All animal experiments were performed as per the laboratory animal—guidelines for ethical review of animal welfare (GBT 35892.-2018) and were approved by the Jilin Agricultural University Committee on the use of live animals (Approval No: 20220311008). The sika deer was procured from Sika Deer Breeding Farm of Jilin Agricultural University. Healthy adult male sika deer (3 years old; n = 6) was used for this study. The sika deer was anesthetized by intramuscular injection of 50 mg/kg chlorpromazine hydrochloride. The sika deer was awakened at the end of experiments with 0.075 mL/kg composite awakening agent consisting of atemizole, naloxone, and flumazenib (0.90:0.95:0.26). Antler tips were collected from six anesthetized male 3-year-old sika deer. According to the previously reported method, the tissues of reserve mesenchyme and cartilage were collected from the growing antler tips at 30 days and 60 days (21). The culture of RMCs and antler chondrocytes (CC) was determined following Li et al. (22). Three biological replicate experiments for each tissue type were performed.

Flow cytometry was used to identify the surface markers of RMCs as previously described (10). In brief, the RMCs were mixed with different fluorescently labeled monoclonal antibodies, including CD34, CD45, CD73, and CD90. After washing the RMCs, we immediately determined them using a flow cytometer.

The RMCs on 24-well plates were fixed with 4% paraformaldehyde and permeabilized by 0.5% Triton X-100. After being blocked by serum, the cells were incubated overnight with primary antibodies (CD34, CD45, CD73, CD90). Then, the RMCs were incubated for 1 h with the secondary antibodies. Finally, the nucleus was stained with DAPI, and the images were analyzed using a fluorescence microscope.

We used a chondrogenic differentiation kit to determine the chondrogenic ability of RMCs. In short, 500,000 cells were placed in centrifuge tubes. Then the culture medium was changed from maintenance medium to chondrogenic medium, changing every 2 days for a course of 14 days. The final formed cartilage-like nodules were embedded in optimal cutting temperature (OCT), cut into 5-μm frozen sections, and stained with Alcian blue. MesenCult adipogenic differentiation kit and osteogenic differentiation kit were used to determine the adipogenic and osteogenic abilities of RMCs. A 12-well plate was cultured with 200,000 RMCs/well. Then the culture medium was changed from maintenance medium to adipogenic medium or osteogenic medium, changing every 3 days. After 14 days, oil red O was used to stain lipid droplets of adipocytes. After 21 days, Alizarin Red was used to stain calcium deposits of osteoblasts.

Total RNA from cells and cartilage-like nodules was isolated using RNA-quick purification kit. Isolated RNAs were reverse-transcribed into cDNAs with the use of PrimeScript^™ RT reagent kit, and cDNA was amplified with the use of TB Green^®Premix Ex Taq^™ II. The primer sequences are shown in Supplementary File S1. Relative expression levels of miRNAs and mRNAs were normalized to U6 and GAPDH internal control.

RIPA lysate (Beibokit, Shanghai, China) was utilized for extracting total proteins from cells and cartilage-like nodules. Quantification of proteins was achieved using a BCA protein kit assay. Protein was separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes, which were subsequently blocked with 5% skimmed milk. The PVDF membranes were incubated with primary antibodies on a shaker overnight at 4°C. The membranes were subsequently incubated with secondary antibodies at ambient temperature. The protein was detected using ECL detection reagents (Tanon, Shanghai, China) and signal was normalized to β-actin.

Total RNA was collected from chondrocytes and RMC-induced chondrogenic differentiation was done for 0, 7, and 14 days to construct small RNA libraries. Small RNAomics were sequenced on the Illumina Novaseq 6000. The small RNA libraries were constructed and sequenced according to the method described previously (18). The differential expression miRNAs (DE miRNAs) were obtained with p < 0.05 and fold-change cutoff of two. Online databases including miRmap, TargetScan, and miRanda were used to predict the targets of miRNAs. A Venn diagram was constructed to display overlapping interactions.

miR-145-overexpression lentivirus, miR-145-knockdown lentivirus, scrambled miRNA controls lentivirus, SOX9-overexpression lentivirus, and SOX9-knockdown lentivirus were purchased from Hanbio Biotechnology Co. Ltd. (Shanghai). The transfection process of lentivirus followed the manufacturer’s instructions. About 2 × 10⁵ RMCs were inoculated overnight in a 12-well plate and infected with lentivirus expressing either target gene lentivirus at a multiplicity of 50 after 48 h for further testing.

The binding sequences of miR-145 and SOX9 3’UTR were predicted according to targetscan database. The SOX9-3’untranslated region-wild type and SOX9-3’untranslated region-mutant type in luciferase reporter plasmids were constructed. Then, the luciferase activity was measured after co-transfection with the plasmids and miR-145 mimic or miRNA-NC in 293 T cells.

Sprague Dawley (SD) rats were procured from Experimental Animal Center of Jilin Agricultural University. The animals were kept in a standard laboratory condition. Healthy adult male rats (weight: 250 ± 30; 8 weeks old; n = 48) were used for this study. The rats were anesthetized by intraperitoneal injection of 20% urethan (1 g/kg). The rats were euthanized at the end of experiments with 20% urethan (4 g/kg). The rats were randomly divided into miR-145 mimic group, miR-145 inhibitor group, miR-145 NC group, RMCs group, PBS group, and Sham group to establish a cartilage defect model. The rat articular cartilage defect model was produced by our previous description (10). miR-145 mimic group received RMCs transfected with miR-145-overexpression lentivirus (10⁶ cells/joint). miR-145 inhibitor group received RMCs transfected with miR-145-knockdown lentivirus (10⁶ cells/joint). miR-145 NC group received RMCs transfected with miR-145-NC lentivirus (10⁶ cells/joint). The RMC group received RMCs (10⁶ cells/joint). The PBS group received 100 μL PBS. Sham control animals only underwent surgery without treatment once a week for 3 weeks. The grip strength level of rats was investigated using the grip strength test. After 8 weeks of treatment, the maximum pulling force of the hind limbs was measured five consecutive times, and the average value was recorded. At 4 and 8 weeks after surgery, the femurs of rats were collected for the evaluation of disease progression. The macroscopic evaluation of cartilage defects was made according to the Innovative Clinical Research Solutions (ICRS) criteria.

Hematoxylin and eosin (HE) staining (Bioss, Beijing, China), Safranin-O-Fast green staining (Phygene, Fuzhou, China), and IHC reagent kit (ZSGB Bio, Beijing, China) were used to assess the histological changes in sampled femurs. Briefly, the whole femurs were incubated in 4% paraformaldehyde for 1 day and decalcified for 1 month using 10% EDTA. All samples were embedded in paraffin blocks and sectioned at a thickness of 4 μm. Then, sections were stained with HE and Safranin-O-Fast green. The methods of Pineda and Wakitani were used to score the severity of the degree of cartilage defects. The parameters of Pineda score included percent filling of the defect, reconstitution of the osteochondral junction, matrix staining, and cell morphology. The range was from the best 0 to the worst 14 (23). The parameters of Wakitani score included cell morphology, matrix-staining, surface regularity, thickness of cartilage, and the integration of donor with host. The range was from the best 0 to the worst 14 (24). For immunohistochemical analysis, the tissue sections were incubated with primary antibodies overnight at 4°C. Horseradish peroxidase-labeled secondary antibody was incubated at 37°C for 60 min. The color was developed by applying 3,3′-diaminobenzidine. The sections were counterstained with hematoxylin.

All statistical analyses were conducted with SPSS 26. The means of two groups were compared using two-tailed independent Student’s t-test, and the means of multiple groups were compared by one-way analysis of variance (ANOVA). The difference was considered statistically significant at p value < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001).

Firstly, we evaluated whether the cultured RMCs exhibited the characteristics of MSCs. RMCs had spindle-shaped, fibroblast-like morphology (Figure 1A). The surface marker profile of RMCs was confirmed by immunofluorescent staining, including the expression of CD73 and CD29 (Figure 1D). Following the induction of adipogenic, chondrogenic, or osteogenic differentiation, the RMCs were positively stained for oil red O, Alcian blue, or Alizarin red, respectively (Figure 1E). Furthermore, real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) showed an increased expression of adipogenic, chondrogenic, and osteogenic genes of RMCs after designated induction (Figure 1F). These results showed that RMCs exhibited typical characteristics of MSCs. Secondly, we evaluated the basal characteristics of antler chondrocytes. They resembled paving stones in morphology (Figure 1B). Alisin blue staining of antler chondrocytes was positive (Figure 1C), and the expression of chondrogenic markers COL II and COMP was further detected by WB (Figure 1G). These results indicate that the cultured cells in this study exhibited typical chondrocyte characteristics.

miRNA sequencing was used to identify and select the differentially expressed miRNAs from chondrocytes and RMC-induced chondrogenic differentiation for 0, 7, and 14 days. A total of 475 known miRNAs were identified, of which 240 were differentially expressed (Figure 2A). The expression levels of 74 differentially co-expressed miRNAs in all comparison groups were more than two-fold different (Figure 2B). Hierarchical clustering analysis was performed to display the dynamic profiles of 74 differentially expressed miRNAs in Figure 2C. Based on our previous miRNA sequencing results on 10 different tissues of sika deer, we identified 12 candidate miRNAs related to antler growth (Figure 2D). Among them, miR-145 was well-known for regulating the proliferation and differentiation of chondrocytes. We further identified the dynamic expression of miR-145 by qRT-PCR (Figure 2E). This expression trend was consistent with miRNA sequencing. Therefore, we speculated that miR-145 might play a crucial role in the chondrogenic differentiation of RMCs.

Lentivirus was selected for RMC transfection to increase or reduce miR-145 expression. Chondrogenic differentiation was induced in RMCs with different treatments by pellet culture for 14 days. It was confirmed that miR-145 mimic group showed a significant increase in miR-145 expression, while miR-145 inhibitor group showed a significant decrease in miR-145 expression by qRT-PCR (Figure 3A). Histological analysis of pellets displayed that the staining of glycosaminoglycan in miR-145 inhibitor group using Alcian blue was stronger than control group, while the results of the miR-145 mimic group were opposite (Figures 3B,C). The expression of chondrogenic and hypertrophic markers in RMC after different treatments was detected through qRT-PCR, IF, and WB. The results consistently showed that the expression of COL II and COMP decreased in the miR-145 mimic group, while the expression of COL X increased. On the contrary, the expression of COL II and COMP increased in the miR-145 inhibitor group, while the expression of COL X decreased (Figures 3D–L). Thus, it can be concluded that knockdown of miR-145 may effectively promote the chondrocyte differentiation of RMCs, while overexpression of miR-145 may promote hypertrophic differentiation after chondrogenic induction in culture.

Using miRmap, TargetScan, and miRanda databases to identify putative target genes of miR-145, we obtained 2,615 intersecting target genes (Figure 4A). Then, the intersected target genes were compared with the differentially expressed genes in the CP_7 vs. CP_0, CP_14 vs. CP_7, CP_14 vs. CP_0, CC vs. CP_7, and CC vs. CP_14 groups, and 128 intersected candidate target genes were further identified (Figure 4B). Among them, 43 intersecting candidate target genes showed a significant negative correlation with the expression of miR-145 during chondrogenic differentiation of RMCs (Figure 4C). According to previous reports, SOX9 and KDM6A were closely related to chondrogenesis, and SOX9 was particularly important as a switch for MSCs to induce chondrogenic differentiation (25, 26). Therefore, we further validated the expression of SOX9 during chondrogenic differentiation of RMCs, and qRT-PCR results showed that the expression trend of SOX9 was opposite to that of miR-145 (Figures 4D,E). Moreover, using dual-luciferase reporting system, we found that miR-145 mimic could significantly reduce the luciferase activity of the tested SOX9 3′UTR reporter (Figure 4F). These data demonstrated that miR-145 can directly target 3′UTR of SOX9. The empty vector or recombinant lentivirus containing the entire coding sequence of miR-145 or miR-145 inhibitor was transfected into RMCs. Then, the SOX9 expression was measured after inducing chondrogenic differentiation of RMCs for 14 days using qRT-PCR, IFA, and WB. The results consistently showed that overexpression of miR-145 inhibited the expression of SOX9, while knockdown of miR-145 promoted the expression of SOX9 (Figures 4G–K). Therefore, it was speculated that miR-145 regulates the chondrogenic differentiation process in RMCs by targeting SOX9.

To verify whether miR-145 regulated the chondrogenic differentiation of RMCs through SOX9, the empty vector or recombinant lentivirus containing the entire coding sequence of miR-145, miR-145 inhibitor, SOX9, and SOX9 shRNA was transfected into RMCs. The results of Alcian blue staining demonstrated the following: miR-145 knockdown or SOX9 overexpressing increased the staining intensity; miR-145 overexpressing or SOX9 knockdown decreased the staining intensity; SOX9 knockdown reversed the promotion effect of the low expression of miR-145 on glycosaminoglycan deposition; SOX9 overexpressing rescued the inhibitory effect of miR-145 overexpression on glycosaminoglycan deposition (Figures 5A,B). Furthermore, the results of qRT-PCR, IF, and WB highlighted that the overexpression of miR-145 knockdown or SOX9 promoted the expression of COL II, COMP, and SOX9, while such overexpression inhibited the expression of COL X. On the contrary, miR-145 overexpression or SOX9 knockdown inhibited the expression of COL II, COMP, and SOX9 but promoted the expression of COL X. SOX9 knockdown reversed the promoting effect of miR-145 low expression on the chondrogenesis of RMCs, and SOX9 overexpression rescued the inhibitory effect of miR-145 overexpression on the chondrogenesis of RMCs (Figures 5C–E, 6A–H). Overall, miR-145 targeted SOX9 to inhibit the chondrogenic differentiation process in RMCs.

The role of miR-145 in regulating cartilage repair by RMCs on rats with full-thickness cartilage defect was evaluated. The in vivo models were injected with RMCs treated with miR-145 mimic, miR-145 inhibitor, and miR-NC. At the macroscopic level, 8 weeks after operation, grip strength testing shown that RMC treatment could significantly improve the hind limb grip level of rats (Supplementary Figure S1). The defect of miR-145 inhibitor group was filled with newly formed tissue, appearing smooth on the surface. The defects of RMCs group and miR-NC group were also filled with the newly formed tissue but with clear boundaries from the surrounding tissue. The defects of miR-145 mimic group and PBS group were only partially filled (Figure 7A). The ICRS scores showed that the miR-145 inhibitor group had the highest score, followed by the RMCs group and miR-NC group, while the miR-145 mimic group and PBS group had the lowest scores (Figure 7B). At the histological level, HE staining and Safranin-O-Fast green staining results revealed that the top layer cartilage of the miR-145 inhibitor group was perfectly integrated with the adjacent cartilage, and well-differentiated chondrocytes were surrounded by abundant matrix proteoglycan. In contrast, the cartilage in the miR-145 mimic group exhibited poor integration with the adjacent cartilage, characterized by a reduction in chondrocyte clusters and a loss of matrix proteoglycans (Figures 7C,E). According to the histological assessment by Pineda and Wakitani, the miR-145 inhibitor group had the lowest score, followed by the RMC group and miR-NC group, while the miR-145 mimic group and PBS group had the highest scores (Figures 7D,F). At the molecular level, the qRT-PCR results confirmed a significant change in the miR-145 expression after treatment with miR-145-modified RMCs (Supplementary Figure S2). The immunohistochemical results demonstrated that the expression of fibrous chondrocyte markers COL I and COL X was strong in the miR-145 mimic group, while the expression of SOX9 and COL II was lower (Figure 8). On the contrary, the expression of COL I and COL X was downregulated in the miR-NC and miR-145 inhibitor groups, while the expression of SOX9 and COL II was upregulated (Figure 8). This indicated that miR-145 tended to promote hypertrophy differentiation and endochondral bone formation of RMC-derived chondrocytes in vivo. In summary, the histological staining results were consistent with the protein expression of SOX9, COL II, COL X, and COL I. These data suggested that a low expression of miR-145 effectively enhanced the repair efficacy of the RMCs and reduced their hypertrophic differentiation in the cartilage defect model.