Circular RNA circANKRD36 regulates Casz1 by targeting miR‐599 to prevent osteoarthritis chondrocyte apoptosis and inflammation

Abstract Osteoarthritis (OA) is an ageing‐related disease characterized by articular cartilage degradation and joint inflammation. circRNA has been known to involve in the regulation of multiple inflammatory diseases including OA. However, the mechanism underlying how circRNA regulates OA remains to be elucidated. Here, we report circANKRD36 prevents OA chondrocyte apoptosis and inflammation by targeting miR‐599, which specifically degrades Casz1. We performed circRNA sequencing in normal and OA tissues and found the expression of circANKRD36 is decreased in OA tissues. circANKRD36 is also reduced in IL‐1β–treated human chondrocytes. FACS analysis and Western blot showed that the knockdown of circANKRD36 promotes the apoptosis and inflammation of chondrocytes in IL‐1β stress. We then found miR‐599 to be the target of circANKRD36 and correlate well with circANKRD36 both in vitro and in vivo. By database analysis and luciferase assay, Casz1 was found to be the direct target of miR‐599. Casz1 helps to prevent apoptosis and inflammation of chondrocytes in response to IL‐1β. In conclusion, our results proved circANKRD36 sponge miR‐599 to up‐regulate the expression of Casz1 and thus prevent apoptosis and inflammation in OA.

In recent years, microRNAs (miRNAs) have been found to play important roles in many inflammatory diseases including OA. 5,6 In addition, RNA sequencing (RNAseq) has identified multiple families of non-coding RNAs, such as long intergenic non-coding RNAs (lncRNA) and circular RNAs (circRNAs). 7,8 Compared with linear RNAs, circRNAs are more stable because of the non-canonical splicing without a free 3′ or 5′ end. 9 circRNA can function as miRNA sponges, whose sequences can competitively bind miRNAs to regulate the expression of target genes. 10 Our investigations focused on cirRNA ankyrin repeat domain 36 (circANKRD36), which is markedly decreased in OA patients and has been reported to be related to inflammatory response in type 2 diabetes. 11 Casz1 is an evolutionarily conserved zinc-finger transcription factor originally characterized in Drosophila. 12 Casz1 has critical functions in differentiation and tumour suppression. 13,14 Recently, it was reported that Casz1 is a regulatory protein controlling T-helper cell inflammation and immunity. 15 In this study, we demonstrated that circANKRD36 up-regulates Casz1 by targeting miR-599 to prevent chondrocyte apoptosis and inflammation in response to IL-1β treatment. Our results suggested that the regulation of Casz1 by circANKRD36-miR-599 may present a novel strategy for the treatment of OA.

| Isolation and cultivation of human chondrocytes
Chondrocytes were enzymatically isolated from human cartilage of nine patients (mean age: 65, range: 51-82 years). In short, full-thickness cartilage was minced and digested for 45 min with 0.2% pronase (Sigma-Aldrich), followed by washing and a second digest with 0.025% collagenase (Sigma-Aldrich) overnight. After washing with PBS and filtration through a 40-μm cell strainer, cells (passage 0) were cultured in serum-containing medium.
Chondrocytes were split at a confluence of 80% and used in passage 1 or 2.

| Real-time PCR analysis of mRNA
Total RNA was isolated with TRIzol reagent. The cDNA was synthesized from 2 μg of RNA by the Quantscript RT Kit (TiANGEN, KR103).
The primer sequences of for RT-PCR were as follows: circAN-

| Western blotting
In brief, the cells were harvested with a scraper and then washed once with cold PBS. The cells were then lysed in lysis buffer containing 50 Mm Tris-HCl, 250 mmol/L NaCl, 5 mmol/L EDTA, 50 mmol/L NaF, 0.1% NP-40 and 1% protease inhibitor cocktail.
Equal amounts of proteins were size-fractionated by 7.5-15% SDS-PAGE. Data collected came from at least three independent experiments.
After that, the cells were hatched (1 hour, dark, 25°C). Results were performed through fluorescence-activated cell sorting (Beckman Coulter). Results were analysed by FlowJo software (Tree Star Software).

| Reporter vectors constructs and luciferase reporter assay
The fragments of Casz1 (containing predicted binding sites) were cloned into the pMIR-REPORT Vector (Promega) to form the reporter vector Casz1 wild-type (Casz1-wt). The sequence replacing the putative binding site was named Casz1 mutant (Casz1-mut). The vector and the miR-599 were cotransfected into HEK 293T cells to test the luciferase activity by Dual-Luciferase Reporter Assay System (Promega).

| Fluorescence in situ hybridization (FISH) analysis
Slides were treated with 0.2 mol/L HCl, washed for 5 minutes, incubated in 4% pepsase for 10 minutes and washed for 5 minutes. The slides were pre-hybridized with pre-hybridization for 2 hours, hybridization using probe was performed overnight at 37°C, and slides were rinsed with 0.3% NP-40 for 30 minutes and counterstained with DAPI for 5 minutes. The images were acquired using a confocal microscope (Leica). F I G U R E 1 circANKRD36 is significantly decreased in osteoarthritis. A, Relative expression of circANKRD36 in OA (osteoarthritis) tissues (n = 36) and normal tissues (n = 9). **P < .01. B, Modified Mankin grading was performed to evaluate OA severity. (In the modified Mankin grading, abnormalities in structure (0-6 points), cellularity (0-3 points) and Safranin-O staining (0-4 points) were assessed up to a maximum score of 13 points.) The expression of circANKRD36 was negatively correlated with modified Mankin scores). C, Relative expression of circANKRD36 in chondrocytes treated with IL-1β in a dose-dependent manner (0, 0.1, 0.4, 0.7, 1.0 ng/mL). Data are presented as means ± SD (n = 3). **P < .01, ***P < .001. D, Human chondrocytes were treated with different doses of IL-1β. Apoptotic cells were analysed by FACS with Annexin V staining. Annexin V-positive cell rates were statistically analysed. Data are presented as means ± SD (n = 3). **P < .01, ***P < .001. E, Human chondrocytes were treated with different doses of IL-1β. Concentration of cytokines in culture medium was measured by ELISA Kit. Data are presented as means ± SD (n = 3). **P < .01, ***P < .001. F, The presence of circANKRD36 was validated in chondrocytes by RT-PCR. Divergent primers amplified circANKRD36 from cDNA, but not from genomic DNA. GAPDH was used as a negative control. G, RNA FISH revealed the predominant localization of circANKRD36 in the cytoplasm. circANKRD36 probes were labelled with Cy-3. Nuclei were stained with DAPI. Scale bar, 50 μm

| Statistical analysis
For all statistical tests, three or more independent experiments were performed, and data are shown as means ± SD P < .05, by unpaired Student's t test, was considered statistically significant. Data were analysed by GraphPad 6.0 (Graph Pad Software).

| Study approval
This study was approved by the Ethics Committee of Renmin Hospital of Wuhan University, Wuhan, China. Informed consent was obtained from each patient before the use of their cartilage tissue.

| circANKRD36 is decreased in OA tissue and is inhibited by IL-1β in human chondrocytes
To investigate the potential role of circANKRD36 in OA, we firstly detected the expression of circANKRD36 in articular cartilage samples from OA or normal patients. As shown in Figure 1A, the expression of circANKRD36 was significantly decreased in tissues from OA patients compared with normal tissues. Moreover, the relative expression of cir-cANKRD36 negatively correlated with modified Mankin Score which we used to evaluate OA severity of the patients according to their pathological gradings ( Figure 1B). As IL-1β has been shown to plan a critical role in the progression of OA by promoting degeneration and apoptosis of chondrocytes, we use IL-1β treatment to establish in vitro OA cell models. A marked decrease in circANKRD36 expression was found in IL-1βtreated chondrocytes ( Figure 1C). Moreover, percentage of Annexin V-positive cells and inflammatory cytokine (IL-6 and TNF-α) both increased in a dose-dependent manner upon IL-1β treatment ( Figure 1D,E).
To rule out the possibility of detecting ANKRD36 liner mRNA along with circANKRD36, we designed convergent primers to amplify ANKRD36 mRNA and divergent primers to amplify circANKRD36 using cDNA and genomic DNA (gDNA). circANKRD36 was amplified by divergent primers in cDNA but not in gDNA ( Figure 1F). Furthermore, RNA fluorescence in situ hybridization (FISH) showed that circANKRD36 mainly localized in cytoplasm ( Figure 1G). These results indicated that circAN-KRD36 might be involved in regulation of human OA.

| circANKRD36 protects chondrocytes from IL-1β apoptosis and inflammation
The decreased expression of circANKRD36 in OA tissues indicated that circANKRD36 may function as a suppressor of OA. To test this hypothesis, we performed flow cytometry to measure the apoptosis induced by IL-1β in chondrocytes with circANKRD36 overexpression or knockdown ( Figure 2A). As shown in Figure  found circANKRD36 might interact with five miRNAs ( Figure 3B).
Among these putative binding miRNAs, miR-599 has been reported to be involved in apoptosis and inflammatory response. Therefore, promoted the apoptosis induced by IL-1β, and miR-599 inhibitor blocked this effect ( Figure 3H). Statistical analysis of apoptotic rate is shown in Figure 3I. Furthermore, long-term proliferation activity of cells was tested by CCK-8 assay and indicated the consistent conclusion ( Figure 3J). Western blot of Bax/Bcl-2 and cleaved caspase 3 confirmed the anti-apoptotic effect of miR-599 inhibitor ( Figure 3K). Moreover, the pro-inflammatory cytokine production also showed consistent inflammation result with the apoptosis ( Figure 3L).
In conclusion, these results suggest that circANKRD36 acts as a sponge for miR-599 and protects apoptosis and inflammation by targeting miR-599.

| circANKRD36 promotes Casz1 expression by targeting miR-599
To investigate the potential target of miR-599, we searched for the putative gene target of miR-599 via bioinformatic analysis. We found that Casz1, a conserved transcription factor, was a direct target of miR-599 ( Figure 4A). To validate the direct targeting of Casz1 by miR-599, the wild-type (wt) Casz1-targeted sequence or a mutant variant was cloned into a dual-luciferase reporter vector. The effect of miR-599 on luciferase activity was detected in 293T cells.
In addition, we analysed the expression of miR-599/Casz1 and cir-cANKRD36/Casz1 in OA tissues and found the consistent correlation of both by Pearson ( Figure 4E). These results showed that Casz1 is a direct target of miR-599.
To  F I G U R E 4 circANKRD36 up-regulates Casz1 expression by targeting miR-599. A, Binding sites between miR-599 and Casz1 are shown. B, Luciferase activity was measured in miR-599 and Casz1 WT, and Mut cDNA plasmids infected 293T. Data are presented as means ± SD (n = 3). **P < .01. C, Relative expression of Casz1 in human chondrocytes with indicated treatment. Data are presented as means ± SD (n = 3). *P < .05, **P < .01. D, Relative expression of Casz1 and miR-599 in human chondrocytes with indicated treatment. Data are presented as means ± SD (n = 3). **P < .01, ***P < .001. E, The correlation between miR-599/Casz1 and circANKRD36/Casz1 expression level in OA tissues analysed by Pearson. F, Apoptosis protein biomarkers Bax(6A7), Bcl-2, cleaved caspase 3 and PARP1 were measured by Western blot of chondrocytes with indicated treatment. G, The inflammation proteins IL-6 and TNF-α were tested by performing Western blot and PCR of chondrocytes with indicated treatment. H, Chondrocytes were transfected as indicated and treated with 1 ng/mL IL-1β. The inflammation cytokine proteins IL-6 and TNF-α in culture medium were tested by ELISA Kit. I, Flow cytometry was conducted to measure the apoptosis rate of human chondrocytes treated as indicated. 1 ng/mL IL-1β was used to treat the cells. J, Annexin V-positive cell rates in Figure 4Iwere statistically analysed. Data are presented as means ± SD (n = 3). **P < .01. K, ACLT-induced OA mice were injected with AAVnegative control, WT AAV circANKRD36 or Mut AAV circANKRD36, and the degree of knee OA was evaluated by OARSI scoring according to the Safranin-O staining (n = 12). *P < .05. L, Catabolic enzymes (MMP-3 and MMP-13) and ECM composition (COL2A1) in mice cartilage tissues were measured by Western blot with indicated treatment

| D ISCUSS I ON
In this study, we used IL-1β-treated chondrocytes to study the molecular mechanism of OA and found circANKRD36 as a novel suppressor of chondrocyte apoptosis and inflammation. As newly discovered non-coding RNAs, circRNAs have been reported to take part in the pathogenesis of OA disease. 17 Several circRNAs, such as circSERPINE2 and circRNA-CDR1as, were found to function as miRNA sponges to regulate chondrocyte proliferation, apoptosis and inflammation. [17][18][19] circANKRD36 was recently discovered in diabetes and was reported to play an anti-inflammatory damage efficacy. 11 However, it has been reported that circANKRD36 serves as a pro-inflammatory factor in LPS-treated HaCaT cells, which were performed as a bedsore model. 20 The dichotomous effect of cir-cANKRD36 may arise with different signalling pathways in specific tissues.
Casz1 orchestrates cell specification and differentiation in many cell lineages, including neuroblasts, cardiomyocytes and lymphoid cells. 13,15 The function of Casz1 in cartilage chondrocytes has not been studied before. Our research also found a novel regulatory pathway of Casz1. It has been reported that Casz1 limits repressive histone marks and enables acquisition of permissive histone marks at several Th17 differentiation gene loci. 13 Considering its function as a transcriptional factor, we speculate that Casz1 may miR-599 has been reported to have dichotomous roles in different cell types. 21,22 miR-599 promotes apoptosis in papillary thyroid carcinoma cells by targeting Hey2, which is a transcription factor involved in cell fate and forming boundary. 21 In our study, miR-599 is proved to inhibit chondrocyte apoptosis by targeting Casz1, which is also a transcription factor. Considering this, we speculate that miR-599 may have to function to regulate different pathways by targeting various transcription factors.
In conclusion, we have identified the circANKRD36-miR-599-Casz1 axis as a novel target for the prevention and treatment of OA.
cirANKRD36 and Casz1 overexpression may remit the apoptosis and inflammation in chondrocytes upon IL-1β treatment. Besides, miR-599 inhibition may also serve as an effective therapeutic strategy for OA treatment.

CO N FLI C T O F I NTE R E S T
The authors have declared that no competing interest exists. qiongjie hu: Conceptualization (supporting).

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this article. Further details are available on request.