Institute of Health Biosciences, The University of Tokushima Graduate School, Tokushima, Japan
Department of Respiratory Medicine and Rheumatology, Institute of Health Biosciences, University of Tokushima Graduate School, 3-18-15 Kuramoto-cho, Tokushima 770-8503, Japan. E-mail: email@example.com
Thymidine phosphorylase (TP) in rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) is induced by tumor necrosis factor α (TNFα) and other cytokines that have been reported to be major inflammation mediators in RA. We previously demonstrated that TP plays an important role in angiogenesis and tumor growth, invasion, and metastasis. The aim of this study was to investigate whether the role of TP in the pathogenesis of RA is similar to its role in tumors.
In FLS obtained from 2 patients with RA, the expression of TP, interferon-γ (IFNγ)–inducible protein 10 (CXCL10), and other cytokines was measured by quantitative real-time polymerase chain reaction, immunoblotting, and enzyme-linked immunosorbent assays. Microarray analysis was performed using FLS transfected with TYMP complementary DNA and treated with a TP inhibitor.
The expression of TP in FLS was up-regulated by TNFα, interleukin-1β (IL-1β), IL-17, IFNγ, and lipopolysaccharide. Microarray analysis of FLS overexpressing TP identified CXCL10 as a thymidine phosphorylase–related gene. The expression of CXCL10 was induced by TNFα, and this induction was suppressed by TYMP small interfering RNA and TP inhibitor. Furthermore, the combination of TNFα and IFNγ synergistically augmented the expression of TP and CXCL10. TP-induced CXCL10 expression was suppressed by the antioxidant EUK-8. In the synovial tissue of patients with RA, TP levels were significantly correlated with CXCL10 expression.
The combination of TNFα and IFNγ strongly induced the expression of thymidine phosphorylase in RA FLS. The induction of thymidine phosphorylase enhanced the expression of CXCL10, which may contribute to the Th1 phenotype and bone destruction observed in RA.
Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by synovial inflammation and progressive destruction of articular cartilage and bone. Recently, several specific molecules, such as tumor necrosis factor α (TNFα), interleukin-6 (IL-6), and CTLA-4, were demonstrated to be therapeutic targets in RA, and antagonists to these molecules were developed (). These biologic agents provide a dramatic effect for the therapeutic management of RA and have updated the treatment strategy for RA (). However, the pathogenesis of RA has not yet been fully clarified.
Meanwhile, the cytokines implicated in RA, such as TNFα, IL-1α, and interferon-γ (IFNγ), have been shown to induce thymidine phosphorylase (TP) in tumor cells (). TP is involved in the metabolism of pyrimidine nucleoside and converts thymidine to thymine and 2-deoxy-d-ribose 1-phosphate. We previously showed that TP is identical to platelet-derived endothelial cell growth factor, an angiogenic factor, and is expressed in many solid tumors (). We also demonstrated that in cancers, TP promoted tumor angiogenesis, proliferation, and metastasis ([4-6]). The overexpression of TP has also been observed in inflammatory diseases such as RA, psoriasis, atherosclerosis, and chronic glomerulonephritis (). In particular, the expression of TP in RA was significantly correlated with disease activity ().
It has been reported that TP levels in synovial fluid and serum were higher in patients with RA than in patients with osteoarthritis (OA) and healthy control subjects (), and that TP was induced by TNFα, IL-1, IL-6, and IL-8 in RA fibroblast-like synoviocytes (FLS) (). Although the molecular basis for TP induction by those cytokines is not fully understood, previous studies demonstrated that the TP promoter region contains a cluster of 6–9 Sp1-binding motifs (), and that TNFα induced the expression of TP through the activation of transcription factor Sp1 in RA FLS (). These findings indicate the putative mechanisms underlying how TNFα stimulates the expression of TP.
RA is characterized as a Th1 disease, and the Th1/Th2 cytokine imbalance with a predominance of Th1 cytokines, including IFNγ, has been suggested to be important in the pathogenesis of RA (). Th1 cells express particular chemokine receptors such as CXCR3 and CCR5 (), and IFNγ-inducible protein 10 (CXCL10) is one of the CXCR3-binding ligands. CXCL10 has been detected in the sera, synovial fluid, and synovial tissue of patients with RA (). CXCL10 in RA has been shown to promote FLS invasion () and increase the expression of RANKL in CD4+ T cells; CXCL10 also had a potential role in joint erosion (). These findings suggest that CXCL10 is implicated in the pathogenesis of RA.
In this study, we investigated whether and how TP is involved in the pathogenesis of RA, using RA FLS. We observed that TP regulated the expression of cytokines, including CXCL10, in RA FLS. TP may contribute to the pathogenesis of RA, such as Th1 cell predominance, invasion of FLS, and bone destruction, through CXCL10 expression.
MATERIALS AND METHODS
FLS were obtained by collagenase digestion of synovial tissue from 2 patients with RA (named KTZRA cells and IORA cells) and 2 patients with OA (named K604OA cells and SNOA cells), as described previously ([16, 17]). The human embryonic lung fibroblast cell line MRC-5 was purchased from ATCC. Cells were grown in Dulbecco's modified Eagle's medium (Nissui Seiyaku) containing 10% fetal bovine serum, 2 mM glutamine, and 100 units/ml penicillin at 37°C in a humidified atmosphere of 5% CO2. All experiments were performed using FLS from passages 3–5.
Recombinant human TNFα (rhTNFα), rhIL-1β, rhIL-6, and rhIL-8 were purchased from R&D Systems. Recombinant human IFNγ, rhIL-17, lipopolysaccharide (LPS), and EUK-8 were purchased from PeproTech, HumanZyme, Sigma-Aldrich, and Calbiochem, respectively. The TP inhibitor (TPI) 5-chloro-6-[1-(2-iminopyrrolidignyl) methyl] uracil hydrochloride was supplied by Taiho Pharmaceutical.
Full-length TYMP complementary DNA (cDNA) was provided by Dr. K. Miyazono and Dr. C-H. Heldin (Ludwig Institute for Cancer Research, Uppsala, Sweden). TYMP and scrambled small interfering RNA (siRNA) duplexes were purchased from Sigma. The transfection of TYMP cDNA and the transfection of TYMP and scrambled siRNA in FLS were carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
Total cellular RNA was extracted from cells, using TRIzol reagent according to the manufacturer's instructions (Invitrogen). One microgram of RNA was reverse transcribed using a ReverTra Aceα First Strand cDNA Synthesis Kit (Toyobo). Quantitative real-time PCR was performed using SYBR Premix Ex Taq (Takara) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) according to the manufacturer's instructions. Quantitative measurements were determined using the ΔΔCt method, and the expression of GAPDH was used as an internal control. Melting curve analyses of all real-time PCR products were performed and shown to produce the sole DNA duplex. The primers used were as follows: for TYMP, 5′-GCTGGAGTCTATTCCTGGATTC-3′ (sense) and 5′-ACTGAGAATGGAGGCTGTGATG-3′ (antisense); for CXCL10, 5′-TGACTCTAAGTGGCATTCAAG-3′ (sense) and 5′-CAGGTACAGCGTACAGTTCT-3′ (antisense); for CXCL11, 5′-AAGCCTCCATAATGTACCCA-3′ (sense) and 5′-TATAAGCCTTGCTTGCTTCG-3′ (antisense); for CD40, 5′-TTTCTGATACCATCTGCGAGC-3′ (sense) and 5′-CAACCAGGTCTTTGGTCTCAC-3′ (antisense); for CCL3, 5′-AACCAGTTCTCTGCATCACT-3′ (sense) and 5′-GCTCGTCTCAAAGTAGTCAG-3′ (antisense); for IL32, 5′-ACAACAA GGAACACTCTGTG-3′ (sense) and 5′-AAACACTCCAGGATCAGTCT-3′ (antisense); for CCL7, 5′-AATACTTCAACTACCTGCTGC-3′ (sense) and 5′-CACAGATCTCCTTGTCCAGT-3′ (antisense); for ANGPTL4, 5′-TGGAGGCTGGACAGTAATTC-3′ (sense) and 5′-GTGATGCTATGCACCTTCTC-3′ (antisense); for ADAMTS4, 5′-ATTGGCTCCAAGAAGAAGTTT-3′ (sense) and 5′-ACCACATTGTTGTATCCGTAC-3′ (antisense).
Whole cell extracts were prepared with M-PER reagent (Thermo Scientific) containing phosphatase and protease inhibitor cocktails (Roche). The concentrations of protein were measured using the Bradford method. Fifty micrograms of total cell extract protein was electrophoresed on 4–10% NuPAGE Bis-Tris Mini Gels (200V for 40 minutes). Gel proteins were then electrophoretically transferred onto polyvinylidene difluoride membranes (Millipore) using a Bio-Rad Trans-Blot SD system (). The membrane was treated with the blocking agent Blocking One (Nacalai Tesque) for 1 hour and incubated with the mouse monoclonal antibody against human TP, as described previously (), or with anti–β-actin antibody (1:1,000 dilution; Santa Cruz Biotechnology) or anti–α-tubulin antibody (Calbiochem) overnight at 4°C. Following 4 washes, the membrane was incubated with horseradish peroxidase–conjugated secondary antibodies (Amersham Pharmacia) in the buffer for 1 hour at room temperature. The membrane was then washed and developed using an Enhanced Chemiluminescence Western blotting Detection system (Amersham Pharmacia). Finally, immunoreactive bands were read under an LAS-4000 mini luminescent image analyzer (Fujifilm Medical Systems).
Total RNA from FLS was isolated using an RNeasy Mini Kit (Qiagen). The relative purity of the RNA was measured using an Agilent 2100 Bioanalyzer. Total RNA (130 ng) was amplified and labeled using an Agilent Low Input Quick Amp Labeling Kit, One-Color (5190-2305), and labeled RNA was hybridized with a Human Gene Expression 4x44K v2 Microarray Kit (catalog no. G4845A). Agilent Feature Extraction version 10.7.3 image analysis software was used to extract data from raw microarray image files. Data visualization and analysis were performed using GeneSpring GX version 12.1 software.
Enzyme-linked immunosorbent assays (ELISAs) of CXCL10.
CXCL10 concentrations in the culture medium were determined by ELISA (R&D Systems) according to the manufacturer's instructions.
Published microarray data sets
Public microarray data sets (the GSE1919 [Ungethuem] data set and the GSE39340 [Chang] data set) were downloaded from the GEO database. Data visualization and analysis were performed using GeneSpring GX version 12.1 software. Monochannel data in the Ungethuem data set and the Chang data set were normalized using the robust multiarray average and the 75th percentile shift, respectively.
Results were analyzed using GraphPad Prism version 5.0 software. Statistical analyses for the experiments were carried out using one-way analysis of variance. Statistical analyses for microarray data obtained from GEO were carried out using the Kruskal-Wallis test and Spearman's correlation. Data are presented as the mean ± SD. P values less than 0.05 were considered significant.
Inflammatory cytokine–augmented TP expression in FLS
Previous studies indicated that TNFα induced the expression of TP in RA FLS (). We observed TP expression in both KTZRA cells (Figure 1A) and IORA cells (results not shown) treated with TNFα. The expression of TYMP messenger RNA (mRNA) in RA FLS increased in a time-dependent manner and reached a maximum level 36 hours after TNFα stimulation and also increased with increasing concentrations of TNFα up to 1,000 pg/ml (Figure 1A). The expression of TYMP mRNA also increased with increasing concentrations of TNFα up to 1,000 pg/ml in OA FLS but decreased in MRC-5 cells (data not shown). Additionally, the expression of TYMP mRNA in RA FLS increased in a time-dependent manner after treatment with IL-17, IFNγ, IL-1β, and LPS but not after treatment with IL-6 or IL-8 (data not shown). TYMP mRNA expression also increased in a dose-dependent manner following treatment with IL-17 and IFNγ, and also partially with IL-1β and LPS (Figure 1B). These results suggested that various cytokines that were associated with RA disease activity augmented the expression of TP in FLS.
TP-induced genes using microarray analysis
TP has been reported to increase the expression of IL-6, IL-8, vascular endothelial growth factor (VEGF), matrix metalloproteinase 1 (MMP-1), and MMP-3 in cancer cells ([3, 7]). We examined whether TNFα-induced TP enhanced the expression of these genes in FLS. The expression of TYMP, IL6, IL8, MMP1, and MMP3, but not VEGF, was induced by TNFα in KTZRA cells (data not shown). However, both an enzymatic inhibitor of TP at a dose of 300 μM and TYMP siRNA did not markedly suppress the expression of these TNFα-induced genes in KTZRA cells or IORA cells (data not shown). TPI at a dose of 300 μM was not cytotoxic to KTZRA cells (data not shown).
We speculated that a large part of the enhanced expression of these genes was caused directly by TNFα. Therefore, in order to explore the TP-induced genes in FLS, we performed microarray analysis of KTZRA cells transfected with a control vector (KTZRA/CV), KTZRA cells transfected with TYMP cDNA (KTZRA/TP), and KTZRA/TP cells treated with TPI for 48 hours. We confirmed the expression of TYMP mRNA in KTZRA/TP cells (Figure 2A).
Among ∼30,000 genes tested using an Agilent Whole-Genome microarray, 1,052 genes were up-regulated more than 2-fold above control in KTZRA/TP cells. When KTZRA/TP cells were treated with TPI, 1,299 genes were down-regulated. Among the genes that were up-regulated by TP, 899 genes were down-regulated by TPI (Figure 2B). Among these 899 genes, we observed 19 that were reportedly related to RA (Table 1). We confirmed the expression of these and other genes by real-time PCR (data not shown). In particular, the expression of CXCL10 and CXCL11 mRNA in KTZRA/TP cells was more than 30-fold higher than that in KTZRA/CV cells (Figure 2C). Both CXCL10 and CXCL11 bind to CXCR3, which enhances the migration of Th1 cells and contributes to the Th1 phenotype of RA (). Moreover, CXCL10 regulates FLS invasion () and bone destruction () in RA. We thus focused our study on CXCL10.
Table 1. Genes shown to be related to thymidine phosphorylase (TP) by microarray analysis
aMessenger RNA level in KTZRA cells transfected with TP (TYMP) cDNA (KTZRA/TP) divided by that in KTZRA cells transfected with a control vector (KTZRA/CV). Values >1 indicate that the gene is induced by TP.
bMessenger RNA level in KTZRA/TP cells treated with 300 μM TP inhibitor (TPI) divided by that in KTZRA/TP cells not treated with TPI. Values <1 indicate that the TP-induced gene is suppressed by TPI.
Enhanced expression of TYMP-induced CXCL10 by the combination of TNFα and IFNγ.
We next examined the expression of TP and CXCL10 in KTZRA cells (Figure 3A) and IORA cells (data not shown). In this experiment, we used clinically relevant doses of TNFα or IFNγ in the synovial fluid of patients with RA ([21-23]). The expression of TYMP mRNA was increased by both TNFα and IFNγ, and this increase was suppressed by TYMP siRNA (Figure 3A). CXCL10 was also induced by TNFα at a dose of 100 pg/ml, and this induction was suppressed by TYMP siRNA (Figure 3A). TPI reduced the expression of CXCL10 mRNA in KTZRA cells treated with TNFα (data not shown). These results suggested that the enzymatic activity of TPI is essential for the induction of CXCL10. The expression of CXCL10 mRNA was not induced by a clinically relevant dose of IFNγ (100 pg/ml) (Figure 3A), but a higher concentration of IFNγ (1 ng/ml) significantly augmented the expression of CXCL10 (data not shown).
The combination of TNFα and IFNγ synergistically enhanced the expression of TP and CXCL10 (Figures 3B and C) in KTZRA cells. Although TNFα alone up-regulated the expression of TP and CXCL10, IFNγ alone did not significantly augment the expression of CXCL10. However, IFNγ potentiated the effect of TNFα on the expression of TP and CXCL10 in KTZRA cells. The induction of CXCL10 by both cytokines was inhibited by TYMP siRNA (Figure 3C). These findings indicated that TNFα and IFNγ synergistically induced the expression of TP, which increased CXCL10 expression. Meanwhile, the expression of CXCL10 mRNA was enhanced with increasing concentrations of IL-1β and LPS but not IL-17 (data not shown). However, IL-1β–induced CXCL10 expression was not significantly suppressed by TPI (data not shown), indicating that the induction of CXCL10 by IL-1β may not be regulated by TP.
Effect of antioxidant on the expression of TYMP-induced CXCL10 in FLS.
We previously reported that TP enhanced the generation of reactive oxygen species (ROS) in cancer cells, and that the enzymatic activity of TP was required for the generation of ROS (). To examine whether TP induces CXCL10 expression through ROS generation in FLS, KTZRA/TP cells (Figure 4A) and KTZRA cells treated with TNFα (Figure 4B) were treated with the antioxidant EUK-8. The increased CXCL10 expression in KTZRA/TP cells was significantly suppressed by EUK-8 (Figure 4A). The increased expression of CXCL10 induced by TNFα was also significantly attenuated by EUK-8 (Figure 4B). These results suggested that TP augments CXCL10 expression through ROS production. Interestingly, the TNFα-induced TYMP expression observed in KTZRA cells was attenuated by EUK-8 (Figure 4B). ROS may also be involved in the increased expression of TYMP by TNFα.
Expression of TP and CXCL10 in the synovial tissue of patients with RA
We investigated TP and CXCL10 levels in synovial tissue using the Ungethuem data set (GSE1919) and the Chang data set (GSE39340) of the GEO. The Ungethuem data set included gene expression levels in synovial tissue from 5 normal donors, 5 patients with OA, and 5 patients with RA. The Chang data set included gene expression levels in synovial tissue from 5 patients with ankylosing spondylitis, 7 patients with OA, and 10 patients with RA.
We first used the Ungethuem data set. The expression levels of TP and CXCL10 in RA tissue were higher than those in tissue from normal donors (Figure 5A). The expression levels of TP and CXCL10 in OA tissue were not significantly different from those in normal donors. TP levels in all of these tissues were correlated with CXCL10 expression (Figure 5B). In the Chang data set, the expression of TP and CXCL10 was higher in RA tissue than in OA and AS tissue (Figure 5C). TP levels were significantly correlated with CXCL10 levels in all tissues and in RA tissue alone (Figure 5D). These results suggested that the participation of TP in the induction of CXCL10 expression is clinically relevant.
In the present study, we demonstrated that TP is involved in the expression of CXCL10 in FLS. Moreover, the expression of TP in FLS was markedly up-regulated by the combination of TNFα and IFNγ. Because the induction of CXCL10 activated by TNFα and IFNγ in FLS was markedly suppressed by the knockdown of TP, TP may play a critical role in the induction of CXCL10 by these cytokines. To the best of our knowledge, this study is the first to demonstrate the augmentation of CXCL10 expression by TP in RA FLS.
CXCL10 levels in the synovial fluid and synovial tissue of patients with RA are increased, and serum levels of CXCL10 are correlated with RA disease activity ([25, 26]). A phase II clinical trial using an anti-CXCL10 monoclonal antibody (MDX-1100) in patients with RA has been previously described (). MDX-1100 administered every 2 weeks for 12 weeks in combination with weekly methotrexate led to a statistically significant improvement in the American College of Rheumatology 20% response () compared with placebo (). These studies showed that CXCL10 contributed to the pathogenesis of RA and is a promising molecular target for anti-RA therapy.
CXCR3, a CXCL10 receptor, is expressed on several immune cell types, such as natural killer cells, plasmacytoid and myeloid dendritic cells, B cells, and, especially, Th1 cells. Accumulating evidence has suggested that CXCR3 ligands are involved in the selective recruitment of Th1 cells into the site of tissue inflammation ([29, 30]). CXCL10 expression was increased in a mouse model of collagen-induced arthritis and played a critical role in the infiltration of Th1 cells (). We observed that the induction of TP by TNFα and IFNγ increased the expression of CXCL10 in FLS, indicating that TP induced by these cytokines may contribute to the Th1 phenotype in RA. Meanwhile, Laragione et al reported that CXCL10 and its receptor CXCR3 enhanced the invasion of synovial fibroblasts in RA ().
We previously demonstrated that TP is involved in the invasion and metastasis of some solid tumors (). We also showed that TP increased the expression of MMP-1, MMP-9, IL-8, and VEGF in cancer cells (). Leek et al suggested that TNFα produced by neighboring tumor-associated macrophages may play a role in the regulation of TP expression in tumor cells, as well as their metastatic behavior (). TP may promote the invasive activity of FLS as well as carcinoma cells by enhancing the expression of CXCL10. Kwak et al demonstrated that CXCL10 played a critical role in the infiltration of Th1 cells and macrophages into inflamed joints (). CXCL10 promoted the expression of RANKL in Th1 cells, which caused bone destruction by enhancing osteoclast differentiation (). We demonstrated that TP induced the expression of CXCL10 in RA FLS. These findings suggested that TP is involved in the Th1 phenotype and bone destruction observed in the pathogenesis of RA.
Besides CXCL10 and CXCL11, many genes, including CCL3, IL32, CD40, ANGPTL4, CCL7, and ADAMTS4, were also induced by TP. CCL3 as well as CXCL10 recruit immune cells, including Th1 cells. Genome-wide haplotype association and gene prioritization identified CCL3 as a risk locus for RA (). IL-32 was a potent inducer of prostaglandin E2 release in mouse macrophages and human monocytes (). The level of IL-32 in synovial tissue from patients with RA was correlated with the erythrocyte sedimentation rate, which is a marker of systemic inflammation (). The interaction of CD40 on the surface of FLS with CD40L expressed on activated T lymphocytes enhanced the production of VEGF and, consequently, neovascularization in RA (). ANGPTL4 also had angiogenic activity in a mouse model of CIA () and enhanced the expression of MMP-1 and MMP-3 in chondrocytes (). TP may cause pleiotropic effects on RA pathogenesis by enhancing the expression of these genes.
Kusabe et al showed that antirheumatic drugs such as aurothioglucose and dexamethasone suppressed IL-1β–induced TP expression in FLS (). These investigators also demonstrated that the immunosuppressant FK-506 (tacrolimus), which has been approved as an antirheumatic drug in Japan, inhibited the level of TP that was increased by TNFα (). We confirmed that aurothioglucose and FK-506 suppressed the expression of TYMP and CXCL10 mRNA in RA FLS (data not shown). These results suggested that the antirheumatic effect of these drugs may be attributable, at least in part, to antiangiogenic and antiarthriogenic activities following the down-regulation of TP.
In the present study, both TYMP siRNA and TPI inhibited the expression of TP as well as CXCL10 induced by TNFα, with or without IFNγ, in FLS. It is still unclear how the enzymatic activity of TP is involved in the expression of TP and CXCL10. Further study is required to clarify this mechanism.
TP generates ROS and increases proinflammatory cytokines in cancer cells (). We previously reported that TP augments ROS generation in cancer cells, and that the enzymatic activity of TP was required for the generation of ROS (). In the current study, we examined whether TP induces CXCL10 expression through ROS generation in RA FLS. The TP-induced CXCL10 expression was suppressed by the antioxidant EUK-8. Furthermore, the induction of TP expression by TNFα in FLS was also suppressed with EUK-8 treatment. These findings suggested that TNFα augmented TP expression through ROS production, and TP also enhanced the ROS level and consequently increased CXCL10 expression.
In conclusion, we demonstrated that TP regulated the expression of CXCL10 in RA FLS via enzymatic activity. Therefore, compounds that inhibit TP activity, such as TPI, may be good candidates for new anti-RA agents.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Nishioka had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Toyoda, Tabata, Akiyama, Nishioka.
Acquisition of data. Toyoda, Tabata, Kuramoto, Mitsuhashi, Saijo, Horikawa.
Analysis and interpretation of data. Toyoda, Tabata, Kishi, Kawano, Goto, Aono, Hanibuchi, Horikawa, Nakajima, Furukawa, Sone, Akiyama, Nishioka.
We thank Ms Tomoko Oka (Institute of Health Biosciences, University of Tokashima Graduate School) for technical assistance.