Circulating miR-146b and miR-27b are efficient biomarkers for early diagnosis of Equidae osteoarthritis

One of the most orthopedic problems seen in the equine is osteoarthritis (OA). The present study tracks some biochemical, epigenetic, and transcriptomic factors along different stages of monoiodoacetate (MIA) induced OA in donkeys in serum and synovial fluid. The aim of the study was the detection of sensitive noninvasive early biomarkers. OA was induced by a single intra-articular injection of 25 mg of MIA into the left radiocarpal joint of nine donkeys. Serum and synovial samples were taken at zero-day and different intervals for assessment of total GAGs and CS levels as well as miR-146b, miR-27b, TRAF-6, and COL10A1 gene expression. The results showed that the total GAGs and CS levels increased in different stages of OA. The level of expression of both miR-146b and miR-27b were upregulated as OA progressed and then downregulated at late stages. TRAF-6 gene was upregulated at the late stage while synovial fluid COL10A1 was over-expressed at the early stage of OA and then decreased at the late stages (P < 0.05). In conclusion, both miR-146b and miR-27b together with COL10A1 could be used as promising noninvasive biomarkers for the very early diagnosis of OA.


Methods
Ethical statement. The present study was approved by the Institutional Animal Care and Use Committee of Cairo University (CU-IACUC) and was performed after receiving ethical approval (approval number: CU/ II/F/4/16). The experiment was performed following ARRIVE's relevant guidelines and regulations.
Experimental study design. Based on our previous study and experimental design of equine osteoarthritis that was recently published by Yassin et al. 1 , The current study was conducted for the investigation of some biochemical, transcriptomic, and epigenetic biomarkers in synovial fluid and serum samples. Briefly, nine healthy adult male local breed donkeys, aged 3-5 years and weighing 150-200 kg, were subjected to the study following the clinical assessment. Osteoarthritis was chemically induced through intraarticular injection of a single dose of MIA (25 mg/ml) (sodium monoiodoacetate, Sigma-Aldrich, St. Louis, MO, USA) at the left radiocarpal joint. Synovial fluid samples were collected from the left radiocarpal joint, and blood samples were collected from the left jugular vein on plain tubes and then centrifuged at 1500×g for 15 min. Samples were collected at day 0 (before the OA induction), then in the 1st week, 1st, 2nd, 3rd, 5th, and 7th months after the induction. Both samples were kept at − 80 °C for further analysis.
Biochemical analysis. The concentration of total glycosaminoglycans (GAGs) and chondroitin sulfate (CS) was assessed by High-performance liquid chromatography (HPLC) on Day 0, 1st week then at the 1st, 2nd, 3rd, 5th, and 7th months post-MIA-injection as modified according to Gässler et al. 44 . Briefly; serum and synovial samples were dissolved with 1:1 a ratio for the sample to solvent (deionized water: acetonitrile 80:20%) and then centrifuged at 1957×g for 10 min. The clear supernatants were ready for analysis by the isocratic HPLC system (Agilent 1200 Series equipped with the computerized solvent delivery system and UV detector, Santa Clara, CA, USA) using Supelco Kromasil NH2 silica gel column (5 µm particle size, pore size 100 A) with flow rate 1.5 ml/min. The effluent is monitored by a UV detector at wavelength 195 nm by using GAGs and CS standard solutions (Sigma Aldrich, St. Louis, MO, USA) and the injection volume was 20 µl.
Transcriptomic and epigenetic quantitative real-time polymerase chain reaction (qRT-PCR) of miRNAs and their target genes. Total  www.nature.com/scientificreports/ extracted using miRNeasy Mini Kit (Qiagen, Hilden, Germany cat. no. 217004) according to the manufacturer's protocol with Serum/Plasma Spike-In Control for data normalization (Qiagen, cat. no. 219610). Total RNA purity and concentration were obtained using a nanodrop ND-1000 spectrophotometer. The isolated RNA was used for cDNA synthesis of target genes using Revert Aid Reverse Transcriptase (Thermo Scientific, Cat. No. EP0441, USA) according to the provided guidelines. Real-time PCR (qPCR) was performed in a total volume of 20-μl using a mixture of 1 μl cDNA, 0.5 mM of each primer (Table 1), iQ SYBR Green Premix (Thermo Scientific, Cat. No. K0221, USA). TaqMan ® microRNA Reverse Transcription kit was used for miRNA reverse transcription of the isolated RNA using (Applied Biosystems, cat. no.4366596) as manufacture protocol. The sequence-specific RT primer was provided in each TaqMan ® microRNA ready-made assay (sequence-specific primers/probes); hsa-miR-146b (assay ID: 001097) and hsa-miR-27b (assay ID:000409). Accurate and reliable results were obtained through the normalization of RT-qPCR data with spike-in control cel-miR-39 (assay ID:000200). Taqman Real-time PCR (qPCR) was performed in a total volume of 20 μl using a mixture of 1.33 μl cDNA,10 μl TaqMan ® universal PCR Master Mix II, No UNG (2X) (Applied Biosystems, Belgium, cat. no. 4440043), 1 μl TaqMan MicroRNA Assay Mix.). PCR amplification and analysis of total and miRNAs expression were achieved using Bio-Rad iCycler thermal cycler and the MyiQ real-time PCR detection system. Each assay includes triplicate samples for each tested cDNAs and no-template negative control, the expression relative to the control is calculated using the equation 2 −ΔΔCT45 .
Statistical analysis. The obtained data were statistically analyzed using the One-Way ANOVA Statistics, version 24.0 software (SPSS Inc., Chicago, IL, USA). Polynomial contrasts, post-hoc Duncan, and descriptive statistics were performed. The level of significance was set at p ≤ 0.05. The represented values are given as a standard error of the mean (SEM).

Results
Biochemical findings of total GAGS and CS concentrations in serum and synovial fluid. In serum and synovial fluid samples; the concentration of total GAGs recorded the highest significant concentration at the 3rd month post-MIA injection (Figs. 1, 2).
The concentration of CS in serum was significantly increased only at the 5th and 7th month (Fig. 3A), while in the synovial fluid; CS showed an irregular significant increase from day 0 with the highest significant level at the 3rd month (Fig. 3B).
Transcriptomic and epigenetic qRT-PCR of miRNAs and their target genes findings. Expression levels of miR-146b and miR-27b. In serum, the expression level of miR-146b was significantly up-regulated immediately at 1 week,1 month, and 3 months after MIA injection by 3.1-, 6.7-, and 2.26-folds, respectively. While at the 2nd, 5th, and 7th months; the expression level of miR-146b was not detected (Fig. 4A). In the synovial fluid sample, miR-146b was recorded to be significantly up-regulated starting from the first month till the 3rd month by 6.1-, 4.6-, and 1.76-folds respectively. While at the 5th and 7th months, the expression of miR-146b was significantly down-regulated to 0.13-and 0.16-folds, respectively (Fig. 4B). Herein, the maximum significant expression of serum and synovial fluid samples for miR-146 was observed in the 1st month.
The expression level of miR-27b in serum and synovial fluid samples was investigated throughout the experiments. The results from serum samples showed that miR-27b was significantly up-regulated rapidly 1-week post-MIA injection up to the third month by 3.6-, 4.8-, 2.1-, and 3.9-folds, respectively. Moreover, our data revealed that miR-27b expression was not detected in the 5th and 7th months (Fig. 5A). While in the synovial fluid samples, the miR-27b up-regulation was identified significantly only on the 1st and the 2nd-month postinjection by 4.5-and 4.4-fold, respectively. The 1st week and the 3rd, 5th, and 7th months exhibited significant down-regulation to 0.7-, 0.9-, 0.11-, and 0.23-folds, respectively (Fig. 5B). For serum and synovial fluid samples, the maximum significant up-regulation for miR-27b was observed in the 1st month.
The expression level of TRAF-6 and COL10A1 genes. Our study showed that the expression of serum TRAF-6 was significantly up-regulated by 471.8-fold in the third month while significant down-regulation in 1st month,2nd month, and 5th month by 0.853-, 0.88-, and 0.32-folds, respectively. Whereas, its expression was not detected in the 2nd and the 7th month (Fig. 6A). Regarding synovial TRAF-6 expression level, it was upregulated significantly at the 2nd and 5thmonth post-MIA injection by 1.4-fold and 1.1-fold respectively, and www.nature.com/scientificreports/ down-regulated significantly at the 3rd month by 0.2-fold. While, the TRAF-6 expression was not detected in 1st week,1st month, and 7th month (Fig. 6B).
In serum samples, COL10A1 was observed to be significantly up-regulated at the 2nd, 3rd, and 5th months post-MIA injection by 1.5-, 9.5-, and 2.6-folds, respectively. Also, it was noticed that the periods of downregulation were at the 1st month and 7th month by 0.93-and 0.2-folds, respectively, and were not detected in the 1st week (Fig. 7A). In synovial fluid samples, COL10A1 was immediately up-regulated following MIA injection up  www.nature.com/scientificreports/ to the 3rd month by 34.6, 35.6, and 19.14, respectively then it was down-regulated at the 5th and 7th month by 0.72-and 0.48-folds, respectively (Fig. 7B).

Discussion
Osteoarthritis is characterized by the progressive degeneration of the articular cartilage matrix, resulting in the release of its components and degradation products, which can be measured biochemically to monitor disease progression 46 .
In the present study, serum and synovial fluid total GAGs in a model of MIA-induced OA in donkeys increased, especially by the 3rd month, then declined in the 5th and 7th months. High levels of GAGs were recorded in chronic and acute joint diseases 47 ; in synovial fluid, serum, and urine of horses with OA 48,49 . While the decline in the synovial fluid total GAGs at the 5th and 7th months can be attributed to the 'floor effect' which is characterized by complete cartilage wear-out and loss of joint space as shown in the SF of patients with grade IV OA 50,51 . Also, these results came in context with our previous histopathological findings, which revealed the loss of proteoglycan from the articular cartilage 1 .
Concerning the findings of CS, serum CS was found to be significantly increased in the 5th and 7th months. In this context, serum CS was reported to be elevated in inflammatory diseases 28 and all OA patients 52 . While our results concerning synovial CS showed that its levels tended to be elevated throughout the experimental period  www.nature.com/scientificreports/ without any discrimination between the early and late osteoarthritic changes. This is consistent with a study on polo horses, which found an increase in synovial CS related to the inflammatory process of joint tissues 53 . Several miRNAs have been involved in bone and cartilage homeostasis and development as well as the progression of OA through the modulation of osteoblastogenesis, osteoclastogenesis, chondrogenesis, synovial inflammation, and cartilage degradation 54 . Some miRNAs have been shown to exhibit anti-inflammatory effects in the synovium during OA. miR-146a/b is considered one of the key regulators in the NF-κB inflammatory response pathway by binding to the 3′ UTR of their mRNA targets, TRAF-6 and IRAK-1, resulting in their inhibition 55 .miR-146b shares sequence similarities with miR-146a, so miR-146b might be up-regulated in OA and target the pro-inflammatory mediators acting as an anti-inflammatory mediator 56 .
In the present study, miR-146b was upregulated in serum at the 1st week, 1st month, and 3rd month, and in synovial fluid at the 1st, 2nd, and 3rd month, while it wasn't detected at the 2nd, 5th, and 7th months in both serum and synovial samples.
In accordance with our results, Overexpression of miR-146 at early OA was reported in human PBMCs (peripheral blood mononuclear cells) 57 , in the knee joint destabilization OA model 58 , in chondrocytes isolated from the human articular cartilage with OA 56 and knocked-down in mouse models 59 . Because of their accumulation in the synovium in the early stages of OA, these miR-146 expressed in the circulating PBMCs of OA patients may contribute to the pathogenesis of OA 57 . Meanwhile, miR-146b down-regulation was observed during the chondrogenic differentiation of human bone marrow-derived SSCs 56 .  These results are in accordance with a previous study by Zhu et al. 62 which demonstrated that TRAF-6 expression in RA synovium was significantly higher than that in OA synovium. Notably, overexpression of miR-146a in OA FLS decreased the expression of many inflammatory mediators, including TRAF-6, IRAK-1, COX-2, IL-8, MMP-13, and ADAMTS5 expression 63 . Besides, TRAF-6 was upregulated in osteoarthritis patients, which indicated that it may be associated with OA 64,65 and FJOA 61 severity and progression.
Many studies refer to miR-146a as a novel target in OA by negative feedback regulation of inflammatory responses 63,[66][67][68] or by promoting chondrocyte autophagy 69 . Other studies showed that miR-146 may contribute to OA pathogenesis by promoting the expression of VEGF and impairing the TGF-β signaling pathway via targeting Smad4 70,71 In contrast to our findings, Liu et al. reported that miR-146b was up-regulated in IL-1β-treated chondrocytes and significantly affected cell viability and matrix gene expressions in chondrocytes. The downregulation of miR-146b profoundly inhibited caspase activation and the expression of the proteolytic enzymes. Moreover, intra-articular injection of antago-miR-146b protected mice with OA from cartilage degradation as well as proteoglycan loss, suggesting that up-regulation of miR-146b can contribute to the development and progression of OA 41 . Meanwhile in our previous investigation 1 ; we found that MMPs activity and Caspase-3 expression increased, accompanied by articular cartilage degeneration and loss of proteoglycan at the 5th and 7th months where the miR-146b was downregulated in the present study.
The non-expected expression pattern of miR-146 and TRAF-6 at the late stage of OA can be explained by the fact that one miRNA could potentially control the expression of a few to several thousand genes. Conversely, each mRNA could be affected by multiple miRNAs 72 .
Few studies have investigated the expression level of miR-27b in OA chondrocytes and OA synovial fluid 29,42,73,74 . In the current study, serum, and synovial fluid miR-27b expression showed the maximum upregulation at the 1st-month post-MIA induction of OA in donkeys and then declined. In context, miR-27b was up-regulated in the synovial fluid and down-regulated in the chondrocytes following IL-1β stimulation, which could be explained as miRNAs that respond to inflammatory stimuli (IL-1β) were released into the synovial fluid from the synovium during OA in attempts from the chondrocyte to compensate for the catabolic actions 29 .
Zhou et al. demonstrated that miR-27b directly inhibits chondrocyte apoptosis and, thus, ameliorates the development of RA 74 . Besides, the expression of miR-27b was sharply down-regulated, and the production of MMP-13 protein was enhanced in normal or OA chondrocytes stimulated with IL-1β 42 .
We previously found that MMP-13 was highly up-regulated in the 2nd month of MIA OA 1 , despite profound up-regulation for synovial miR-27b. These unexpected expression patterns of MMP-13 and miRNA-27b could www.nature.com/scientificreports/ be explained as synovial miR-27b did not correspond to the MMP-13 activity in the synovial fluid, but it was correlated to the miR-27b expressed in the chondrocytes and cannot be anticipated as a perfect one-to-one correspondence 29 . In addition, miR-27b interacts with the 3-UTR of MMP-13 mRNA with imperfect complementation and down-regulates its expression at the post-transcriptional level 42 . The context scores of miR-27b recorded low values among the microRNAs predicted to target the conserved sites in the 3-UTR of MMP-13 mRNA 42 , giving the chance for other miRNAs (miR-181b) 75 and miR-33a 76 to directly or miR-145 77 indirectly increase MMP-13 expression. Further investigations are needed to clear up the correlation between synovial miR-27b and MMP-13 expression patterns. Our results demonstrated that both serum and synovial fluid miR-27b go nearly in the same direction of expression, giving an idea that serum miR-27b could be used as a predictor for the joint state and a very early biomarker for OA.
On the other hand, miR-27b-3p was reported to be highly expressed in both samples collected from OA patients and rat models. Suppression of miR-27b-3p promotes the expression of the osteogenic differentiation markers while inhibiting the expression of the adipogenic differentiation markers, inflammatory factors, cellular senescence of bone marrow mesenchymal stem cells (BMSCs), and alleviating OA pain in rats by demethylation of KDM4B (Lysine demethylase 4B) 78 . These observations may explain our previous finding that the lameness score in donkeys has decreased with the progression of OA 1 .
Concerning the COL10A1 expression level in the current study, it increased in the 2nd and 3rd months, with the highest level at the 1st month after MIA injection, then downregulated at the 5th and 7th months.
Overexpression of miR-27b upregulated the mRNA expression of the hypertrophy marker COL10A1 and the levels of COL10A1 protein in human bone marrow mesenchymal stem cells 65 . miR-27b expression was found to be inversely correlated with chondrocyte hypertrophic differentiation in postnatal rat knee articular cartilage 79 .
In agreement, COLX was significantly higher with KL2, KL3, and KL4 osteoarthritic patients compared to patients with no radiographic sign of OA (KL0). Type X collagen is normally not expressed in human healthy articular cartilage, but its expression is detected at protein and mRNA levels in human OA cartilage 80 .
Also, Fukui et al. studied the regional and zonal differences in gene expression in OA cartilage and demonstrated that the overall expression of COL10A1 was elevated but that expression showed significant local variation. Notably, the expression of type COL10A1 was higher in the less degenerated joint region than in the more degenerated areas, and this explains our finding concerning the significant downregulation of COL10A1 at the late stages of OA 81 .
It is worth mentioning that, hypertrophy markers such as MMP-13 and type X collagen have been observed in several models of OA in different species, such as mice 82 and rats 83 .
For instance, osteoarthritis involves the dysfunction of articular chondrocytes leading to cartilage degradation through chondrocyte maturation and MMP production 82 . During growth-plate chondrogenesis, the chondrocytes become hypertrophic with the expression of collagen type X, the collagen matrix is removed through the production of MMP-13, and finally, it undergoes apoptosis and is replaced by bone 84,85 .
Despite the limitation concerning the small number of animals used in our experiment, as well as the response of MIA-induced OA donkeys, which may differ from the naturally occurring OA ones, and based on our previous findings, we were able to track for the first-time changes in the expression of some biochemical, transcriptomic, and epigenetic factors in the serum and synovial fluid of Egyptian donkeys for the establishment of grading diagnostic biomarkers.
In conclusion, based on our previous classification for the stages of OA, biochemical markers (total GAGs and CS) showed their low efficiency in the early diagnosis or prognosis of OA since serum GAGs can not differentiate between the early and late stages of OA and synovial GAGs can not differentiate between the normal and late OA stages. CS in the serum cannot be used as an early diagnostic biomarker since it increases only at the late stage; also synovial CS cannot differentiate between early and late OA. Conversely, transcriptomic and epigenetic factors showed promising utility as very early noninvasive diagnostic biomarkers. Serum and synovial miR-146b showed a significant up-regulation at the very early stages of OA. To our knowledge, this is the first study investigating the expression level of miR-27b in serum during the different stages of OA, and we suppose that it would be a promising biomarker for the very early diagnosis of OA as it was up-regulated immediately 1-week post-MIA injection and nearly showed the same expression pattern as in the synovial fluid. Synovial COL10A1 showed its early, significant upregulation, representing an early biomarker for OA. All the above-mentioned findings, in combination with our previous grading for Equidae OA where the early stage of the disease is represented in the first month from MIA induction to OA, give the superiority to the serum samples to be selected as a noninvasive method for the early diagnosis of OA.

Data availability
All data generated or analyzed during this study are included in this published article. www.nature.com/scientificreports/