Exosomes from dysfunctional chondrocytes affect osteoarthritis in Sprague-Dawley rats through FTO-dependent regulation of PIK3R5 mRNA stability

Subject enrolment

The present study was authorized by the Ethics Committee of the the Second Xiangya Hospital of Central South University. All patients with OA and the normal controls who participated in the present study signed a written informed consent. A total of 53 cases of cartilage tissues from patients with varying degrees of OA, and 16 cases of normal cartilage tissues, were collected from 2016 to 2018. Normal knee cartilage tissues were from patients with post-traumatic amputations and femoral condylar fractures. These normal controls were free of osteophytes or joint space narrowing on radiograph, and without osteochondral deformity of the knee. OA cartilage tissues were categorized according to the Kellgren-Lawrence (KL)¹⁴ grade. The Kellgren-Lawrence scoring system is a method for grading the severity of knee OA. According to knee radiographs, the patients were scored as grade 0 (normal knee), grade 1 (mild narrowing of the knee joint, i.e. cartilage wear), grade 2 (small osteophytes and narrowing of the joint space on radiograph), grade 3 (large amount of moderate osteophytes, narrowing of the joint space, or deformity of knee joint), and grade 4 (large amount of osteophytes on radiograph, severe narrowing of the joint space, and deformity of knee joint).

Chondrocyte culture

Knee cartilage tissues from eight-week-old rats (n = 6, 200 g (standard deviation (SD) 20), from Beijing Vital River Laboratory Animal Technology, China) were flushed twice with phosphate-buffered saline (PBS) and trimmed with a #11 sterile surgical blade on the surface of the specimen and at the edge of the abraded area. The specimen was rinsed again using PBS and detached with 0.15% type II collagenase (diluted ten times) in the tube in a shaking table (37°C, 80 rpm) for four to six hours. The tubes were then centrifuged at 1,000 rpm for eight minutes, and the supernatant was discarded. The cells were then identified by light microscopy of the cell morphology and immunofluorescence of the chondrocyte marker collagen II in the cells.

After identification, chondrocytes were treated with IL-1β (50 μg/ml, Sigma-Aldrich, USA). Changes in the expression of matrix metalloproteinase family (MMP1 and MMP9), and ECM-related proteins collagen II and aggrecan in chondrocytes, were detected by reverse transcription-quantitative (RT-q) polymerase chain reaction (PCR) and immunofluorescence.

Extraction and identification of DC-exo

DC-exo were extracted using a modified ultracentrifugation method. The cell culture medium after IL-1β induction was collected, and DC-exo were separated using Exoquick Reagent (System Biosciences, USA). The conditioned medium was incubated with Exoquick reagent (5:1) for more than 12 hours and centrifuged at 1,500 ×g for 30 minutes. The precipitated exo were re-suspended in 100 μl PBS and stored at -80°C. Nanosight LM10 particle tracking analysis, transmission electron microscopy (TEM), and western blot detection of exosome marker proteins CD63, CD9, CD81, and ALG-2 interacting protein X (ALIX) were used for DC-exo identification.

OA rat modelling

Animal experiments were reviewed and approved by the ethical review committee of the Second Xiangya Hospital of Central South University. A total of 42 Sprague-Dawley rats were randomly allocated into an OA group (n = 36) and a sham group (n = 6). After anaesthesia using intraperitoneal injection of ketamine hydrochloride (0.08 mg/100 g) and xylazine (0.04 mg/100 g), the right knee area was sterilized with alcohol. A 28-gauge needle was inserted perpendicularly to the inferior aspect of the patella, penetrated the patellar ligament, and turned proximally into the knee joint cavity until a sensation of subsidence was felt. Papain and L-cysteine were added to saline to make a 4% papain solution and 0.3 mol/l cysteine solution before use. The model rats were injected with the mixture of 4% (w/v) papain and 0.3 mol/l cysteine at 0.25 ml/kg into the right knee joint cavity.¹⁵ Injections were given on days 1, 4, 7, 10, and 13. On the fifth day after the last papain injection (day 18), 20 μg/ml DC-exo or an equivalent amount of PBS was administrated into the knee joint. Injections were given every five days for four consecutive injections. The rats were euthanized at day 40 for a modified Mankin score.¹⁶ The operation of the sham-operated group was the same as that of the model group, but an equal amount of saline was injected. Serum from rats was collected and frozen at -80°C for subsequent testing. Extended procedures can be found in the Supplementary Material. For all experiments involving animals, we have adhered to the ARRIVE guidelines and have supplied an ARRIVE checklist in the Supplementary Material.

Macrophage culture and treatment

Bone marrow-derived macrophages were produced from non-adherent bone marrow cell suspensions of rats from which chondrocytes were extracted. Bone marrow was flushed from the femur and tibia by centrifugation at 3,000 ×g for five minutes in a microcentrifuge and cultured for four hours to remove differentiated macrophages. Non-adherent cells were counted, seeded at 0.375 × 10⁶ cells/ml in 96-well plates, and cultured with 0.2 ml Dulbecco’s Modified Eagle Medium (DMEM) with 10% foetal bovine serum (FBS), 100 IU/ml penicillin/streptomycin and 20% L929. The cells were cultured for one week after changing the medium on day 3.

Macrophages in good growth condition were infected with lentiviruses to reduce the expression of OANCT or PIK3R5. The lentiviral (Lv) vectors containing short hairpin RNAs (shRNAs) were purchased from Shanghai GenePharma Bioengineering (China). Briefly, Lv-Enhance was first diluted in DMEM plus 10% FBS to obtain the appropriate Lv-Enhance (1 ×). Macrophages were seeded at 1 × 10⁶ cells/ml. After washing with PBS, the cells were cultured in medium containing Lv-Enhance (1 ×) and mixed with Lv vectors having a multiplicity of infection (MOI) of 30 to 50. To obtain effective infection, infected macrophages were kept growing for three days. If cell viability was impaired, culture media were replaced with fresh DMEM and promptly supplemented with 10% FBS. After examination under an inverted microscope, infection efficiency was verified by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and western blot.

RT-qPCR

Total RNA was extracted from the cells using the TRIzol method. RNA was extracted from cartilage tissues using RNeasy fibrous tissue kit (Qiagen company, Germany). In brief, the tissues were stabilized in RNA, later grounded with liquid nitrogen, and then lysed in lysis solution. After the addition of 70% ethanol and a centrifugation on RNeasy spin column with buffer, RNA pellet was obtained. Total RNA (1 μg) was used for reverse transcription using a Primescript RT kit (Takara Bio, Japan) following the manufacturer’s protocol. Real-time quantitative PCR was performed on a ViiA 7 real-time PCR system using SYBRPremix Ex Taq II (Takara Bio). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was regarded as a reference gene. The PCR reaction experiment of each sample was repeated three times, and the RT-qPCR data were analyzed by the 2^−ΔΔCt method. The primer sequences of these targets are provided in Supplementary Table i.

Statistical analysis

GraphPad Prism 6 (GraphPad Software, USA) software was used for data analysis. The data are reported as the mean and standard deviation (SD). The data in the present study were derived from more than three biologically independent experiments. Independent-samples t-test or one-way/two-way variance analysis (ANOVA) with Tukey’s post hoc test was employed. Statistical significance was accepted for p < 0.05.

DC-exo promote the polarization of M1 macrophages

Under the light microscope, the isolated and cultured rat chondrocytes were small, mostly triangular or irregularly shaped, and highly shaded, resembling paving stone. We used immunofluorescence to detect the chondrocyte marker collagen II, and found that this was mainly localized in the cytoplasm (Supplementary Figure a). To mimic the OA-resembling environment, we added IL-1β to the chondrocyte medium. We found that after IL-1β treatment, the levels of MMP1 and MMP9 in chondrocytes were increased significantly, and aggrecan and collagen II accumulation was significantly reduced (Supplementary Figure a). Subsequently, we further extracted exosomes from PBS- or IL-1β-treated chondrocytes. The particle size of the extracted exosomes was analyzed using Nanosight LM10 particle tracking, and we found that the extracted exosomes were all in the range of 80 to 120 nm (Supplementary Figure a). Transmission electron microscope (TEM) revealed that the exosomes of chondrocytes had distinct ellipsoidal or barrel-like structures (Supplementary Figure a). We further verified the expression of the exosomal marker proteins CD63, CD9, CD81, and ALIX using western blot. CD63, CD9, CD81, and ALIX protein expression was found to be significantly higher in exosomes relative to chondrocytes (S1C-1: p < 0.001; S1C-2: p = 0.002; S1D-1: p < 0.001; S1D-2: p = 0.001; S1E-1: p < 0.001; S1E-2: p = 0.002; S1F-1: p < 0.001; S1F-2: p < 0.001; all independent-samples t-test), indicating that exosomes were successfully extracted (Supplementary Figure a).

Subsequently, we used exosomes from PBS-treated or different doses of IL-1β-treated chondrocytes, respectively, to treat the macrophages. As the dose of exosomes from IL-1β-treated chondrocytes increased, the proportion of macrophages polarized toward the M1 type was significantly increased (1a: PBS vs 10: p = 0.006, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; 1b: PBS vs 10: p < 0.001; PBS vs 20: p < 0.001; PBS vs 40: p < 0.001, all one-way ANOVA) (Figures 1a and 1b). Exosomes from IL-1β-treated chondrocytes significantly promoted an inflammatory phenotype in macrophages, as evidenced by a significant increase in the levels of TNF-α (all p < 0.001; one-way ANOVA), IL-12 (PBS vs 10: p = 0.001, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; one-way ANOVA), and IL-6 (all p < 0.001; one-way ANOVA), and a significant decrease in the level of IL-10 (PBS vs 10: p = 0.003, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; one-way ANOVA) and TGF-β (PBS vs 10: p = 0.001, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; one-way ANOVA) (Figure 1c). Furthermore, the polarized phenotype of macrophages is closely linked to the modulation of autophagy, and decreased levels of autophagy in macrophages can promote their M1 polarization.^17,18 Thus, we used monodansylcadaverine (MDC) staining to detect the level of autophagy in macrophages, and we found that the level of autophagy in macrophages co-cultured with DC-exo was significantly reduced (PBS vs 10: p = 0.038, PBS vs 20: p = 0.003, PBS vs 40: p < 0.001; one-way ANOVA) (Figure 1d). Furthermore, we examined the expression of autophagy-related genes Atg4B, p62, and Beclin-1 in macrophages and observed that co-culture with DC-exo significantly decreased the expression of Atg4B (1e: PBS vs 10: p = 0.030, PBS vs 20: p = 0.005, PBS vs 40: p < 0.001; 1f: PBS vs 10: p = 0.006, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; two-way ANOVA) and Beclin-1 (1e: PBS vs 10: p = 0.036, PBS vs 20: p = 0.006, PBS vs 40: p < 0.001; 1f: PBS vs 10: p = 0.049, PBS vs 20: p = 0.006, PBS vs 40: p < 0.001; two-way ANOVA), and increased the expression of p62 (1e & 1f: all p < 0.001; two-way ANOVA) in macrophages (Figures 1e and 1f). Moreover, the proportion of LC3-positive particles in the cells was observed under a fluorescence microscope, and we found a significant decline in the number of positive cells in DC-exo-treated macrophages (Figure 1g). TEM was conducted to detect the formation of autophagic vesicles in the cells. The number of autophagic vesicles in the cytoplasm of DC-exo-treated macrophages was significantly reduced (Figure 1h), in which the cytoplasmic autophagosomes had a multilayered membrane structure.

Fig. 1

Dysfunctional chondrocyte-derived exosomes (DC-exo) promote M1-type polarization of macrophages. a) Flow cytometry detection of macrophage M1 polarization (CD133^-/CD86⁺) and b) M2 polarization (CD133^-/CD206⁺). c) Enzyme-linked immunosorbent assay for the detection of inflammatory factors tumour necrosis factor alpha (TNF-α), interleukin (IL)-12, IL-6, transforming growth factor beta (TGF-β), and IL-10 produced by macrophage culture. d) Changes in autophagy levels in macrophages evaluated by monodansylcadaverine staining. e) Reverse transcription-quantitative polymerase chain reaction and f) western blot detection of messenger RNA (mRNA) and protein expression of autophagy-related factors Atg4B, p62, and Beclin-1 in macrophages. g) Proportion of cells with LC3-positive granules in macrophages by fluorescence detection. h) transmission electron microscopy (TEM) observation of the formation of autophagic vesicles in macrophages. The results are representative of three independent experiments. All data are represented as mean and standard deviation, and analyzed using one-way or two-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001. AV, autophagic vesicles; DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFP-LC3, green fluorescent protein-microtubule associated protein 1 light chain 3; PBS, phosphate-buffered saline.

DC-exo exacerbate symptoms in rats with OA

To investigate the effect of DC-exo on OA, we used papain injections into the rat knee joint cavity to construct a rat model of OA. Five days after model establishment, exosomes derived from PBS- or IL-1β-treated chondrocytes were injected intra-articularly at an interval of five days until the endpoint (Figure 2a). The OA in rats was evaluated using Mankin’s modified score. It was found that OA symptoms were significantly exacerbated after DC-exo treatment (Figure 2b). The contents of pro-inflammatory factors were significantly increased (all p < 0.001; one-way ANOVA), whereas those of anti-inflammatory factors were reduced, in the knee joint cavity of rats (Figure 2c). Subsequently, we extracted knee cartilage tissues from rats to detect the level of cartilage degradation in knee tissues using toluidine blue staining. The degradation of cartilage tissues from OA rats was aggravated after DC-exo treatment (Figure 2d). We further used haematoxylin and eosin (H&E) staining to observe the knee tissues of rats. The knee joints of rats with OA had more obvious damage and immune cell infiltration after DC-exo treatment (Figure 2e). We then used immunohistochemistry to detect a significant augment in the number of inducible nitric oxide synthase (iNOS)-positive cells (both p < 0.001; one-way ANOVA) and a significant decline in the number of Arg-positive cells (both p < 0.001; one-way ANOVA) in cartilage tissues (Figures 2f and 2g). The results of terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay showed an elevation (both p < 0.001; one-way ANOVA) in the number of apoptotic cells in cartilage tissues after DC-exo treatment (Figure 2h). Finally, we measured the collagen content in rat cartilage tissues using a collagen deposition kit and observed that the collagen content in OA rat cartilage tissues was significantly reduced (both p < 0.001; one-way ANOVA) after DC-exo administration (Figure 2i). These results suggest that the inflammatory conditions in OA lead to chondrocyte dysfunction, and that the secreted exosomes will promote M1-type macrophage polarization and further exacerbate the symptoms of OA.

Fig. 2

Dysfunctional chondrocyte-derived exosomes (DC-exo) exacerbate symptoms in osteoarthritic (OA) rats. a) Schematic diagram of the rat modelling and interventions. b) Symptoms in rats assessed by Mankin score. c) Enzyme-linked immunosorbent assay for the detection of tumour necrosis factor alpha (TNF-α), interleukin (IL-12), IL-10, monocyte chemoattractant protein-1 (MCP-1), and transforming growth factor beta (TGF-β) in rat knee joint fluid. d) Cartilage degradation in rats evaluated by toluidine blue staining. e) Pathologically damaged structures in rat knee assessed by haematoxylin and eosin staining. f) Positive cell rate of inducible nitric oxide synthase (iNOS) and g) arginine in rat knee tissues by immunohistochemical staining. h) Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay detection of apoptosis in rat cartilage tissue. i) Collagen content in rat cartilage tissue assessed by the collagen deposition kit. Each group contains six rats, with each dot representing one rat. The results are representative of three independent experiments. All data are represented as mean and standard deviation, and analyzed using one-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001.

OANCT expression is significantly increased in DC-exo

To investigate the molecular composition of DC-exo, we used high-throughput sequencing to analyze and identify the DC-exo collected. A new lncRNA-AC002091.2 with the highest expression was identified, which we found to be consistent with AC002091.2 in the LncRNA database by DNAMAN comparison (6.0; LynnonBiosoft, Canada).¹⁹ It was thus named lnc-OANCT (Figure 3a). Subsequently, we further verified the expression of OANCT in IL-1β-treated chondrocytes. RT-qPCR revealed that the expression of OANCT in IL-1β-treated chondrocytes and their exosomes was much higher (both p < 0.001; independent-samples t-test) than that of PBS-treated chondrocytes (Figure 3b). Furthermore, we examined the expression and distribution of OANCT in chondrocytes and cartilage tissues of OA rats by fluorescence in situ hybridization (FISH) experiments, and we found that OANCT was distributed in both chondrocytes and cartilage tissues (Figures 3c and 3d). We then examined the expression of OANCT in macrophages after co-culture with exosomes, or in cartilage tissues of rats after exosome administration, and found that there was a significant upregulation in OANCT expression in macrophages (PBS vs 10: p = 0.008, PBS vs 20: p < 0.001, PBS vs 40: p < 0.001; one-way ANOVA) or cartilage tissues (all p < 0.001; one-way ANOVA) after exosomal treatment relative to their expression after PBS treatment. We further observed that the expression of OANCT in the knee tissues of OA rats was much higher than that of sham-operated rats (Figures 3e and 3f).

Fig. 3

The expression of osteoarthritis non-coding transcript (OANCT) is significantly increased in dysfunctional chondrocyte-derived exosomes (DC-exo). a) Heatmap of microarray screening of exosome from normal chondrocytes and interleukin beta (IL-1β)-treated chondrocytes. b) OANCT expression in chondrocytes and their exosomes detected by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). c) Fluorescence in situ hybridization (FISH) assay to detect the expression and distribution of OANCT in cartilage tissues or d) chondrocytes. e) RT-qPCR for OANCT expression in exosome-treated macrophages or f) rat cartilage tissues. g) RT-qPCR detection of OANCT expression in knee tissues from 53 osteoarthritis (OA) patients and 16 non-OA patients requiring meniscal arthroplasty due to fracture injury. h) Spearman correlation analysis of the correlation between the expression of OANCT and the patients’ Kellgren-Lawrence (KL) grade in 53 patients with OA. Each group contains six rats, with each dot representing one rat. The results are representative of three independent experiments. All data are represented as mean and standard deviation and analyzed using independent-samples t-test, and one-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001. DAPI, 4′,6-diamidino-2-phenylindole; EMP3, epithelial membrane protein 3; FXYD6, FXYD domain containing ion transport regulator 6; LILRB5, leucocyte immunoglobulin like receptor B5; MARCO, macrophage receptor with collagenous structure; NOXA1, NADPH oxidase activator 1; PBS, phosphate-buffered saline; PIK3R5, phosphoinositide-3-kinase regulatory subunit 5; PTGDS, prostaglandin D2 synthase; TNFAIP2, TNF alpha induced protein 2.

To further define the role of OANCT in the progression of OA, we clinically collected cartilage tissues from 53 OA patients and 16 non-OA patients who required meniscal arthroplasty due to fracture injury. We first examined the expression of OANCT by RT-qPCR and found that its expression was significantly higher (independent-samples t-test; p < 0.001) in the cartilage tissues of OA patients than in those of normal controls (Figure 3g). Furthermore, we classified 53 patients with OA according to the KL classification, and found a positive correlation with the severity of OA with the expression of OANCT (one-way ANOVA; I vs II: p = 0.045, I vs III: p < 0.001, I vs IV: p < 0.001, II vs III: p < 0.001, II vs IV: p < 0.001, III vs IV: p = 0.003) (Figure 3h).

Knockdown of OANCT inhibits M1-type polarization of macrophages

Based on our previous results, we believed that OANCT has a critical role in the progression of OA and the polarization of macrophages. To validate our conjecture, we transfected shRNA targeting OANCT in macrophages treated with DC-exo, and verified the transfection efficiency by RT-qPCR (one-way ANOVA; both p < 0.001) (Figure 4a). Subsequently, we used flow cytometry to detect the proportions of polarized macrophages, and found a significant decline in the proportion of macrophages that were polarized toward M1 (shScr vs shONACT-#1: p = 0.016, shScr vs shONACT-#2: p = 0.006; one-way ANOVA) and a significant elevation in the proportion of macrophages that were polarized toward M2 (both p < 0.001; one-way ANOVA) after knocking down the expression of OANCT (Figures 4b and 4c). Consistently, the secretion of TNF-α (shScr vs shONACT-#1: p = 0.006, shScr vs shONACT-#2: p = 0.003; one-way ANOVA), IL-12 (one-way ANOVA; shScr vs shONACT-#1: p = 0.018, shScr vs shONACT-#2: p = 0.003), and IL-6 (both p < 0.001; one-way ANOVA) was significantly decreased, and IL-10 (shScr vs shONACT-#1: p = 0.006, shScr vs shONACT-#2: p = 0.002; one-way ANOVA) and TGF-β, both p < 0.001; (one-way ANOVA) levels were significantly increased in macrophages after knocking down OANCT (Figure 4d). Additionally, changes in autophagy levels in macrophages were examined, and we found that knocking down OANCT further promoted autophagy in macrophages (Figures 4e to 4g).

Fig. 4

Knockdown of osteoarthritis non-coding transcript (OANCT) inhibits M1-type polarization of macrophages. Macrophages co-cultured with dysfunctional chondrocyte-derived exosomes (DC-exo) were infected with short hairpin (sh)RNAs targeting OANCT. a) Expression of OANCT in macrophages detected by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). b) Flow cytometry detection of macrophage M1 polarization (CD133^-/CD86⁺) and c) M2 polarization (CD133^-/CD206⁺). d) Enzyme-linked immunosorbent assay for the detection of inflammatory factors tumour necrosis factor alpha (TNF-α), interleukin (IL)-12, IL-6, transforming growth factor beta (TGF-β), and IL-10 produced by macrophage culture. e) Changes in autophagy levels in macrophages evaluated by monodansylcadaverine staining. f) RT-qPCR and g) western blot detection of messenger RNA (mRNA) and protein expression of autophagy-related factors Atg4B, p62, and Beclin-1 in macrophages. h) to k) Expression of: h) matrix metalloproteinase (MMP)1; i) MMP9; j) aggrecan; and k) collagen II in chondrocytes by RT-qPCR and immunofluorescence. l) Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay of the level of apoptosis in chondrocytes. The results are representative of three independent experiments. All data are represented as mean and standard deviation, and analyzed using one-way or two-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

In a previous study, it was shown that macrophages with different polarization types have different effects on the development of OA.^3,20 After co-culturing primary chondrocytes with macrophages, it was strikingly found that ECM degradation in chondrocytes was significantly reduced (4h, first graph: both p < 0.001; 4h, second graph: shScr vs shONACT-#1: p = 0.002, shScr vs shONACT-#2: p < 0.001; 4i, first graph: shScr vs shONACT-#1: p = 0.001, shScr vs shONACT-#2: p < 0.001; 4i, second graph: shScr vs shONACT-#1: p = 0.010, shScr vs shONACT-#2: p = 0.006; 4j, first graph: shScr vs shONACT-#1: p = 0.002, shScr vs shONACT-#2: p < 0.001; 4j, second graph: shScr vs shONACT-#1: p = 0.007, shScr vs shONACT-#2: p= 0.002; 4k, first graph: both p < 0.001; 4k, second graph: shScr vs shONACT-#1: p < 0.002, shScr vs shONACT-#2: p < 0.001; all one-way ANOVA) after co-culturing with macrophages with shOANCT, and the percentage of apoptosis was significantly reduced (shScr vs shONACT-#1: p = 0.002, shScr vs shONACT-#2: p < 0.001; one-way ANOVA) (Figures 4h to 4l).

Knockdown of OANCT relieves symptoms in rats with OA

To further clarify the role of OANCT in the progression of OA, we injected shRNAs targeting OANCT into the knee joint cavity of rats treated with DC-exo (Figure 5a). We first detected a significant reduction (one-way ANOVA; both p < 0.001) in the expression of OANCT in cartilage tissues using RT-qPCR (Figure 5b). Subsequently, we measured the levels of inflammatory factors in the knee joint cavity of rats using ELISA, and found that the inflammatory response in the knee joint was significantly reduced (TNF-α: both p < 0.001; IL-12: shScr vs shONACT-#1: p = 0.027, shScr vs shONACT-#2: p= 0.006; IL-10: both p < 0.001; MCP-1: shScr vs shONACT-#1: p = 0.005, shScr vs shONACT-#2: p = 0.019; TGF-β: both p < 0.001; all one-way ANOVA) after knocking down OANCT (Figure 5c). Toluidine blue staining and H&E staining further revealed that OANCT knockdown resulted in reduced cartilage degradation, pathologically damaged structures in knee cartilage tissues, and immune cell infiltration (Figures 5d and 5e). The results of immunohistochemistry exhibited a significant decline (shScr vs shONACT-#1: p = 0.022, shScr vs shONACT-#2: p = 0.002; one-way ANOVA) in the number of iNOS-positive cells and a promotion in the number of Arg-positive cells in the cartilage tissues of rats harbouring OANCT knockdown (Figures 5f and 5g). In addition, we observed a significant decline (both p < 0.001; one-way ANOVA) in the percentage of apoptosis in chondrocytes (Figure 5h). Knocking down OANCT significantly reduced the collagen deposition degradation caused by DC-exo (both p < 0.001; one-way ANOVA) (Figure 5i).

Fig. 5

Knockdown of osteoarthritis non-coding transcript (OANCT) alleviates symptoms in osteoarthritic (OA) rats. a) Schematic diagram of the rat modelling and interventions. b) Expression of OANCT in rat knee tissue by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). c) Enzyme-linked immunosorbent assay for the detection of tumour necrosis factor alpha (TNF-α), interleukin (IL)-12, IL-10, monocyte chemoattractant protein-1 (MCP-1), and transforming growth factor beta (TGF-β) in rat knee joint fluid. d) Cartilage degradation in rats evaluated by toluidine blue staining. e) Pathologically damaged structures in rat knee assessed by haematoxylin and eosin staining. f) Positive cell rate of inducible nitric oxide synthase (iNOS) and g) arginine in rat knee tissues by immunohistochemical staining. h) Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay detection of apoptosis in rat cartilage tissue. i) Collagen content in rat cartilage tissue assessed by the collagen deposition kit. Each group contains six rats, with each dot representing one rat. The results are representative of three independent experiments. All data are represented as mean and standard deviation and analyzed using one-way analysis of variance with Tukey’s post-test. ***p < 0.001. ShScr, small hairpin RNA-scramble.

OANCT regulates PIK3R5 mRNA stability by binding to the FTO protein

Our previous unpublished studies have revealed that OANCT could bind to the FTO protein. We found that OANCT was expressed in the cytoplasm and nucleus (green fluorescence), while FTO was mainly expressed in the cytoplasm (red fluorescence). OANCT and FTO co-localized in cartilage tissues and cells (Figures 6a and 6b). Subsequently, we further employed RNA pull-down and RIP experiments to verify whether OANCT could bind to the FTO protein. The level of FTO enrichment was significantly increased in macrophages overexpressing OANCT (Figures 6c and 6d), indicating that OANCT can bind to the FTO protein in macrophages. To further investigate the binding sites of OANCT to FTO, we constructed a series of partially deleted OANCT mutants. The results show that OANCT∆2 had the closest binding relation with FTO (Figures 6e and 6f), suggesting that the sequence located on the OANCT nucleotide 636 to 1,272 bp acts as a major site for FTO binding. Insignificant changes in FTO mRNA expression were found following knockdown of OANCT in macrophages (Figure 6g).

Fig. 6

Osteoarthritis non-coding transcript (OANCT) regulates phosphoinositide-3-kinase regulatory subunit 5 (PIK3R5) messenger RNA (mRNA) stability by binding to fat mass and obesity-associated protein (FTO). a) The localization of OANCT and FTO protein in rat cartilage tissues or b) chondrocytes detected by double-labelled fluorescence. c) The binding relation between OANCT and FTO protein examined by RNA pull-down and d) ribonucleoprotein immunoprecipitation (RIP) assays. e) Construction of OANCT sequences with different deletions. f) FTO protein levels bound to different mutant forms of OANCT detected by RNA pull-down assay. g) Changes in FTO mRNA expression after knockdown of OANCT in macrophages. h) Differentially expressed genes and i) enriched signalling pathways in cartilage tissues of dysfunctional chondrocyte-derived exosomes (DC-exo)-treated osteoarthritis (OA) rats analyzed by RNA-seq high-throughput sequencing. j) Phosphoinositide-3-kinase regulatory subunit 5 (PIK3R5) expression in knee tissues from 53 OA and 16 non-OA patients requiring meniscal arthroplasty due to fracture injury examined by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). k) Spearman correlation analysis of the correlation between the expression of PIK3R5 and the expression of OANCT or l) the Kellgren-Lawrence grade in 53 patients with OA. m) RT-qPCR and western blot for mRNA and protein expression of PIK3R5 in chondrocytes or n) rat cartilage tissues. o) The expression of PIK3R5 mRNA in macrophages transfected with shOANCT and treated with IZCZ-3 at different timepoints, as examined by RT-qPCR. The results are representative of three independent experiments. All data are represented as mean and standard deviation, and analyzed using independent-samples t-test, and one-way or two-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001. DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HTLV, human T-cell lymphotropic virus type 1; IgG, immunoglobulin G; IP, immunoprecipitation; KEGG, Kyoto Encyclopedia of Genes and Genomes; MMP, matrix metalloproteinase; PBS, phosphate-buffered saline; ShScr, small hairpin RNA-scramble.

To explore the downstream signalling pathways of OANCT, we further used RNA-seq high-throughput sequencing to analyze differentially expressed genes and enriched signalling pathways in cartilage tissues of DC-exo-treated OA rats. The autophagy inhibitory pathway PI3K/AKT/mammalian target of rapamycin (mTOR)-related genes were upregulated, with the PI3K receptor PIK3R5 showing the highest correlation with OANCT expression (Figures 6h and 6i). We further analyzed the clinically collected cases and found that the expression of PIK3R5 was significantly higher (p < 0.001; independent-samples t-test) in the cartilage tissues of OA patients than in non-OA patients (Figure 6j). The expression of PIK3R5 was positively correlated with the expression of OANCT and with the KL grade (I vs II: p = 0.043, I vs III: p < 0.001, I vs IV: p < 0.001, II vs III: p = 0.0234, II vs IV: p < 0.001, III vs IV: p = 0.045; one-way ANOVA) (Figures 6k and 6l). Moreover, the expression of PIK3R5 increased significantly (all p < 0.001; one-way ANOVA) after DC-exo treatment in both cartilage tissues and macrophages, but decreased significantly (all p < 0.001; one-way ANOVA) after further knockdown of OANCT (Figures 6m and 6n). We verified the mRNA half-life of PIK3R5 by treating macrophages infected with shOANCT or shScr, separately with the actinomycin D, a RNA synthesis inhibitor. It was found that the degradation rate of PIK3R5 in macrophages with OANCT knockdown was significantly higher (6 h: shScr vs shONACT-#1: p = 0.006, shScr vs shONACT-#2: p < 0.001; for the other timepoints and groups: all p < 0.001; two-way ANOVA) than that in macrophages with shScr (Figure 6o).

OANCT activates the PI3K/AKT/mTOR pathway by regulating the mRNA stability of PIK3R5 in a m6A-dependent manner

In a study by Panoutsopoulou et al,²¹ a correlation was found between FTO expression and the development of rheumatoid arthritis or OA. Since FTO acts as an eraser of m6A, we speculated that the stability of PIK3R5 mRNA was maintained due to the binding of OANCT to FTO, which reduced the demethylation modification of PIK3R5 mRNA by FTO. Therefore, we first predicted the m6A modification sites of PIK3R5 mRNA through the SRAMP website.²² A total of six m6A modification sites were noted with very high confidence (Figures 7a and 7b). Therefore, we tested the binding relationship between FTO and PIK3R5 mRNA by RIP experiments, and found that FTO could bind significantly to the 1,201 to 1,245 bp segment of PIK3R5 mRNA (p < 0.001; independent-samples t-test; Figure 7c). Therefore, we used methylated RNA immunoprecipitation (MeRIP)-qPCR to detect the level of m6A modification of PIK3R5 mRNA in macrophages. We found significant m6A modification at this site, but further treatment with DC-exo resulted in a further increase in m6A modification. Still, a significant decrease in m6A modification was noted after knockdown of OANCT expression (all p < 0.001; one-way ANOVA; Figure 7d). The above results indicated that OANCT reduced the demethylation modification of the m6A sites of PIK3R5 by FTO binding, thereby maintaining the mRNA stability of PIK3R5.

Fig. 7

Osteoarthritis non-coding transcript (OANCT) regulates the messenger RNA (mRNA) stability of phosphoinositide-3-kinase regulatory subunit 5 (PIK3R5) in a N6-methyladenosine (m6A)-dependent manner. a) Possible m6A modification sites for PIK3R5 mRNA predicted by the SRAMP website.²² b) m6A sites with high confidence. c) The binding of fat mass and obesity-associated protein (FTO) to PIK3R5 mRNA validated by ribonucleoprotein immunoprecipitation (RIP). d) FTO-mediated demethylation of modified m6A on PIK3R5 mRNA verified by MeRIP-quantitative polymerase chain reaction (qPCR). e) Extent of PI3K, AKT serine/threonine kinase 1 (AKT1), and mammalian target of rapamycin (mTOR) phosphorylation in rat cartilage tissues and f) macrophages. The results are representative of three independent experiments. All data are represented as mean and standard deviation, and analyzed using independent-samples t-test, and one-way or two-way analysis of variance with Tukey’s post-test. **p < 0.01. CDS, coding sequence; DC-exo, dysfunctional chondrocyte-derived exosomes; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IgG, immunoglobulin G; OA, osteoarthritis; PBS, phosphate-buffered saline; UTR, untranslated region.

Based on our prediction of RNA-seq high-throughput sequencing, we found that the autophagy-associated PI3K/AKT/mTOR signalling pathway was activated in DC-exo-treated rat cartilage tissues. The activation of the PI3K/AKT/mTOR signalling pathway in rat knee tissue was significantly activated by papain treatment and further promoted by DC-exo treatment, but the extents of the PI3K (all p < 0.001; two-way ANOVA), AKT1 (all p < 0.001; two-way ANOVA), and mTOR phosphorylation (sham vs OA: p < 0.001, PBS vs Dc-exo: p < 0.001, Dc-exo + shScr vs Dc-exo + shOANCT-#1: p < 0.001, Dc-exo + shScr vs Dc-exo + shOANCT-#1: p = 0.003, Dc-exo + shScr vs shOANCT-#2: p < 0.001; two-way ANOVA) were significantly downregulated by shOANCT (Figure 7e). Similarly, the PI3K/AKT/mTOR signalling pathway was significantly activated in DC-exo-treated macrophages, but levels of phosphorylated PI3K, AKT1, and mTOR were significantly reduced after knockdown of OANCT expression (all p < 0.001; two-way ANOVA; Figure 7f).

Knockdown of PIK3R5 inhibits DC-exo-promoted M1 macrophage polarization

In Figures 3 to 7, we found that exosomal OANCT from DC exacerbated the symptoms of OA by binding to FTO and inhibiting the ability of FTO of demethylation, thereby maintaining the mRNA stability of PIK3R5 and further promoting M1-type polarization of macrophages. Thus, to investigate the effects of PIK3R5 on macrophage polarization and autophagy, we further transfected shRNAs targeting PIK3R5 in DC-exo-treated macrophages and verified the transfection efficiency by RT-qPCR (both p < 0.001; one-way ANOVA) and western blot (all p < 0.001; two-way ANOVA) (Figures 8a and 8b). We found that the PI3K/AKT/mTOR signalling pathway in cells was also significantly blocked after knockdown of PIK3R5 (all p < 0.001; Figure 8b). Furthermore, our analysis revealed that after knockdown of PIK3R5, the proportion of macrophages with M1-type polarization decreased significantly (both p < 0.001; one-way ANOVA), whereas the proportion of M2-type polarization increased significantly (both p < 0.001; one-way ANOVA; Figures 8c and 8d). In addition, knocking down the expression of PIK3R5 further inhibited the secretion of pro-inflammatory cytokines in macrophages (TNF-α: shScr vs shPIK3R5-#1: p = 0.002, shScr vs shPIK3R5-#2: p < 0.001; IL-12: shScr vs shPIK3R5-#1: p = 0.004, shScr vs shPIK3R5-#2: p = 0.002; IL-6: shScr vs shPIK3R5-#1: p = 0.002, shScr vs shPIK3R5-#2: p = 0.001; one-way ANOVA), whereas the levels of the anti-inflammatory cytokines IL-10 (shScr vs shPIK3R5-#1: p = 0.025, shScr vs shPIK3R5-#2: p = 0.016; one-way ANOVA) and TGF-β1 (both p < 0.001; one-way ANOVA) were significantly increased (Figure 8e). We then used MDC staining to detect the level of autophagy in macrophages. In the presence of shPIK3R5, the level of autophagy in macrophages was significantly enhanced (both p < 0.001; one-way ANOVA), which occurred concomitant with decreased expression of p62 (Figure 8g: shScr vs shPIK3R5-#1: p = 0.004, shScr vs shPIK3R5-#2: p = 0.003; Figure 8h: both p < 0.001; two-way ANOVA) and increased expression of Atg4B (all p < 0.001; two-way ANOVA) and Beclin-1 (all p < 0.001; two-way ANOVA) (Figures 8f to 8h). Furthermore, after co-culturing chondrocytes with macrophages with shPIK3R5, the level of collagen deposition in chondrocytes increased significantly ((Figure 8i to 8k): all p < 0.001; shScr vs shPIK3R5-#1: p = 0.002, shScr vs shPIK3R5-#2: p < 0.001; Figure 8I: both p < 0.001; shScr vs shPIK3R5-#1: p = 0.003, shScr vs shPIK3R5-#2: p < 0.001; all one-way ANOVA) (Figures 8i to 8l) and the level of apoptosis in chondrocytes decreased significantly (both p < 0.001; one-way ANOVA; Figure 8m).

Fig. 8

Knockdown of phosphoinositide-3-kinase regulatory subunit 5 (PIK3R5) inhibits dysfunctional chondrocyte-derived exosomes (DC-exo)-promoted macrophage M1 polarization. Macrophages co-cultured with DC-exo were infected with short hairpin (sh)RNAs targeting PIK3R5. a) Expression of PIK3R5 in macrophages detected by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). b) PIK3R5 protein and PI3K, AKT serine/threonine kinase 1 (AKT1), and mammalian target of rapamycin (mTOR) phosphorylation levels in macrophages detected by western blot. c) Flow cytometry detection of macrophage M1 polarization (CD133^-/CD86⁺) and d) M2 polarization (CD133^-/CD206⁺). e) Enzyme-linked immunosorbent assay (ELISA) for the detection of inflammatory factors tumour necrosis factor alpha (TNF-α), interleukin (IL)-12, IL-6, transforming growth factor beta (TGF-β), and IL-10 produced by macrophage culture. f) Changes in autophagy levels in macrophages evaluated by monodansylcadaverine (MDC) staining. g) RT-qPCR and h) western blot detection of messenger RNA (mRNA) and protein expression of autophagy-related factors Atg4B, p62, and Beclin-1 in macrophages. i) to l) Expression of: i) matrix metalloproteinase (MMP)1; j) MMP9; k) aggrecan; and l) collagen II in chondrocytes by RT-qPCR and immunofluorescence. m) Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay of the level of apoptosis in chondrocytes. The results are representative of three independent experiments. All data are represented as mean and standard deviation. Data were analyzed using one-way or two-way analysis of variance with Tukey’s post-test. **p < 0.01, ***p < 0.001. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ShScr, small hairpin RNA-scramble.