Aminothiazoles inhibit osteoclastogenesis and PGE 2 production in LPS‐stimulated co‐cultures of periodontal ligament and RAW 264.7 cells, and RANKL‐mediated osteoclastogenesis and bone resorption in PBMCs

Abstract Inflammatory mediator prostaglandin E2 (PGE 2) contributes to bone resorption in several inflammatory conditions including periodontitis. The terminal enzyme, microsomal prostaglandin E synthase‐1 (mPGES‐1) regulating PGE 2 synthesis is a promising therapeutic target to reduce inflammatory bone loss. The aim of this study was to investigate effects of mPGES‐1 inhibitors, aminothiazoles TH‐848 and TH‐644, on PGE 2 production and osteoclastogenesis in co‐cultures of periodontal ligament (PDL) and osteoclast progenitor cells RAW 264.7, stimulated by lipopolysaccharide (LPS), and bone resorption in RANKL‐mediated peripheral blood mononuclear cells (PBMCs). PDL and RAW 264.7 cells were cultured separately or co‐cultured and treated with LPS alone or in combination with aminothiazoles. Multinucleated cells stained positively for tartrate‐resistant acid phosphatase (TRAP) were scored as osteoclast‐like cells. Levels of PGE 2, osteoprotegerin (OPG) and interleukin‐6, as well as mRNA expression of mPGES‐1, OPG and RANKL were analysed in PDL cells. PBMCs were treated with RANKL alone or in combination with aminothiazoles. TRAP‐positive multinucleated cells were analysed and bone resorption was measured by the CTX‐I assay. Aminothiazoles reduced LPS‐stimulated osteoclast‐like cell formation both in co‐cultures and in RAW 264.7 cells. Additionally, aminothiazoles inhibited PGE 2 production in LPS‐stimulated cultures, but did not affect LPS‐induced mPGES‐1, OPG or RANKL mRNA expression in PDL cells. In PBMCs, inhibitors decreased both osteoclast differentiation and bone resorption. In conclusion, aminothiazoles reduced the formation of osteoclast‐like cells and decreased the production of PGE 2 in co‐cultures as well as single‐cell cultures. Furthermore, these compounds inhibited RANKL‐induced bone resorption and differentiation of PBMCs, suggesting these inhibitors for future treatment of inflammatory bone loss such as periodontitis.

resorption and differentiation of PBMCs, suggesting these inhibitors for future treatment of inflammatory bone loss such as periodontitis.

| INTRODUCTION
In chronic inflammatory conditions including periodontitis and rheumatoid arthritis, bone loss is initiated by unresolved inflammation in the neighbouring tissues leading to pathological osteoclastogenesis. 1 During osteoclastogenesis, osteoclast progenitors, recruited from the monocytic linage, differentiate into pre-osteoclasts, fuse and form large multinucleated cells and finally mature into activated osteoclasts that start to express tartrate-resistant acid phosphatase (TRAP) and release acid and collagen-degrading enzymes resulting in bone resorption. 1,2 Osteoclastogenesis is regulated by a broad range of molecules and stimuli, including cytokines, prostaglandins and growth factors, which directly/indirectly regulate the production of osteoclastogenic molecules by various cell types, including stromal cells, osteoblast, osteocytes or resident fibroblasts. 1 The main signalling molecules regulating osteoclast formation are macrophage colony stimulating factor (M-CSF), receptor activator of nuclear factor κB (RANK), RANK ligand (RANKL) and osteoprotegerin (OPG). RANKL and OPG are mainly produced by osteoblasts, osteocytes, resident cells such as periodontal ligament (PDL) cells, and in case of inflammation also by activated T-cells. 1 Whereas OPG is the main osteoclastogenesis-inhibitory molecule, specifically increasing bone density, RANK-RANKL and M-CSF are main molecules inducing bone resorption. 1,3 The latter can also be induced through periodontal bacteria or bacterial components such as lipopolysaccharide (LPS) acting via Toll-like receptors (TLRs) 4 and 2, leading to increased expression of RANKL and other proinflammatory cytokines and osteoclastogenesis-stimulating molecules, including TNFα, interleukin (IL)-1 and M-CSF. 1,3 In vitro experiments have reported that PDL cells, mainly producing OPG, adapt to bacterial stimuli by producing increased levels of osteoclastogenesis-stimulating molecules, thus increasing bone resorption. 3 The inflammatory mediator prostaglandin E 2 (PGE 2 ) may affect inflammation-induced osteoclastogenesis by stimulation of osteoblasts/stromal cells or by direct stimulation of osteoclast progenitor cells. 4 PGE 2 is synthesised sequentially by the enzymes phospholipase A 2 , cyclooxygenases (COX-1 and COX-2) and the final prostaglandin E synthases (PGES). 4 Three isoforms of PGES have been identified; the inducible microsomal PGES-1 (mPGES-1), and the constitutively expressed membrane-associated PGES-2 and cytosolic PGES. [5][6][7][8][9] In the chronic inflammatory disease periodontitis, elevated levels of PGE 2 have been reported in gingival crevicular fluid of patients with periodontitis suggesting this mediator as a predictor of alveolar bone loss. 10,11 Increased levels of PGE 2 correlate well with enhanced levels of COX-2 and mPGES-1, expressed also in inflamed gingival tissues. 9,11 Controlling the production of PGE 2 through COX-inhibition, using non-steroidal anti-inflammatory drugs (NSAIDs) or selective COX-2 inhibitors, have been reported to decrease the progression of alveolar bone loss and gingival inflammation in periodontitis. [11][12][13][14] However, long-term treatment with NSAIDs and selective COX-2 inhibitors is associated with adverse side effects such as gastrointestinal toxicity and increased the risk of druginduced heart failure. [15][16][17][18][19][20] The downstream enzyme mPGES-1, responsible for PGE 2 production, is therefore an attractive target for new anti-inflammatory drugs with potentially fewer side effects compared to COX-inhibitors. 21 However, to the best of our knowledge, mPGES-1 inhibitors are not yet available for clinical use.
The aminothiazoles TH-848 and TH-644 have been identified as mPGES-1 inhibitors by docking models towards the three-dimensional crystal structure of mPGES-1. 22 These compounds inhibit cytokine-induced PGE 2 synthesis in gingival fibroblasts and in vitro enzyme activity of mPGES-1 as well as reduce alveolar bone resorption in vivo. 22 In addition, the aminothiazoles inhibited both RANKLand LPS-stimulated osteoclast formation of the osteoclast progenitor cell line RAW 264.7. 23 In the present study, we aimed to further elucidate the effect of aminothiazoles on (a) osteoclastogenesis and the production of PGE 2 , IL-6, OPG and mPGES-1 in a co-culture model using PDL and RAW 264.7 cells stimulated by LPS and (b) osteoclastogenesis and bone resorption in peripheral blood mononuclear cells (PBMCs) stimulated with RANKL.

| PDL cells
Periodontal ligament cells were obtained from the root surface of extracted teeth without any clinical signs of periodontal disease from three healthy donors and isolated as previously described. 24 The study was approved by the Regional Ethical Review Board Stockholm and informed consent was obtained from the donors.  100% humidity in 5% CO 2 as previously described. 25 For the experiments, RAW 264.7 cells were seeded in 24-well plates (5000 cells/well) and treated with LPS (1 μg/mL), alone or in combination with aminothiazoles TH-848 (0.2 μmol/L) or TH-644 (15 μmol/L) in a total volume of 0.5 mL. After 2 days, the medium containing all the reagents described above was replaced and stored in −20°C for later PGE 2 , OPG and IL-6 analysis. The cells were allowed to further differentiate into osteoclast-like cells; and after 4 days, the experiment was stopped, the cells were fixed in 2% formaldehyde and stained for TRAP (Supporting Information). previously determined by dose-response experiments. 23 The cells were incubated for 48 hours at 37°C in 100% humidity with 5% CO 2 and 800 μL of medium was collected and stored in −20°C for PGE 2 , OPG and IL-6 analysis. Thereafter, the cells were re-stimulated with 800 μL medium containing the same reagents as described above to further allow osteoclast differentiation of RAW 264.7 cells.

| Co-culture experiments
After 4 days in total, the experiment was stopped and the cell layers were fixed in 2% formaldehyde and stained for the osteoclast marker TRAP (Supporting Information).

| Differentiation of PBMCs
Peripheral blood mononuclear cells were purified from buffy coats donated by five healthy males who had given their informed consent The cells from the plastic wells and bone slices were stained for TRAP, as described in Supporting Information.

| Bone resorption
The TRAP-stained bone slices from day 12 were imaged in coverslipbottomed chambers, immersed in H 2 O, using 60x water immersion objective and NikonA1+ confocal laser microscope system (Nikon, Tokyo, Japan). We used reflection signal obtained with 405 nm laser to image bone contour and 647 nm laser to excite the TRAP stain.
To visualise the resorption pits, Z-stacks of bone contour and TRAP stain were captured and processed to volumetric 3D projections with ImageJ plugin collection Fiji (2.0.0-rc-65/1.52b) and to videos scrolling through the stacks (Videos S1-S4).
To quantify bone resorption, culture medium from the days 9 and 12 was analysed for C-terminal cross-linking telopeptides of type I collagen (CTX-I) using Crosslaps for Culture CTX-I ELISA kit (AC-07F1, Ids, Frankfurt am Main, Germany) according to the manufacturer's instructions.

| PGE 2 analysis
The levels of PGE 2 in the supernatants collected from the cell cultures were analysed using a commercial PGE 2 enzyme immunoassay (EIA) kit (Cayman Chemicals, Ann Arbor, MI, USA) according to the manufacturer's instructions.

| Analysis of OPG and IL-6
Levels of OPG and IL-6 were analysed in the culture supernatants according to the manufacturer's instructions. The minimum detection levels of the assay were 1.9 pg/mL for OPG and 1.4 pg/mL for IL-6.

| Quantitative RT-qPCR
Periodontal ligament cells were seeded in Petri dishes (500 000 cells/dish) and cultured as described above. After 24 hours of treatment with LPS in the absence or presence of TH-848 (0.2 μmol/L) and TH-644 (2 μmol/L), total RNA was isolated from the cells using

| Statistical analysis
One-or two-way ANOVA with Tukey's or Sidak's corrections for multiple analyses were used to determinate statistical significance between treatment groups. Differences in treatment outcome were considered significant if P ≤ 0.05.

| Aminothiazoles decrease the formation of osteoclast-like cells in co-cultures of PDL and RAW 264.7 cells
In agreement with previous findings, LPS treatment stimulated the formation of TRAP-positive multinucleated cells compared to control cells in cultures of RAW 264.7 cells alone. 23 Similarly, but to a lower To verify the origin of PDL cells, we analysed the expression of S100A4, which have previously been suggested as a specific marker for these cells. 27 All PDL cells used in this study were positive for S100A4 ( Figure S1).

| Aminothiazoles do not affect IL-6 or OPG production in co-cultures of PDL and RAW 264.7 cells
The levels of the inflammatory cytokine IL-6 were analysed in the supernatants from the PDL and RAW 264.7 co-cultures as well as from PDL cells alone. LPS stimulated the production of IL-6 in all cultures, although the concentrations of IL-6 were 2.5 times higher in cell-cell compared to separated co-cultures and 14 times higher as compared to PDL cells alone ( Figure 3A). The addition of aminothiazoles in combination with LPS did not significantly affect the levels of IL-6 induced by LPS ( Figure 3A).
The levels of OPG, the RANKL decoy receptor, were also analysed in the same culture supernatants. The results showed that the production of OPG was not affected either by LPS or the aminothiazoles in the cultures ( Figure 3B).  up-regulated by LPS ( Figure 4A). The aminothiazoles, on the other hand, did not affect the LPS-stimulated mPGES-1 expression in PDL cells ( Figure 4A). Similar to mPGES-1, the mRNA expression of OPG and RANKL was up-regulated by LPS, but not affected by the aminothiazoles (Figure 4B,C, respectively).

| Aminothiazoles decrease osteoclastogenesis and bone resorption in PBMC cultures
In the next series of experiments, human CD14 + PBMCs were used to investigate the effect of aminothiazoles on osteoclast formation.  Figure 5A,B). The inhibitor TH-644, at 2 μmol/L, did not affect the formation of TRAP-positive multinucleated cells ( Figure 5A,B).
To confirm that the TRAP-positive cells were osteoclasts, bone resorption experiments were performed on bone slices for 12 days, where cells were stimulated with M-CSF and RANKL in the presence or absence of inhibitors. Using a confocal microscope to visualise the resorption pits, a difference in bone resorption was observed in the 3D images, reconstructed from Z-stacks, where the cells treated with aminothiazole TH-848 formed shallower pits, as compared to the control cells treated with RANKL only (Figure 5C and Videos S1 and S2).
As our results did not show any effect of 2 μmol/L TH-644 on formation of osteoclasts or resorption pits, we tested higher concentration, 15 μmol/L, of this inhibitor, as this concentration was also used for RAW 264.7 cells. Similar to TH-848, TH-644 also inhibited formation of both osteoclasts ( Figure 5D) and bone resorption pits ( Figure 5E).   The precise role of PDL cells in inflammatory bone loss is not fully clarified. These cells play an integral role in the production of the extracellular matrix of the PDL 33 but apart from that, these fibroblast-like cells have been shown to influence the migratory capacity, phagocytic activity and phenotypic maturation of the dendritic cells and macrophages. 34 PDL cells have also been shown to up-regulate RANKL when stimulated with PGE 2 , indicating that they are not only structural cells but also serve a regulatory role in inflammatory bone loss. 35 In the current study, we The inhibition of bone resorption by aminothiazoles, at the concentrations used in the current study, was higher than 50%. These results are in line with our previous findings demonstrating that the aminothiazoles decreased alveolar bone destruction, by almost 50%, in experimental periodontitis in rat model. 22 In the current study, dif-  NFATc1- 45 and Wnt/β-catenin pathways, 42 and therefore both may be sensitive to aminothazoles. This suggestion is further supported by the findings that the expression of NFATc1, its binding and transcriptional activity of NFAT is reduced by antagonists of PGE 2 receptors 46 that are known to play important role in regulating bone formation and resorption. 45,47 Furthermore, RANK-RANKL signalling pathway acting through NF-kB, MAPKs and AKT, induces the expression of different osteoclast-specific genes, including TRAP, MMP-9 and cathepsin K (CTSK), leading to osteoclast differentiation and bone resorption. 48,49 It is noteworthy to mention that the levels of CTSK, the major protease in bone resorption, are reduced by aminothiazoles in LPS as well as RANKL-stimulated RAW 264.7 cells. 23 Periodontitis is a common chronic inflammatory disease resulting in the loss of alveolar bone supporting the teeth leading to tooth loss. Inhibitors of mPGES-1, including aminothiazoles, are relatively new compounds included to the group of substances with antiinflammatory properties. To our knowledge, up to date there are no mPGES-1 inhibitors approved for clinical use. Our group is the first to report and elucidate the impact of aminothiazoles as mPGES-1 inhibitors in a model of periodontitis in vitro. 22 The findings that aminothiazoles inhibit osteoclast differentiation and bone resorption in cell types tested, suggest these compounds as potential inhibitors of bone loss.