Background
Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by marked hyperplasia of the lining layer of the synovium, leading to the destruction of articular cartilage and bone. In RA pathogenesis, fibroblast-like synovial cells (FLS) are pivotal. FLS contribute to the production of pro-inflammatory cytokines, small molecule mediators of inflammation, and proteolytic enzymes that degrade the extracellular matrix [
1]. Moreover, FLSs are resistant to programmed cell death [
2], resulting in an aggressive, invasive phenotype similar to that of an invasive cancer, and the hyperplastic synovial tissue, also called the pannus, destroys cartilage and bone. Although rapid development of cytokine-targeted therapeutic agents such as tumor necrosis factor (TNF) inhibitors has provided better clinical outcomes including achievement of remission for patients with RA, there are many unfavorable problems such as inadequate response, high cost, and adverse events such as infections [
3,
4]. RA-FLS-targeted therapies have thus been explored, and several key mediators that activate cytokine production from FLS [
5,
6] or acquire anti-apoptosis property [
7] had been elucidated. However, despite the enthusiasm for developing new treatments that directly target FLS, no directly FLS-targeted therapy is available at this time.
In RA, FLS of mesenchymal origin conserve mesenchymal properties. The gene expression pattern of FLS is similar to that of mesenchymal stem cells [
8], and in vitro studies have shown that appropriate stimulation in culture induces differentiation of FLS into chondrocytes, adipocytes, muscle cells, and osteoblasts [
9‐
11]. FLS are bone marrow (BM)-derived mesenchymal cells (MSCs) [
12], and the multi-linage potential of arthritic FLS is thought to be arrested at the early stage of differentiation by activation of nuclear factor-κB (NF-κB) [
13]. Forced cell differentiation might become a candidate therapeutic option for RA; for example, mesenchymal stromal cells showed reduced interleukin-6 (IL-6) production after their differentiation into adipocytes [
14]. Until now, there has been no evidence that FLS differentiate into osteoblasts in joints. However, if the induction of intrinsic transdifferentiation of markedly proliferating FLS in the joints causes differentiation of FLS into osteoblasts, it might become a treatment option for RA.
MicroRNAs (miRNAs) are small non-coding RNAs, which regulate gene expression post-transcriptionally by binding to the 5′ untranslated region (UTR), coding regions or 3′ UTR of mRNA [
15]. Altered expression of miRNAs has been reported in many diseases such as cancer, infections, and autoimmune diseases including RA, and this might arise from a modulation of diverse biological processes such as cell proliferation, apoptosis, metabolism and cell differentiation by miRNAs [
16‐
18]. There is growing evidence that miRNAs are critical in osteoblast differentiation [
19].
Jie et al. reported that miR-145 suppressed the osteogenic differentiation of mouse osteoblastic and myoblastic cell lines (MC3T3 and C2C12) by targeting Sp7 [
20]. Other reports revealed that several miRs, e.g., miR-218, miR-34 and miR-195 modulate osteogenic differentiation by suppressing their targets [
21‐
23]. However, the effect of miRNAs on RA-FLS differentiation including osteoblast differentiation had not yet been elucidated.
In the present study, we identified a miRNA (miR-218) that was altered during the osteogenic differentiation of RA-FLS, and we confirmed that this miRNA indeed promoted the osteogenic differentiation of RA-FLS. Our findings also revealed that Wnt/β-catenin signaling is involved in the promotion of the osteogenesis of RA-FLS by miR-218.
Methods
Isolation of FLS and stimulation assays
We obtained synovial tissues from patients with RA who fulfilled the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) classification criteria for RA [
24] or the 1987 ACR classification criteria for RA [
25] at the time of orthopedic surgery. Each patient provided a signed consent form to participate in the study, which was approved by the Institutional Review Board of Nagasaki University and the Swiss Ethical commission. FLS were isolated from synovial tissues as described previously [
26,
27]. Cells were cultivated in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% heat-inactivated fetal bovine serum (FBS), 100 units/ml penicillin and 100 ng/ml streptomycin (all from Gibco, Basel, Switzerland). FLS from passages 3–8 in monolayer culture were used for the experiments. In the stimulation experiments, FLS were stimulated for 24 h with recombinant TNF-α (10 ng/ml), interleukin-1β (IL-1β) (1 ng/ml) (R&D Systems, Abingdon, UK), recombinant interleukin-6 (IL-6) (100 ng/ml) and recombinant soluble IL-6 receptor (sIL-6R) (100 ng/ml) (Peprotech, Rocky Hill, NJ, USA).
Osteogenic differentiation in vitro
RA-FLSs were plated at a cell density of 1 × 104 in 12-well plates. After they were 70% confluent, medium was replaced with osteogenic differentiation Bulletkit™ medium containing dexamethasone, ascorbate, glycerophosphate, L-Glutamine, Pen/Strep and MCGS (Lonza, Walkersville, MD, USA) to differentiate to osteoblasts. RA-FLS was cultured in the induction medium for up to 21 days. The medium was changed every 3 days. Osteoblast differentiation was evaluated by alkaline phosphatase (ALP) staining and Alizarin Red staining.
ALP staining and Alizarin Red staining
After the osteogenic induction or transfection experiments, cells were fixed in 4% paraformaldehyde and stained with ALP using ALP staining kit (Cosmo Bio, Tokyo). In another set of experiments, we performed Alizarin Red staining to detect the calcification after 21 days of culture in induction medium (late period of induction). Cells were fixed in methanol and stained with Alizarin Red using Calcified Nodule Staining kit (Cosmo Bio). ALP-positive cells were stained blue by ALP staining, and calcium nodules were detected as red bodies by Alizarin Red staining.
Transfection experiments
For a transient transfection approach with the aim to inhibit or enhance the miR-218 function, RA-FLSs were transfected with synthetic precursor miRNA (Pre-miR), with inhibitors of miR-218 (anti-miR), or with scrambled controls (Pre-miR/Anti-miR Negative Control #1; Ambion/Applied Biosystems, Foster City, CA, USA) at a final concentration of 100 nM with the use of Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA). In another set of experiments, RA-FLSs were transfected with specific small interfering RNA (siRNA) that target ROBO1 mRNA using FlexiTube siRNA Premix (Qiagen, Hilden, Germany) at a final concentration of 25 nM according to the manufacturer’s protocol. AllStars Negative control siRNA (siRNA-premix, Qiagen) served as a control. Transfection efficiency of pre/anti-mir-218 and siRNA were controlled by TaqMan-based real-time polymerase chain reaction (PCR).
RNA isolation and quantitative real-time PCR analysis
A
mirVana miRNA Isolation kit was used for isolation of total RNA (Ambion/Applied Biosystems). Specific single TaqMan miRNA assays (Ambion/Applied Biosystems) were used to measure the expression levels of selected miRNA in a model light cycler 1.5 (Roche Diagnostics). Expression of the U6B small nuclear RNA (RNU6B) was used as endogenous control to normalize the data. In the analysis of the expression of specific mRNA, gene expression was quantified using SYBR Green Real-time PCR, as previously described [
27]. The primers were obtained from Takara Bio (Tokyo) and Integrated DNA Technologies (Coraville, IA, USA). The primer sequences are shown in Table
1. The amounts of loaded complementary DNA (cDNA) were normalized using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an endogenous control. For relative quantification, the comparative threshold cycle (Ct) method was used.
Table 1
SYBR green primers used for real-time PCR
ALP forward | 5′-AGCTCAACACCAACGTGGCTAA-3′ |
ALP reverse | 5′-TTGTCCATCTCCAGCCTGGTC-3′ |
RUNX2 forward | 5′-CTTTGTAGCACAAACATTGCTGGA-3′ |
RUNX2 reverse | 5′-AAAGCTGTGGTACCTGTTCTGGA-3′ |
CTNNB1 forward | 5′-CATCCTAGCTCGGGATGTTCAC-3′ |
CTNNB1 reverse | 5′-TCCTTGTCCTGAGCAAGTTCAC-3′ |
CDH11 forward | 5′-CAGGTGCTACAGCGCTCCAA-3′ |
CDH11 reverse | 5′-TTAATGTTCCCATCACCAGAGTCAA-3′ |
ROBO1 forward | 5′-CGGCAGAGTATGCTGGTCTGAA-3′ |
ROBO1 reverse | 5′-CTAGGGCACTGAGACGCATGAA-3′ |
DKK1 forward | 5′-CCAGACCATTGACAACTACCAG-3′ |
DKK1 reverse | 5′-AGGCGAGACAGATTTGCAC-3′ |
GAPDH forward | 5′-GCACCGTCAAGGCTGAGAAC-3′ |
GAPDH reverse | 5′-TGGTGAAGACGCCAGTGGA-3′ |
In silico prediction analysis of miRNA targeting genes
MiRecords (
http://c1.accurascience.com/miRecords/) was used to predict miRNA transcript targets. MiRecords is a database that combines the following miRNA target prediction tools: DIANA-microT, Micro inspector, miRanda, Mir Target2, mi Target, NB miRTar, Pic Tar, PITA, RNA hybrid, and TargetScan. Results were filtered based on the observation that a given miRNA targeted a transcript present in a minimum of five of these miRNA target prediction tools.
ELISA
Protein in cell supernatant was detected by ELISA with an ELISA kit specific for Dickkopf-1 (DKK1) according to the manufacturer’s instructions (R&D Systems). Absorption was measured at 450 nm.
miRNA and DNA microarray assay analyses
miRNA expression profiles during osteogenic differentiation were established by applying SurePrint G3 Human miRNA, 8 × 60 K (release 18.0) microarrays containing 1887 human miRNA oligonucleotide probes (Agilent Technologies, Santa Clara, CA, USA). DNA microarray analysis was performed using whole human genome DNA microarray SurePrint G3 Human Gene Expression, 8 × 60 K (ver. 2) microarrays (Agilent Technologies). All procedures were carried out according to the manufacturer’s recommendations. Microarray data were analyzed by GeneSpring software ver. 12.5.0 or 12.6.1 (Agilent Technologies). The raw signals were log2 transformed and normalized using the percentile shift normalization method: the value was set at the 90th percentile for miRNA microarray and the 75th percentile for DNA microarray.
Statistical analyses
GraphPad Prism software (GraphPad, San Diego, CA, USA) was used for statistical analyses. Normal distributions of the data were confirmed using the Kolmogorov-Smirnov test. Statistical significance was evaluated by Student’s paired t test (for parametric data) or the Wilcoxon matched-pairs signed rank test (non-parametric data) for related data. All data are expressed as the mean ± standard error of the mean (SEM). p values < 0.05 were considered significant.
Discussion
This is the first study to show that a miRNA could induce the osteogenic differentiation of FLS from RA, a bone-erosive disease. Our findings demonstrated that the expression of miR-218 was altered during osteogenic induction and most interestingly, miR-218 directly promoted the osteogenic differentiation of RA-FLS through the suppression of DKK1.
Skeletal homeostasis is a continuous process that is maintained by a balance between bone resorption by osteoclasts and bone formation by osteoblasts. In RA, bone erosion is considered to be the result of a disruption of this balance, inadequate bone formation, and an enhancement of osteoclast activity [
30]. Inadequate bone formation in RA was recently elucidated. Two studies reported that IL-6, a key pro-inflammatory cytokine of RA, decreased osteoblast proliferation and induced osteoblast apoptosis [
31,
32]. IL-6 inhibited the formation of mineralized bone nodules in an in vitro rat osteogenesis model [
32]. Another study focused on DKK1, which we observed to be a key regulator of the promotion of osteogenesis by miR-218. DKK1 expression was increased in FLS and endothelial cells in an animal model of arthritis, and TNF markedly increased the production of DKK1 from cultured FLS. In addition, serum DKK1 was elevated in patients with RA [
33].
An in vivo study by Walsh et al. using an animal model of RA demonstrated that the presence of inflammation modified osteoblast-lineage cell function, resulting in impaired osteoblast maturation and significant reduction of mineralized bone formation within the site of arthritic erosion [
34]. In clinical practice, the repair of bone erosion is uncommon but it has been demonstrated to occur. For example, 6% of patients with RA treated with adalimumab were shown to have bone repair [
35], and 1-year treatment with TNF inhibitor was shown to reduce the mean depth of erosion detected by high-resolution computed tomography [
36]. Although the mechanisms underlying the repair of bone erosion in RA have been not elucidated, the possible main mechanism might be the correction of the imbalance of bone remodeling that arises from inflammation. It is not elucidated that the osteogenic differentiation of RA-FLS, which we showed in an in vitro study, occurs in the joints in RA. However, if it does occur in the joints in RA, it is possible that proliferation of FLS contributes to bone repair by induction of osteogenic differentiation by miR-218.
Although the role of miR-218 in human disease and cell physiology has not been widely addressed, several studies of miR-218 have been reported. For example, miR-218 suppresses gastric cancer cell proliferation via regulation of angiopoietin-2 [
37], and miR-218 inhibits proliferation of glioma cells by targeting ROBO1 [
38]. Two studies revealed that miR-218 promotes osteogenic differentiation of mesenchymal stem cells through regulation of Wnt/β-catenin signaling, targeting DKK2, sclerostin, and secreted frizzled related protein 2 [
22,
39]. The difference in targets compared to our present study might be due to the difference in the types of cells examined, because miRNA may have different effects depending on cell type.
Our study suggests that the ROBO1-DKK1 axis is important for osteogenesis in RA-FLS. ROBO1 is a member of the ROBO family; it serves as a transmembrane receptor of Slit, and emerging evidence has indicated that a ROBO/Slit signaling pathway is crucial in axon guidance [
40]. In addition to axon guidance, the ROBO/Slit pathway is also involved in cell processes such as cell proliferation, cell motility, and angiogenesis [
41,
42]. The effect of the ROBO/Slit signaling pathway in osteogenesis remains unknown, but Sun et al. reported that slit2 reduced ALP expression and osteoblastic gene expression in the osteoblastic cell line MC3T3-E1 [
28]. Our present findings also showed that knockdown of ROBO1 significantly reduced DKK1 secretion from RA-FLS.
Wnt/β-catenin signaling is known as one of the important molecular cascades and is central to osteogenesis, and DKK1 is a potent inhibitor of this signaling pathway, causing deregulation of bone formation [
43]. As described above, in vivo and in vitro studies have shown an increase of DKK1 in both an arthritic animal model and in patients with RA. In fact, patients with RA with radiological progression within 2 years have been shown to have higher baseline levels of serum DKK1 compared to the patients without radiological progression [
44]. Wang et al. reported that serum DKK1 is significantly correlated with bone erosion, and that treatment with a TNF-α inhibitor or IL-1 receptor antagonist decreased serum DKK1 levels [
45]. Considering these results, the reduction of DKK1 secretion by miR-218 might provide a protective effect against RA bone erosion besides the effect of miR218 toward RA-FLS osteogenesis.
In the present study, miR-218 promoted osteogenic differentiation despite a significant decrement of miR-218 after osteoblast differentiation. A negative and positive feedback loop between microRNA and its target gene or cellular response have been observed [
46‐
48]. This crosstalk was also seen in the Wnt/β-catenin signaling pathway; miR-122 inhibits the Wnt/β-catenin signaling pathway, which negatively regulates the expression of miR-121 in glioma cells [
49]. miR-372 and miR-373 activate the Wnt/β-catenin signaling pathway by targeting Wnt/β-catenin signaling inhibitors including DKK1, and these miRs are induced by Wnt/β-catenin signaling-dependent transcription [
50]. Such crosstalk with miR-218 might be implicated in RA-FLS osteogenesis.
Conclusions
In conclusion, our study showed that the expression of miR-218 was altered during the osteogenic differentiation of RA-FLS, and that miR-218 promoted the osteogenic differentiation of RA-FLS by targeting ROBO1 and suppressing DKK1. The induction of the osteogenic differentiation of proliferated FLS in RA synovial tissue has two potential effects; the attenuation of RA disease progression derived from FLS as effector cells, and the repair of destruction of bone. Therefore, strategies to provide miR-218 to RA-FLS or to boost the cellular reservoir of miR-218 might become a therapeutic strategy for RA. This attractive hypothesis should be further tested in animal models. At the least, overexpression of miR-218 might contribute to bone repair and suppression of bone erosion by the inhibition of DKK1 secretion, which we observed herein as an effect of miR-218 in RA-FLS, and modification of the inflammatory and invasive phenotype of RA-FLS.