The (-)-enantiomer of DHMEQ, which is simply represented as DHMEQ in this manuscript, is a more potent inhibitor of NF-κB than its (+)-enantiomer, and was synthesized as described previously 19.

Animal experiments were approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University. Male 8-week-old DBA/1J mice were purchased from Oriental Yeast (Tokyo, Japan). Bovine collagen type II (Collagen Research Center, Tokyo, Japan) was dissolved in 50 mM acetic acid at 4 mg/ml and was emulsified in an equal volume of Freund's complete adjuvant (Difco Laboratories, Detroit, MI, USA). Mice were immunized with 100 μl emulsion intradermally at the base of the tail. After 21 days (day 0), the same amount of the antigen emulsified in the same adjuvant was intradermally injected at the base of the tail as a booster immunization. From day 0 to day 10, 100 μg DHMEQ (5 mg/kg body weight) dissolved in 50 μl dimethyl sulfoxide was injected subcutaneously every day to mice in the experimental group. Mice in the control group received 50 μl dimethyl sulfoxide similarly injected.

The thickness of each hind paw was measured on day -4 and on day 10 using a pair of digital slide calipers. Radiographs of both ankle joints were obtained on day 10 and were scored on a scale of 1–3 (1 = no change, 2 = mild osteoporosis without bone erosion, 3 = severe osteoporosis with or without bone erosion) by three investigators who were blinded to the assignment of mouse groups.

Mice were sacrificed on day 10, 3 hours after the last injection, and their hind paws were excised for experiments. After the skin was scarified with a surgical blade, the left hind paws were preserved in 10% buffered formalin for 3 hours, and the skin was totally removed. The paws were then decalcified in 10% ethylenediamine tetraacetic acid, 5% polyvinylpyrolidone, 100 mM Tris (pH 7.4) for 4 weeks, were dehydrated in graded ethanol, were permeated serially by methyl benzoate and benzene, and were embedded into paraffin in a vacuum oven. Longitudinally sectioned paraffin blocks were fixed in citrate-acetone (2:3 mixture of 380 mM citrate and acetone) for 30 seconds, and were stained with 0.5 mg/ml naphthol AS-BI phosphoric acid and 0.3 mg/ml fast red violet LB salt (Sigma-Aldrich, St Louis, MO, USA) in 27 mM sodium tartrate and 100 mM sodium acetate (pH 5.2) for 1 hour at 37°C. Nuclei were stained with hematoxylin. The mean number of TRAP-positive giant cells with four or more nuclei in the individual ankle joints of arthritic mice was counted under a microscope by two investigators in a manner blinded to the assignment of mouse groups.

The right hind paws were frozen in liquid nitrogen, and bone-containing sections were prepared using Cryofilm (Finetec, Tokyo, Japan) 20. Phycoerythrin-labeled anti-NFATc1 monoclonal antibody 7A6 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Monoclonal antibody to the nuclear localization signal in the p65 subunit of NF-κB was purchased from Chemicon (Temecula, CA, USA), was labeled with FITC (Wako, Tokyo, Japan), and was dialyzed against 100 mM Tris, 200 mM NaCl (pH 7.4). After fixing with acetone for 10 min, permiabilization with PBS containing 0.5% Triton X-100 for 10 minutes, and blocking with PBS containing 10% FCS for 1 hour, the sections were incubated with a mixture of the two antibodies for 1 hour, and observed under a confocal microscope (Olympus, Tokyo, Japan).

Serum samples were obtained at euthanasia and the concentrations of soluble receptor activator of NF-κB ligand (sRANKL), osteoprotegerin(OPG) and M-CSF were measured by ELISA. The ELISA kits for sRANKL and OPG were purchased from Biomedica (Vienna, Austria), and the ELISA for M-CSF was from R&D Systems (Minneapolis, MN, USA).

Synovial tissues were obtained at the time of total knee joint replacement from six patients with RA; these patients were female, aged (mean ± standard deviation) 67.3 ± 9.1 years, and their serum C-reactive protein levels were 3.4 ± 2.4 mg/dl. Of the six RA patients, five women took prednisolone, four women took methotrexate, two women took bucillamine, and one woman took leflunomide. Signed consent forms were obtained prior to the operation, and the experimental protocol was approved in advance by the Ethics Committee of Tokyo Medical and Dental University. RA was diagnosed according to the criteria of the American College of Rheumatology 21.

RA fibroblast-like synovial cells (FLS) were prepared from the synovial tissues as described previously 22, and were cultured in DMEM with heat-inactivated 10% FCS (Sigma-Aldrich). After incubation with or without DHMEQ for 24 h, RA-FLS were collected and lysed with RIPA lysis buffer (Upstate, Lake Placid, NY, USA). After debris was eliminated by centrifugation, 5 μg proteins in the supernatant were separated by 10% SDS-PAGE under reducing conditions, and were transferred to a polyvinylidene difluoride membrane. The membranes were blocked with 4% Block Ace (Snow Brand Milk Products, Sapporo, Japan) in PBS containing 0.1% Tween 20 overnight, were incubated with anti-RANKL monoclonal antibody 70513 (R&D Systems) or with anti-β-actin monoclonal antibody AC-15 (Sigma-Aldrich) in PBS containing 0.1% Tween 20 with 0.4% Block Ace for 1 hour, and were then incubated with peroxidase-conjugated rabbit anti-mouse IgG (Dako Cytomation, Carpinteria, CA, USA) for 1 hour. The immunoblots were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ, USA). To analyze M-CSF production by RA-FLS, cells (2 × 10⁴/well) were cultured in 96-well plates with or without DHMEQ for 24 hours. The culture supernatants were collected and the concentration of M-CSF was measured using an ELISA kit (BioSource, Camarillo, CA, USA).

Peripheral blood mononuclear cells from healthy donors were collected by Ficoll-Conray gradient centrifugation, and monocytes were positively selected using MACS microbeads (Miltenyi Biotec, Auburn, CA, USA). The monocytes (5 × 10⁴/well) were incubated in 96-well plates in αMEM with heat-inactivated 10% FCS (Sigma-Aldrich), 25 ng/ml M-CSF (Peprotech, Rocky Hill, NJ, USA) and 40 ng/ml RANKL (Peprotech). The indicated concentration of DHMEQ was added throughout the culture period. On day 3 the medium was replaced with fresh medium. After incubation for a further 4 days, the number of TRAP-positive multinucleated cells with three or more nuclei was counted under a microscope. To analyze the expression of matrix metalloprotease-9 (MMP-9), osteoclasts were differentiated as above without DHMEQ; then DHMEQ was added after medium replacement at day 7, and the culture supernatant was collected at day 8. The concentration of MMP-9 in the supernatant was measured using an ELISA kit (GE Healthcare Bio-Sciences, Tokyo, Japan).

We first confirmed the effect of DHMEQ on collagen-induced arthritis by comparing the paw thickness and radiographic changes in the mice treated with DHMEQ (n = 12) and vehicle alone (n = 13), as well as in nonimmunized age-matched normal mice (n = 6), 10 days after booster immunization. As shown in Figure 1, treatment with DHMEQ ameliorated both inflammation and bone destruction.

To examine the effect of DHMEQ on in vivo differentiation of osteoclasts, the ankle joints of the mice were excised and processed for histochemical staining. The specimens from arthritic mice treated with vehicle alone showed marked synovitis accompanying invasion of pannus into the marrow space (Figure 2d). Numerous TRAP-positive cells were attached on the eroded bone surface and the inner surfaces of bone lacunae, and some of them were multinucleated (Figure 2g). Radiographically, the ankles of these mice showed remarkable periarticular osteoporosis and bone erosion (Figure 2a). In contrast, the joints of arthritic mice treated with DHMEQ showed milder synovial inflammation. Osteoclasts were mainly observed on the inner surfaces of the bone marrow, and their number and size were less than those in vehicle-treated mice (Figure 2e, h). Radiographs showed mild periarticular osteoporosis (Figure 2b). In ankle joints of normal control mice, virtually no TRAP-positive cells were observed (Figure 2f, i).

For quantitative evaluation, the number of TRAP-positive giant cells with four or more nuclei in each ankle joint was counted. As shown in Figure 3, the mice treated with DHMEQ exhibited significantly fewer osteoclasts than those given vehicle alone, indicating the suppressive effect of DHMEQ on in vivo osteoclastogenesis.

Many in vitro studies adopt a culture system in which monocyte/macrophage precursor cells are stimulated with sRANKL and M-CSF for induction of osteoclasts. In the in vivo bone metabolism, the naturally occurring decoy receptor OPG also plays a key role by preventing the binding of RANKL to its receptor, receptor activator of NF-κB (RANK). The ratio of circulating OPG to sRANKL in early RA patients has been demonstrated to predict later joint destruction 23. We therefore tested whether administration of DHMEQ changed expression of these soluble factors. As shown in Figure 4a, serum levels of OPG in arthritic mice treated with DHMEQ and with vehicle alone were both significantly higher than those of normal mice. sRANKL in both arthritic groups also tended to be higher than nonarthritic normal mice, although this was not statistically significant (Figure 4b). No differences in OPG, sRANKL or the sRANKL/OPG ratio were observed, however, between the DHMEQ-treated and vehicle-treated groups (Figure 4a–c). In addition, no significant difference was observed in the serum levels of M-CSF among three groups (Figure 4d).

To further examine the effect of DHMEQ on expression of RANKL and M-CSF, we carried out in vitro experiments using human RA-FLS. As shown in the results of western blotting, RA-FLS spontaneously expressed RANKL without the addition of proinflammatory cytokines, and incubation with DHMEQ did not change the level of RANKL expression (Figure 5a). Similarly, the result of ELISA revealed that RA-FLS secreted M-CSF without any stimulation. Incubation with DHMEQ did not suppress the levels of M-CSF, but rather enhanced it slightly at 3 μg/ml (Figure 5b). Stimulation with TNFα did not further increase the production of RANKL or M-CSF by RA-FLS (data not shown). These results suggest that production of RANKL and M-CSF by proliferating RA-FLS are not particularly dependent on NF-κB, and the suppressive effect of DHMEQ on osteoclastogenesis resulted from the downregulation of proosteoclastogenic factors other than RANKL, RANK or OPG.

In the presence of RANKL and M-CSF, DHMEQ inhibits differentiation of osteoclasts in cultures of mouse bone-marrow-derived monocyte/macrophage precursor cells by downregulation of NFATc1 12. We therefore examined the expression of NFATc1 as well as NF-κB in the joints of arthritic mice by immunofluorescent staining. Using monoclonal antibody that recognizes only an activated form of the p65 subunit of NF-κB, distinct staining was observed along the inner surface of bone lacunae (Figure 6a) and in eroded regions of arthritic bone from mice treated with vehicle alone, but was not observed in those mice treated with DHMEQ (Figure 6d). Staining of NFATc1 was also obvious on the inner surfaces of bone lacunae (Figure 6b) and in the eroded regions of vehicle-treated mice, but not from DHMEQ-treated mice (Figure 6e). Normal control mice exhibited no staining of NF-κB (Figure 6g) nor of NFATc1 (Figure 6h). These results suggest that inhibition of NF-κB activation by DHMEQ leads to suppression of NF-κB-dependent expression of NFATc1 by osteoclasts in arthritic joints.

Human peripheral blood monocytes cultured with M-CSF and RANKL differentiate into osteoclasts 24. To test whether the suppressive effect of DHMEQ on osteoclastogenesis can be applied to human cells, monocytes from peripheral blood of healthy volunteers were cultured with DHMEQ together with M-CSF and RANKL. The result showed that the number of TRAP-positive multinucleated cells was decreased by incubation with DHMEQ in a dose-dependent manner (Figure 7a).

MMP-9 is one of the enzymes released by osteoclasts, and the enzyme plays a role in degradation of the extracellular matrix. Its expression is reported to be modulated by NFATc1 25, and to be upregulated in serum of patients with active RA 26. To examine the effect of DHMEQ on MMP-9 production by human osteoclasts, DHMEQ was added to the culture after formation of mature osteoclasts and secreted MMP-9 was measured. The results showed that concentration of MMP-9 in the culture supernatant was partially but significantly decreased by DHMEQ (Figure 7b). These results indicate that DHMEQ suppresses osteoclast differentiation from human peripheral blood monocytes as well as the activity of mature osteoclasts.