Introduction
In the past decade, a series of potent new biologic therapeutics have demonstrated remarkable clinical efficacy in several autoimmune diseases, including rheumatoid arthritis (RA). In the case of RA, a chronic progressive autoimmune disease that targets joints and occurs in approximately 0.5 to 1% of adults, biologic agents, such as TNF inhibitors, have proven effective in patients not responding to disease-modifying anti-rheumatic drugs, such as methotrexate. However, about 30% of patients treated with a TNF inhibitor are primary non-responders. Moreover, a substantial proportion of patients experience a loss of efficacy after a primary response to a TNF inhibitor (secondary non-responders) [
1‐
3]. More recently, as new therapies have become available, including biological agents targeting IL-6, B cells and T cells, it has become clear that a notable proportion of patients respond to these new biological agents even among primary and secondary non-responders to TNF inhibitors [
3‐
10]. These individual differences in response to each agent highlight the difficulty and limit of treating multifactorial disease by targeting single cytokine or single cell type. Patient-tailored therapy might be able to overcome this issue, but good biomarkers to predict treatment responses have not yet been elucidated.
Therefore, as described above, biological drugs have limited values. In addition, such drugs may be accompanied by serious side effects [
11,
12]. Furthermore, the high cost of these biological drugs may make access to these reagents prohibitive for the general public. Alternative therapeutic options, such as small molecule-based drugs, continue to be an important challenge.
The involvement of prostaglandin pathways in the pathogenesis of arthritis has been shown in animal models by using mice lacking genes, such as cycolooxygenase-2 (COX-2), prostaglandin E synthase, or prostacyclin receptor [
13‐
15]. As COX-2 knockout mice normally develop autoreactive T cells in collagen-induced arthritis (CIA) [
13], prostaglandin pathways appear to be involved mainly in the effector phase of arthritis. However, treatment with celecoxib, a prototype drug belonging to a new generation of highly specific COX-2 inhibitors has been reported to have only mild suppressive effects on animal models of arthritis, and strong inhibition of arthritis was achieved only when mice were treated in the combination of celecoxib with leukotriene inhibitors [
16‐
19]. In humans, although celecoxib is widely used as an analgesic agent in patients with RA or osteoarthritis, there is no evidence that celecoxib therapy modulates the clinical course of RA. In addition, recently it has been shown that celecoxib enhances TNFα production by RA synovial membrane cultures and human monocytes [
20].
Celecoxib has been reported to exhibit COX-2-independent effects, such as tumor growth inhibition and immunomodulation [
21,
22]. Previously, we demonstrated that celecoxib treatment suppressed experimental autoimmune encephalomyelitis (EAE) in a COX-2 independent manner [
22]. We recently developed a trifluoromethyl analogue of celecoxib (TFM-C; full name: 4-[5-(4-trifluoromethylphenyl)-3-(trifluoromet-hyl)-1
H-pyrazol-1-yl]benzenesulfonamide), with 205-fold lower COX-2-inhibitory activity. In studies using recombinant cell lines, TFM-C inhibited secretion of the IL-12 family cytokines, IL-12, p80 and IL-23, through a COX-2-independent, Ca
2+-dependent mechanism involving chaperone-mediated cytokine retention in the endoplasmic reticulum coupled to degradation via the ER stress protein HERP [
23,
24]. In the present study, we demonstrate that TFM-C inhibits innate immune cells and animal models of arthritis, including CIA and type II collagen antibody-induced arthritis (CAIA), in contrast to the limited inhibitory effect of celecoxib. TFM-C suppresses the activation of mast cells in arthritic joints. Moreover, TFM-C treatment suppresses the production of inflammatory cytokines by macrophages and leukocyte recruitment. These findings indicate that TFM-C may serve as an effective new drug for the treatment of arthritis, including RA.
Materials and methods
Differentiation and stimulation of U937 cells
Human U937 cells were obtained from the American Type Culture Collection (Rockville, MD, USA) and cultured in RPMI 1640 supplemented with 10% FCS. To differentiate U937 cells, 5 × 105 cells were treated with PMA (25 ng/ml) for 24 hours. At 22 hours of PMA treatment, 50 μM of TFM-C was added for 2 hours. Subsequently, cells were stimulated with 5 μg/ml of LPS and PMA (25 ng/ml) for 0, 3, 6, 12 and 24 hours in the presence or absence of TFM-C. Supernatants were harvested and assayed for cytokine production by means of Quansys Q-Plex™ Array (Quansys Bioscience, Logan, Utah, USA). RNA isolation was performed following the manufacturer's instructions (Macherey-Nagel, Düren, Germany).
Quantitative RT-PCR (qPCR)
A total of 200 ng of RNA extracted from U937 cells was retrotranscribed to cDNA using random primers according to the manufacturer's protocol (Applied Biosystems, Carlsbad, California, USA). qPCR was performed with the Supermix for SsoFast EvaGreen (Biorad, Hercules, California, USA) on a 7500 Fast Real-Time PCR System (Applied Biosystems). For each target gene, qPCR QuantiTect Primer Assays were used (Qiagen Hilden, Germany). For each sample, expression levels of the transcripts of interest were compared to that of endogenous GAPDH. The levels of mRNA are calculated as 2-Ct.
Quansys Q-Plex™ Array chemiluminescent
A total of 30 μl of medium from differentiated U937 cells treated with PMA/LPS/TFM-C or LPS/PMA were analyzed. Human Cytokine Stripwells (16-plex) were used following the manufacturer's instructions. The image was acquired using Bio-Rad Chemidoc camera and analyzed with Q-View Software (Quansys Bioscience, Logan, Utah, USA)
DAPI staining
Differentiated U937s were treated with LPS/PMA/TFM-C for 6, 12 and 24 hours and then fixed with 2% PFA. The cells were washed three times with PBS and then incubated with DAPI (1:50000; Molecular Probes, Carlsbad, California, USA) in PBS. Coverslips were embedded in Fluoro-Gel (Electron Microscopy Science, Hatfield, Pennsylvania, USA). Images were recorded using the ApoTome system (AxioVision, Carl Zeiss, Inc., Oberkochen, Germany) and analyzed using the ImageJ program (version 1.40, Bethesda, Maryland, USA).
AlarmBlue staining of U937 cells
The number of viable cells was tested at 6, 12, and 24 hours after TFM-C exposure by adding the AlamarBlue reagent (AbD Serotec, Cambridge, UK). Absorbance was measured at wavelengths of 570 nm and 600 nm after required incubation, using a Varioskan Flash (Thermo Fisher Scientific, Fremont, CA, USA). Absorbance values of samples were normalized with values of the cell culture media without cells. The results are presented as the proportion of viable cells, calculated by dividing the absorbance values of drug-treated samples by the absorbance values of untreated control samples.
Mice
DBA1/J mice were purchased from Oriental Yeast Co., Ltd. (Tokyo, Japan). C57BL/6J (B6) mice were purchased from CLEA Laboratory Animal Corp. (Tokyo, Japan). Animal care and use were in accordance with institutional guidelines and all animal experiments were approved by the Institutional Animal Care and Use Committee of the National Institute of Neuroscience.
Induction of CIA
DBA1/J male mice (n = 5 to 6 per group, 7 to 8 weeks old) were immunized intradermally at the base of the tail with 150 μg of bovine type II collagen (CII) (Collagen Research Center, Tokyo, Japan) emulsified with an equal volume of complete Freund's adjuvant (CFA), containing 250 μg of H37Ra Mycobacterium tuberculosis (Mtb) (Difco, Detroit, MI, USA). DBA1/J mice were boosted 21 days after immunization by intradermal injection with 150 μg of CII emulsified with incomplete Freund's adjuvant (IFA).
Induction of CAIA
B6 female mice (n = 5 to 6 per group, 7 to 8 weeks old) were injected intravenously with 2 mg of a mixture of anti-CII monoclonal antibodies (mAbs) (Arthrogen-CIA mAb (Chondrex. LLC. Seattle, WA, USA)), and two days later with 50 μg of lipopolysaccharide (LPS) was injected intraperitoneally.
Clinical assessment of arthritis
Mice were examined for signs of joint inflammation and scored as follows: 0: no change, 1: significant swelling and redness of one digit, 2: mild swelling and erythema of the limb or swelling of more than two digits, 3: marked swelling and erythema of the limb, 4: maximal swelling and redness of the limb and later, ankylosis. The average macroscopic score was expressed as a cumulative value for all paws, with a maximum possible score of 16.
Thioglycollate-induced peritonitis
Mice were injected with 1 ml of 4% sterile thioglycollate intraperitoneally. Four hours later, mice were killed and peritoneal lavage fluid was collected by washing the peritoneal cavity with cold PBS containing 5 mM EDTA and 10 U/ml heparin.
Administration of TFM-C or celecoxib
TFM-C and celecoxib were synthesized as previously described [
23]. We injected TFM-C or celecoxib intraperitonealy (i.p.) in 0.5% Tween/5% DMSO/PBS. In CIA experiments, mice received 10 μg/g TFM-C or celecoxib every other day from 21 days after immunuization. In CAIA, we injected the mice with 10 μg/g of TFM-C or celecoxib every other day starting at two days before disease induction. In thioglycollate-induced peritonitis experiments, mice received 10 μg/g of TFM-C or celecoxib two days and one hour before thioglycollate injection. The control animals were injected with vehicle alone.
Histopathology
Arthritic mice were sacrificed and all four paws were fixed in buffered formalin, decalcified, embedded in paraffin, sectioned, and then stained with H&E. Histological assessment of joint inflammation was scored as follows: 0: normal joint, 1: mild arthritis: minimal synovitis without cartilage/bone erosions, 2: moderate arthritis: synovitis and erosions but joint architecture maintained, 3: severe arthritis; synovitis, erosions, and loss of joint integrity. The average of the macroscopic score was expressed as a cumulative value of all paws, with a maximum possible score of 12.
Mast cells in synovium were visually assessed for intact versus degranulating mast cells using morphologic criteria. Mast cells were identified as those cells that contained toluidine blue-positive granules. Only cells in which a nucleus was present were counted. Degranulating cells were defined by the presence of granules outside the cell border with coincident vacant granule space within the cell border as described previously [
25].
Measurement of CII specific IgG1 and IgG2a
Bovine CII (1 mg/ml) was coated onto ELISA plates (Sumitomo Bakelite, Co., Ltd, Tokyo, Japan) at 4°C overnight. After blocking with 1% bovine serum albumin in PBS, serially diluted serum samples were added onto CII-coated wells. For detection of anti-CII Abs, the plates were incubated with biotin-labeled anti-IgG1 and anti-IgG2a (Southern Biotechnology Associates, Inc., Brimingham, AL, USA) or anti-IgG Ab (CN/Cappel, Aurora, OH, USA) for one hour and then incubated with streptavidin-peroxidase. After adding a substrate, the reaction was evaluated as OD450 values.
Stimulation of or macrophages
B6 mice received 10 μg/g of TFM-C or control vehicle on Day 0 and Day 2, and on Day 3, splenic macrophages were collected and were stimulated by LPS in vitro in the presence of TFM-C or vehicle.
Detection of cytokines
Cytokine levels in the culture supernatant were determined by using a sandwich ELISA. The Abs for IL-1β ELISA were purchased from BD Biosciences (San Jose, CA, USA) and the ELISA Abs for IL-6 and TNFα were purchased from eBioscience (San Diego, CA, USA).
Statistical analysis
CIA and CAIA clinical or pathological scores for groups of mice are presented as the mean group clinical score + SEM, and statistical differences were analyzed with a non-parametric Mann-Whitney U-test. Data for cytokines were analyzed by an unpaired t-test.
Discussion
In the present study we demonstrate, using arthritis models, that TFM-C, a celecoxib analogue with 205-fold lower COX-2-inhibitory activity, inhibits autoimmune disease. TFM-C differs from celecoxib by the substitution of the 4-methyl group by a trifluoromethyl group. This substitution drastically increases the IC
50s for inhibition of COX1 (15 μM to >100 μM for celecoxib and TFM-C, respectively) and COX2 (0.04 μM to 8.2 μM, respectively), but does not affect the apoptotic index measured in PC3 prostate cancer cells, indicating independence between structural requirements for COX-2 inhibition and apoptosis induction [
36]. Celecoxib perturbs intracellular calcium by blocking ER Ca
2+ ATPases, and this activity is shared with TFM-C [
23,
37]. In a HEK293 recombinant cell system, this Ca
2+ perturbation is associated with inhibition of secretion and altered intracellular interaction of IL-12 polypeptide chains with the ER chaperones calreticulin and ERp44, and results in the interception of IL-12 by HERP followed by degradation of the cytokine [
23,
24,
26]. While IC
50s for inhibition of IL-12 secretion by celecoxib or TFM-C are similar [
23,
24], in the present paper, we show that TFM-C inhibits production of various cytokines from activated macrophages (Figures
1 and
5) and exerts a strikingly stronger inhibitory effect on arthritis models compared to celecoxib. Given that the main biological difference between celecoxib and TFM-C resides in the extent of COX-1 and -2 inhibition, it is, therefore, likely that the less potent effect of TFM-C on COX1/2 inactivation is a contributing, disease-limiting rather than disease-promoting factor in these arthritis models. Indications supporting this concept come from a study showing increased LPS-induced macrophage production of TNF-α by inactivation of COX-2 with celecoxib [
38]. Up-regulation of TNF-α by celecoxib was also reported in human PBMCs, rheumatoid synovial cultures and whole blood [
20]. The (please delete inverse)relation (please replace correlation with relation)between the anticipated extent of COX inhibition and production of TNF-α was observed in the present study (Figure
5), where activated macrophages showed a tendency toward increased or decreased TNF-α production in the presence of celecoxib or TFM-C, respectively, compared to vehicle-treated cells. In this cell system (Figure
5), celecoxib significantly increased production of the pro-inflammatory cytokine IL-6 while TFM-C suppressed it. Pending future mechanistic studies, this data indicate that prostaglandin-mediated suppressive effects, or other, as yet to be identified differential TFM-C/celecoxib-related effects on TNF-α production may extend to other cytokines as well, and provide an important clue as to the more potent beneficial effects of TFM-C compared to celecoxib in the arthritis models presented here.
The suppression of antibody-induced arthritis, which requires innate but not acquired immune cells [
29‐
34,
39], suggests that TFM-C also inhibits the activation of innate immune cells while celecoxib does not. In fact, TFM-C suppresses the production of inflammatory cytokines from macrophages and the activation of mast cells as well as the subsequent recruitment of leukocytes. Mast cells are essential for the initiation of antibody-induced arthritis [
29]. Moreover, mast cells are present in human synovia [
40‐
43] and are an important source of both proteases and inflammatory cytokines, including IL-17, in patients with rheumatoid arthritis [
42‐
44]. The clear difference between the effects of TFM-C and celecoxib on the suppression of mast cell activation could explain the differential impact of these compounds on arthritis models. Mast cells are important not only in arthritis but also in other conditions, such as allergy, obesity and diabetes [
45]. Therefore, the suppression of mast cell activation by TFM-C may be applicable for the inhibition of these diseases in addition to autoimmune diseases.
Cytokines and chemokines, such as TNF-α and MCP-1, produced by macrophages, are suggested to play important roles for neutrophil influx in thioglycollate-induced peritonitis [
46]. Mast cells were shown to produce TNF-α, which recruits neutrophils into the peritoneum in an immune complex peritonitis model [
47]. Thus, it is likely that TFM-C suppressed macrophages and mast cells produce such chemoattractants, which in turn inhibited neutrophil influx into the peritoneum. However, it is also possible that TFM-C directly suppressed neutrophil activation. Further studies are required to address this possibility.
As described above, the major players in CAIA are innate immune cells, while adaptive immune cells are not required for disease development. Therefore, CAIA has value as an animal model for the study of the effector phase of arthritis. However, it is well known that adaptive immune cells play a significant role in the pathogenesis of RA and the strongest genetic link in RA is the association with HLA-DR, which is thought to present autoantigens to T cells. The activation of T cells and B cells is believed to initiate and/or enhance the effector inflammation phase of arthritis. In fact, massive infiltration of T and B cells is observed in RA synovium. Therefore, the ideal therapeutic agents for RA are those displaying the capacity to suppress both the induction and effector phases of arthritis. TFM-C treatment suppresses CIA, which requires both innate and adaptive immune cells for the development of arthritis. We previously demonstrated that celecoxib treatment suppresses EAE induced by immunizing B6 mice with myelin oligodendrocyte glycoprotein
35-55 (MOG) peptide [
22]. The suppression of EAE by celecoxib was COX-2 independent and was accompanied by reduced IFN-γ production by MOG-reactive T cells. We observed a trend of reduced anti-CII antibody levels in serum upon TFM-C treatment. As TFM-C inhibited secretion of both recombinant IL-12 and IL-23 using a pIND ponasterone-inducible vector system in HEK293 cells [
23,
24], TFM-C treatment may have also influenced CII-specific immune responses by suppressing antigen-presenting cells.
Specific inhibition of COX-2 has some adverse effects. Rofecoxib, a highly specific COX-2 inhibitor, was withdrawn from the world market because of an increased rate of cardiovascular events in patients with colorectal polyps [
48]. Celecoxib was also shown to augment cardiovascular and thrombotic risk in colorectal adenoma patients, especially in the subgroup suffering from pre-existing atherosclerotic heart disease [
49]. Moreover, inhibition of COX-2 activity has been reported to exacerbate brain inflammation by increasing glial cell activation [
50]. It has been suggested that the inhibition of COX-2-dependent prostaglandin I
2 from endothelial cells may be the major cause of thrombosis [
51]. As the COX-2-inhibitory activity of TFM-C is 205-fold lower than that of celecoxib, the arthritis suppression by TFM-C appears to be independent of COX-2 inhibition. Therefore, TFM-C, which has strong immunoregulatory abilities but low COX-2-inhibitory activity, could serve as a new disease-modifying agent to prevent the progression of autoimmune diseases such as RA.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AC, MM, CT, RT and AP performed and evaluated experiments. AC, TY, KV and SM designed and supervised the experiments. IA and KV provided TFM-C. AC, KV and SM prepared the manuscript. All authors have read and approved the manuscript for publication.