Targeting thymidine phosphorylase as a potential therapy for bone loss associated with periprosthetic osteolysis

Abstract Macrophages are generally thought to play a key role in the pathogenesis of aseptic loosening through initiating periprosthetic inflammation and pathological bone resorption. The aim of this study was to identify macrophage‐derived factors that promote osteoclast differentiation and periprosthetic bone destruction. To achieve this, we examined the effects of 12 macrophage‐derived factors that were identified by RNA‐seq analysis of stimulated macrophages on osteoclast differentiation. Surprisingly, thymidine phosphorylase (TYMP) was found to trigger significant number of osteoclasts that exhibited resorbing activities on dentine slices. Functionally, TYMP knockdown reduced the number of osteoclasts in macrophages that had been stimulated with polyethylene debris. TYMP were detected in serum and synovial tissues of patients that had been diagnosed with aseptic loosening. Moreover, the administration of TYMP onto calvariae of mice induced pathological bone resorption that was accompanied by an excessive infiltration of inflammatory cells and osteoclasts. The RNA‐seq for TYMP‐induced‐osteoclasts was then performed in an effort to understand action mode of TYMP. TYMP stimulation appeared to activate the tyrosine kinase FYN signaling associated with osteoclast formation. Oral administration of saracatinib, a FYN kinase inhibitor, significantly suppressed formation of bone osteolytic lesions in a polyethylene debris‐induced osteolysis model. Our findings highlight a novel molecular target for therapeutic intervention in periprosthetic osteolysis.


| INTRODUCTION
Total joint arthroplasty (TJA) is an excellent and the most appropriate approach for the treatment of end-stage arthritic diseases in that it allows joint function to be preserved and also provides pain relief. Implant design and fixation, surgical procedures, infections, and aseptic loosening are the main cases of TJA failure. However, aseptic loosening, a process that is referred to as periprosthetic osteolysis, is the most frequent cause of TJA failure with incidence rates estimated to be 10-70%. It is estimated that over 25% of all TJAs fail due to aseptic loosening, a condition that requires revision surgery. 1 Osteolysis mainly occurs due to inflammatory response initiated by wear particles that are released from sliding surface of prosthetic materials. Persistent local inflammatory response promotes a bone remodeling process typified by increased osteoclast activities and bone loss that leads to the loss of prosthesis fixation and implant failure. [2][3][4] Neither anti-inflammatory nor anti-resorptive agents have proven to prevent the progression of osteolysis or prolong the lifespan of an implant. Effective therapy is needed to prevent osteolysis as a step toward improving the quality of life for patients and reducing economic burden due to expenses of health service.
Ultra-high-molecular-weight-polyethylene (UHMWPE) remains the gold standard for prosthetic bearing materials with substantial resistance to abrasion and wear. However, despite the strength of UHMWPE, wear continues to be the major determinant of the lifetime of a prosthesis. 2,3 Indeed, wear debris that is deposited in the periprosthetic space with rates exceeding 0.15 mm/year have been documented with a high risk of occurrence of periprosthetic osteolysis. 2 UHMWPE particles with sizes of 0.1 and 10 μm has been frequently been detected in periprosthetic tissues, and particles with sizes of 0.1-2.0 μm are thought to be the most biologically active trigger of high inflammatory responses to pathological bone resorption. 4 With respect to implants, macrophages are the first line of cells that interact with implanted medical devices. Wear particles are recognized and phagocytosed by resident macrophages in periprosthetic tissues and bone that produce an array of proinflammatory mediators that facilitate cell recruitment, maturation, proliferation, and differentiation. In addition to their function in inflammation, macrophages, including those that are resident in periprosthetic tissues, bone and bone marrow play prominent roles in bone remodeling through producing osteoclastogenic factors and their ability to differentiate into bone resorbing osteoclasts. 5,6 Indeed, it is known that macrophage-derived cytokines including TNF-α, IL-1, IL-6, and TNFSF15 promote the differentiation of osteoclast, either directly or indirectly, by inducing the expression of the receptor activator of nuclear factor kappa B ligand (RANKL), a key regulator of osteoclastogenesis, in lymphocytes, osteoblasts and fibroblasts. [6][7][8][9][10] Consistent with this view, our recent study demonstrated that macrophages stimulated with UHMWPE debris express gene signatures similar to those for rheumatoid arthritis including molecules associated with inflammatory response and osteoclast differentiation. 10 Specifically, inflammation and pathological bone resorption are the primary features of the pathogenesis of periprosthetic osteolysis and rheumatoid arthritis. 11 In both diseases, macrophages play a principal role in the development of inflammatory lesions and joint erosion. [11][12][13] Therefore, studying crosstalk between inflammatory macrophages that are resident in periprosthetic tissues and osteoclast progenitors around an implant may provide clues to the development of effective medical therapies for the treatment of periprosthetic osteolysis.
Given the importance of molecular mediators of macrophages that are stimulated by wear particles as attractive therapeutic targets for preventing periprosthetic osteolysis, the objective of this study was to identify macrophage-derived factors that are directly involved in osteoclast differentiation. Our study identified a novel potential F I G U R E 1 Stimulation macrophages by UHMWPE particles induced osteoclast differentiation. (a) Heat map of regulated genes in macrophages that were stimulated with UHMWPE particles for 24 h (n = 3). Log2 fold change >1.0. A scale bar for intense color change from À1 or below indicated by blue color and 1 or above indicated by red. (b) KEGG pathways enrichment analysis of upregulated genes in response to UHMWPE particles stimulation. (c) Acyclic graph for NET-work-based gene enrichment of the enriched KEGG pathways (p ≤ 1eÀ07). (d) Representative images for TRAP-stained macrophages stimulated with UHMWPE particles. Arrows indicate osteoclasts. Scale bars represent 100 μm. (e) MA plot analysis for transcript expression levels of significantly up-or downregulated genes in stimulated macrophages. Selected targeted molecules are indicated osteoclastogenic factor and highlighted a promising molecular target for therapeutic interventions in implant loosening.

| TYMP is macrophage-derived factor inducing osteoclast differentiation
To identify macrophage-secreted factors that are involved in development of pathological bone resorption associated with periprosthetic osteolysis, we first analyzed gene profiles of macrophages that had been stimulated with UHMWPE particles. In the case of macrophages that had been stimulated by UHMWPE particles for 24 h, 2143 genes that were most significantly enriched in osteoclast differentiation based on an analysis of the KEGG pathway were upregulated ( Figure 1a,b). To assess the potential functional associations between the genes and biological cellular processes, a NETwork-based Gene Enrichment was performed for the significantly upregulation genes.
The findings revealed that the stimulated macrophages expressed the gene signature of rheumatoid arthritis, including genes that are involved in inflammatory responses and osteoclast differentiation F I G U R E 2 Identification of TYMP as osteoclastogenic factor. (a) Effects of macrophage derived factors on osteoclasts differentiation. Human monocytes were stimulated with each recombinant protein for 8 days and then stained by TRAP for osteoclast detection. The results represent the means ± SEM for triplicates and * indicates a significant difference of cell count as compared to control. (b) Inhibitory effects of OPG on osteoclast differentiation. Human monocytes were stimulated with RANKL or TYMP plus inhibitory dose of OPG. The results represent the means ± SEM for triplicates and significant differences were determined based on one-way ANOVA, followed by the Tukey's multiplecomparison procedure. **p < 0.01. No significant differences were indicated as ns. (c) Representative images for TRAP-stained cells stimulated with RANKL or TYMP in presence or absence of OPG. Scale bars represent 100 μm. (d) Representative images for osteoclasts induced by RANKL or TYMP. Left panel, cells stained by TRAP, and right panel for cells fluorescently stained by phalloidin for detection of Actin ring formation. Actin ring (red), and cell nuclei (blue). Scale bars represent 100 μm. (e) Quantitative analysis of bone resorption area on dentine slices (n = 4). The right figure shows the bone resorption area on a dentine slice. Scale bars represent 50 μm. (f) Effects of TYMP knockdown on osteoclast differentiation in macrophages stimulated with UHMWPE particles. Results represent the means ± SEM for triplicates and significant differences were determined Student's t-test. **p < 0.01 ( Figure 1c). To confirm this finding, we cultured human macrophages for 6 days in the presence or absence of UHMWPE particles. Matured osteoclasts were clearly observed in cell cultures that had been stimulated by UHMWPE particles (Figure 1d), suggesting that stimulation with wear particles induced macrophages to differentiate into osteoclasts. Based on these results, we speculated that wear particlestimulated macrophages secreted factors that could potentially promote osteoclast differentiation. Of inflammatory-related molecules, 12 target molecules that were found to be highly expressed in rheu-  (Figure 2f). These collective results suggest that TYMP is a potential osteoclastogenic factor that is released from macrophages stimulated with UHMWPE particles.

| TYMP is present in periprosthetic tissues around loosening implants
To verify whether TYMP is, in fact, the osteoclastogenic factor derived from inflammatory macrophages and is involved in periprosthetic osteolysis, the expression of TYMP was examined in stimulated macrophages, and periprosthetic tissues and sera of patients diagnosed with aseptic loosening. The protein level of TYMP was significantly increased in UHMWPE particles that had been stimulated and M1 polarized macrophages (Figure 3a Figure 2). More interestingly, a pull-down assay showed that TYMP was also able to bind a complex protein that con-

| Therapeutic effects of FYN inhibition in murine osteolysis model
To test our hypothesis that FYN is a promising molecular target for therapy of periprosthetic osteolysis, saracatinib, a Src kinase inhibitor, was used to inhibit FYN function in a UHMWPE particle-induced osteolysis model. UHMWPE particles were implanted into calvarial bone and saracatinib was orally administered on days À1, 1, 3, and bodies to these molecules have been clinically approved for the treatment of rheumatoid arthritis, a disease that is closely related to periprosthetic osteolysis, namely, anti-TNF-α drugs (infliximab, adalimumab, certolizumab, and golimumab) 15 and anti-IL-6 drug (Tocilizumab). 16 Moreover, several studies have suggested that blocking cytokines could be a potential therapeutic strategy for periprosthetic osteolysis with promising effects in experimental models. [17][18][19] In the current study, we identified TYMP as a novel osteoclastogenic factor associated with the development of bone loss in periprosthetic osteolysis.
Our results showed that TYMP was most potent osteoclastogenic factor among the molecules that were tested, including molecules that were earlier reported to induce osteoclast differentiation namely TNFSF14, TNFSF15, and VEGFA. 10,20,21 TYMP appeared to induce osteoclast production in presence of OPG, a RANKL receptor that functions as its inhibitor and the resulting osteoclast phenotype exhibited substantial bone resorption activities. TYMP was further detected in macrophages that had infiltrated into periprosthetic tissues in patients who were undergoing revision surgery for hip arthroplasty. Interestingly, the knockdown of TYMP in macrophages resulted in a significant reduction in the number of osteoclasts in UHMWPE particle-stimulated macrophages cultures. TYMP triggered the formation of lytic bone lesions that were comparable to these induced by RANKL in mouse osteolysis models. These collective data suggest that TYMP is one of the osteoclastogenic factors associated with development of periprosthetic osteolysis. Consistent with this view, the elevated expression of TYMP has been documented in inflammatory diseases characterized by focal lytic bone lesions, including RA and malignancies of bone, breast, prostate, and lung. [22][23][24][25] In bone-metastatic tumors, TYMP binds to integrins in osteoclast progenitors resulting in the activation of PI3K/Akt signaling and an increased methylation of interferon regulatory factor 8 (IRF8), a molecule that enhances the expression of nuclear factor of activated T cells, cytoplasmic 1 (NFATc1), thereby leading to osteoclast differentiation and increased bone resorption. 23 Generally, TYMP is known as an enzyme that catalyzes the reversible phosphorolysis of thymidine and deoxyuridine. It is mainly expressed in macrophages and plays an important role in angiogenesis and tumor growth, invasion, and metastasis. An increased expression of TYMP is frequently associated with the aggressiveness of cancer with a poor prognosis. [25][26][27] Intracellular TYMP acts a cytosolic enzyme that is essential for the stability of mitochondrial DNA, and gene mutations cause mitochondrial neurogastrointestinal encephalomyopathy associated with ptosis and progressive external ophthalmoplegia, peripheral neuropathy, severe gastrointestinal dysmotility, cachexia and leukoencephalopathy. 28 Such facts make targeting this gene difficult due to its essential function in cells. On the other hand, extracellular TYMP secreted from immune cells enhances the expression of proinflammatory mediators, including CXCL10, TNF-α, IL-6 and IL-8, and the production of oxygen species (ROS) through NF-κB signaling. [22][23][24][25][26][27]29 Extracellular TYMP activates platelet and thrombosis through binding to SH3 domain-containing proteins of the Src family kinase (SFK) and increasing integrin activity. In fact, TYMP released from macrophages contributes to the formation of atherosclerotic plaques, thus leading to fatal cardiovascular disorders, including myocardial infarction and stroke. 30 These findings suggest that TYMP acts as a signaling molecule that participate in the activation of multiple signaling pathways associated with tissue remodeling.
Our further data showed that TYMP stimulation resulted in a significant upregulation of FYN in macrophages in addition to its ability to pull down complex proteins containing FYN and ITGβ1. These results are in agreement with an earlier study showing that TYMP acts as a signaling molecule by directly binding to Src family kinases (SFKs) through its N-terminus residue, leading to platelet activation and aggregation. 30 SFKs include Src, Yes, Fyn, Fgr, Lck, Hck, Blk, Lyn, and Frk are a group of non-receptor tyrosine kinases that are implicated in the survival and function of osteoclasts. 31 Nonetheless, among the SFKs members, FYN appears to be the most functionally important factor in this process that facilitates the proliferation and differentiation of osteoclasts and retarding their apoptosis. 32 The engagement of FYN with the integrin receptor family has been documented to be associated with initiating intracellular signaling cascades leading to cell activation, proliferation and differentiation. 30,33,34 Specifically, an earlier study showed a functional link between ITGβ1 and FYN kinase in mediating the differentiation, maturation and survival of oligodendrocytes. 33,34 These results highlight the importance of ITGβ1/FYN signaling in driving osteoclast differentiation in response to TYMP stimulation.
Our results suggest a new mechanism that underlines the action of TYMP in osteolysis, because it has been documented that TYMP promotes osteoclastogenesis through downregulating the expression of IRF8, which results in increasing the expression of NFATc1. 23 Our RNA-seq data did not show any significant reduction in the expression of IRF8 in TYMP-stimulated cells, implying a different mode of action.
The discrepancy in the results of the two studies can be explained by the different methodologies used in the current study. In fact, we cultured monocytes with TYMP alone, while Luo et al., cultured osteoclast precursors with myeloma cells plus a low dose RANKL. 23 Generally, osteoclastogenesis is mediated by a canonical pathway (RANKL-dependent) and a noncanonical pathway. In the canonical pathway, osteoclastogenic effects are dependent on intracellular signaling pathway initiated by the binding of RANKL to its receptor RANK that recruits adaptor proteins TNF receptor-associated factors (TRAF), mainly TRAF6, which turn on a range of signal transduction pathways, including NF-κB, MAPKs and AP-1. 35

| Osteoclasts differentiation and bone resorption assays
Human monocytes obtained from the blood of healthy donors (Asian, three males with ages of 30-45 years) were isolated by density gradient centrifugation (Ficoll-PaqueTM PLUS: GE Healthcare, Waukesha, WI) followed by MACS Pan monocyte isolation kit (Miltenyi Biotec, Auburn, CA). 43 Cells were suspended in growth medium containing minimum essential medium Eagle (MEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), a 5% penicillin/streptomycin solution, and 5% Lglutamine and cultured in a 37 C-humidified atmosphere containing 5% CO2 in a 75 cm 2 flask for 3 h. Thereafter, the medium was discarded and adherent monocytes (CD14 + ) were cultured in growth medium supplemented with 25 ng/mL human recombinant macrophage colony-stimulating factor (MCSF; Peprotech, Japan).

| Western blotting
Human synovial fluids and cell lysate were mixed samples buffer EzApply (ATTO, Osaka, Japan) and heated at 100 C for 5 min.

| Preparation of polyethylene particles
Polyethylene particles prepared from hip-bearing materials of ultrahigh-molecular-weight polyethylene as earlier reported. 43 Briefly, UHMWPE was manufactured from GUR1020 powder (Celanese Japan, Tokyo, Japan) after 95 kGy irradiation and annealed below Tm  and gene-specific primers listed in supplementary information Table 1 and our earlier study. 44 Gene expression was determined by the 2ÀΔΔ Ct method with amplification efficiencies ranging between 90 and 110% for the target and reference genes.

| Statistical analysis
Statistical analyses were performed using GraphPad Software (GraphPad Software Inc., La Jolla, CA). One-way analysis of variance (ANOVA) followed by Tukey's multiple-comparison and Student's ttest procedures were used to compare osteoclasts number, percentages of resorbed areas, gene expression, band intensity, bone pits, and cell infiltrates. Results were presented as means ± standard errors of the means (SEM) and were considered statistically significant when p < 0.05.

| CONCLUSIONS
In summary, the findings reported here identified TYMP as a potential osteoclastogenic factor that is released from inflammatory macrophages in response to UHMWPE wear stimulation. Our results show a functional association between TYMP stimulation and FYN signaling leading to osteoclast differentiation. FYN therefore has the potential for being a promising therapeutic target for preventing bone loss associated with implant failure in periprosthetic osteolysis.