 4D) and RA and OA patients (Fig. 5A). The reason for this discrepancy remains unclear. However, the discrepancy may be due to the different methods for stimulating T cells; T cells were stimulated with PHA and IL-2 for RT-PCR and ELISA, whereas T cells were stimulated with PMA and ionomycin for flow cytometry. We are now investigating which cytokines RANKL+ T cells in IFN-γ– T cells produce.
In the current study, we demonstrated that activated human T cells by PHA and IL-2 induced human osteoclastogenesis through the expression of RANKL. On the other hand, it has been reported that activated murine T cells by CD3ϵ antibody alone inhibit osteoclastogenesis 7. To explain this discrepancy, Wyzga et al.36 have recently demonstrated that the effect of activated T cells on osteoclastogenesis are complex and depend on the method of T cell activation using mouse T cells. In addition, Kanamaru et al.37 have demonstrated that the membrane-bound RANKL expression on human T cells is strictly limited, and the majority of RANKL protein produced by human T cells may be active in the soluble form after shedding. Consistently, we have previously reported the low proportion of membrane-bound RANKL+ cells in activated human T cells from peripheral blood 2. In contrast to human T cells, murine T cells have been demonstrated to express much higher membrane-bound RANKL as assessed by the binding of the RANKL-Fc fusion protein 38. From these findings, Kanamaru et al. 37 speculated that the differential regulation of membrane RANKL between human and mice may be attributed to the differential regulation of ectodomain shedding and of RANKL isoforms in the transcriptional levels. Thus, this difference in the amount of soluble form of RANKL from activated T cells between human and mice may explain the discrepancy in the osteoclastogenesis by activated T cells between human 2 and mice 7. In human, sRANKL from activated T cells may play an important role in osteoclastogenesis.
In summary, contrary to our initial hypothesis, the present findings demonstrated that IFN-γ+ human T cells induce human osteoclastogenesis via the expression of RANKL. The present findings also suggest that Th1 cells may play a direct role in bone resorption in patients with Th1 dominant diseases such as RA. Modulation of the RANKL, especially sRANKL, expression by human T cells may represent a useful approach in the development of new treatment strategies to inhibit bone destruction.
Materials and methods
Recombinant human sRANKL and human OPG were obtained from Pepro Tech EC Ltd. (London, UK) and R&D Systems, Inc. (Minneapolis, MN), respectively. Recombinant human M-CSF (Leukoprol®) was obtained from Yoshitomi Pharmaceutical (Osaka, Japan), recombinant IL-2 (TGP-3) was from Takeda Co. (Tokyo, Japan), and recombinant human IFN-γ was from R&D Systems. Anti-human IFN-γ antibodies were obtained from R&D Systems.
PBMC for flow cytometry, ELISA, RT-PCR, and co-culture inducing osteoclastogenesis were obtained from five healthy normal volunteers. PBMC for flow cytometry were obtained from patients with RA or OA. Fifteen female patients with RA (54.3 ± 12.6 years, range 27–76 years, median 57 years) with active arthritis who fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for RA (1987) were studied. The mean of disease duration of RA patients was 10. 1 ± 3.8 months (range 1.5–55 months, median 3 months). The disease duration of 12 of 15 patients with RA was less than 9 months and that of the other three patients was 24, 34, and 55 months, respectively. The mean of CRP levels of RA patients was 2.54 ± 0.82 mg/100 mL (range 0.1–12.3 mg/100 mL, median 0.9 mg/100 mL). All patients with RA were treated with non-steroidal anti-inflammatory drugs alone and were not receiving any DMARDs or prednisone when we obtained PBMC from these patients. These patients were treated with DMARDs after we obtained PBMC. Control individuals were 11 female patients with OA (55.7 ± 7.0 years, range 42–64 years, median 58 years). Age did not significantly differ between patients with RA and those with OA. Informed consent for subsequent procedures was obtained from all normal volunteers and all patients.
Determination of osteoclast characteristics
Adherent multinuclear cells were fixed and stained for vitronectin receptors. For immunohistochemical determination of osteoclast characteristics, cells were fixed and incubated with a monoclonal antibody against vitronectin receptors αv (CD51/61) (Becton Dickinson, NJ) or a monoclonal antibody against anti-CD11b (DAKO Cytomation, Glostrup, Denmark). The bound antibodies were visualized as described previously 8. TRAP activity was detected by acid phosphatase staining in the presence of tartrate as previously reported 1. Pit formation assay was also performed as previously described 1. Phagocytosing activity was detected by phagocytes of fluoresbrite YG microspheres® (Polysciences, Inc., Warrington, PA) after incubation of osteoclasts with the microspheres for 1 h. Actin rings in matured osteoclasts were stained by rhodamine phalloidin (Molecular Probes, Inc., Eugene, OR) and observed using fluorescence microscopy.
We measured the amount of mRNA of RANKL, IFN-γ and β-actin using RT-PCR 39. β-Actin mRNA was detected as a control. Total RNA was prepared from synovial specimens using the Tri ReagentTM method (Molecular Research Center, Inc. Cincinnati, OH), and complementary DNA (cDNA) was synthesized from mRNA by priming total RNA. We designed a sense primer, (S1) and an antisense primer (AS1) for RANKL cDNA, as follows: S1, 5′-AGACACAACTCTGGAGAGTCAAG-3′; AS1, 5′-TACGCGTGTTCTCTACAAGGTC-3′. Primers for IFN-γ (o-S and o-AS) and β-actin were previously reported 18. The product sizes corresponding to cDNA of RANKL and IFN-γ were 815 bp and 356 bp, respectively.
Co-culture system for osteoclastogenesis
Human peripheral blood was collected from five healthy normal volunteers in syringes containing 1000 U/mL preservative-free heparin. Written informed consent was obtained in all cases before blood aspiration. PBMC were isolated by centrifugation over Histopaque 1077 (Sigma Chemicals Co., St. Louis, MO) density gradients, washed and resuspended at 1.3 × 106 cells/mL in α-minimal essential medium (α-MEM) (GIBCO BRL, Gaithersburg, MD) supplemented with 10% fetal bovine serum (FBS) (JRH Biosciences, Lenexa, KS). PBMC were then cultured for 3 days in 48-well plates (Corning Glass Inc., Corning, NY) (5 × 105 cells/0.3 mL/well) in the presence of M-CSF (100 ng/mL). Nonadherent cells were then removed before adding activated IFN-γ+ T cells, IFN-γ– T cells, IFN-γ, or IL-4 to the wells. The adherent PBMC, as described above, were cultured with IFN-γ+ T cells (103, 3 × 103, 104/well), IFN-γ– T cells (103, 3 × 103, 104/well) or IFN-γ (1–1000 pg/mL) with or without anti-IFN-γ antibodies (1.0 μg/mL) in the presence of M-CSF (100 ng/mL) alone or human sRANKL (100 ng/mL) and M-CSF (100 ng/mL). We used the adherent PBMC as monocytes in the co-culture system. Cultures were incubated in quadruplicate. Culture medium and activated T cells were replaced every 2 or 3 days with fresh media supplemented, respectively, with the agents described above and with activated T cells, respectively. Osteoclast formation was evaluated by immunohistological staining for vitronectin receptors and the absence of phagocytic activity after culturing for 7 or 8 days.
Isolation of IFN-γ+ T cells and IFN-γ– T cells
CD3+ cells were separated from PBMC obtained from the same volunteer by magnetic cell sorting (Miltenyi Biotech, Auburn, CA). The separated CD3+ cells (106/well) were cultured and activated in α-MEM supplemented with 10% FBS in the presence of PHA (800:1) and IL-2 (0.5 ng/mL) in 24-well plates for 24 h. Then, IFN-γ+ T cells and IFN-γ– T cells were separated by magnetic cell sorting. The ratio of IFN-γ+ T cells to IFN-γ– T cells was about 3:1. RNA was extracted from activated T cells (1 × 106) for RT-PCR.
Concentrations of soluble and membrane type RANKL in synovial fluids
Concentrations of sRANKL were measured using an ELISA we devised 40. Fully bioactive recombinant human sRANKL and anti-human sRANKL (rabbit) were obtained from Pepro Tech. Recombinant human OPG/Fc chimera was obtained from R&D Systems. In brief, 50 μL/well of the captured OPG/Fc (1 μg/mL PBS) was transferred to an ELISA plate (Nunc, Denmark, 96 wells, flat bottom) and incubated overnight at 4°C. Each well was then aspirated, and the plate was blocked by adding 250 μL PBS containing 1% bovine serum albumin (buffer-1) to each well. Incubation was carried out at room temperature for a minimum of 1 h. The sample or standards (sRANKL) in an appropriate diluent were incubated at 37°C with vigorous shaking in a final volume of 50 μL buffer-1. After 1 h the reaction medium was removed by aspiration, each well was washed three times with 1 mL chilled washing buffer (0.05% Tween 20 in PBS), and 50 mL anti-RANKL IgG (100 ng/mL buffer-1) was added to each well and incubated for 1 h at 37°C. After washing, 50 μL/well of an anti-rabbit IgG conjugated with peroxidase (Bio-Rad, 1:1000 of original solution) was added to each well, followed by incubation for 1 h at 37°C. The bound enzyme protein was assayed with tetramethylbenzidine/H2O2 as a substrate, and stopped by adding 50 μL/well 1 M phosphoric acid. The measurement was conducted at 450 nm using Biolumin 960 (Amersham Biosciences, Sunnyvale, CA). Each experiment was repeated three times. The working range of the assay was between 0.1 and 10 ng/mL for human sRANKL. The detection limit of the assay (2× blank) was 0.05 ng/mL for human sRANKL. The coefficients of variation for the intraassay were less than 11%. Membrane type RANKL was measured using the same method after sonication of T cells with 50 mM sodium phosphate buffer (pH 7.5), and a cocktail consisting of 0.1% Triton X-100, 0.1 mM PMSF, and a 2% protease inhibitor (Sigma). The amount of protein was measured using a Bio-Rad DC protein assay kit (Bio-Rad, Hercules, CA).
Flow cytometry analysis for intracellular IFN-γ and IL-4
After separating IFN-γ+ T cells and IFN-γ– T cells, these T cells were stimulated with 25 ng/mL PMA (Sigma) and 2 μg/mL ionomycin (Sigma) in the presence of 10 mg/mL brefeldin-A (BFA, Sigma) for 4 h at 37°C in 7% CO2. T cells (400 μL) were incubated with 2 mL of 1× FACS lysing solution (Becton Dickinson, Mountain View, CA) for 10 min at room temperature. T cells were washed and incubated with 500 μL of 1× FACS permeabilizing solution (Becton Dickinson) for 10 min at room temperature. T cells were washed again and further incubated with FITC-conjugated anti-human IFN-γ antibodies (Becton Dickinson) and PE-conjugated anti-human IL-4 antibodies (Becton Dickinson) for 30 min at room temperature in the dark. The stained cells were analyzed using FACScan (Becton Dickinson).
Flow cytometry analysis for intracellular IFN-γ and RANKL
T cells were also stimulated with 25 ng/mL PMA and 2 μg/mL ionomycin in the presence of 10 mg/mL BFA for 4 h at 37°C in 7% CO2. T cells (400 μL) were incubated with 2 mL of 1× FACS lysing solution for 10 min at room temperature. T cells were washed and incubated with 500 μL of 1× FACS permeabilizing solution for 10 min at room temperature. T cells were washed again and further incubated with FITC-conjugated anti-human IFN-γ antibodies (Becton Dickinson) and mouse anti-human RANKL antibodies (kindly provided by Sankyo, Inc. Tokyo, Japan) with R-phycoerythrin-indodicarbocyanine (RPE-CY5)-conjugated anti-mouse antibody for 30 min at room temperature in the dark. The stained cells were analyzed using a FACScan.
Detection of IFN-γ in the culture supernatants of T cells by ELISA
After separating IFN-γ+ T cells and IFN-γ– T cells, these T cells (106/well) were cultured in α-MEM supplemented with 10% FBS in the presence of PHA (800:1) and IL-2 (0.5 ng/mL) in 24-well plates for 24 h. The concentration of IFN-γ in the culture supernatants of IFN-γ+ T cells or IFN-γ– T cells was measured by ELISA (Bender Med Systems, Vienna, Austria).
The data were analyzed using the Mann-Whitney test and Spearman's rank correlation test (StatView®, Abacus Concepts Inc. Berkeley, CA). A significant difference was defined as p<0.01. All values are presented as mean ± SD. All values in figures were represented as box plots, in which the upper and the lower bars show the 90th and 10th percentiles, respectively, while the upper, center and lower lines of the box show the 75th, 50th and 25th percentiles, respectively.
This work was supported in part by a grant-in-aid from the Ministry of Education, Science, Sports and Culture of Japan. We thank Ms. N. Akiyama and Ms. H. Kikuchi (Tokyo Women's Medical University) for her valuable technical assistance. We would also like to thank Ms. M. Ebisawa (SRL) for performing the flow cytometry.