All the reagents were analytical grade and purchased from commercial providers. Lipopolysaccharide (LPS), 5-carboxyfluorescein (FAM) and PCR kit were purchased from Sigma Co., United States. Human chondrocyte CHON-001 and HUVEC (Human umbilical vein endothelial cells) were purchased from ATCC. TRIzol reagent was from Invitrogen. Using a reverse transcription kit, reverse transcription was carried out (TaKaRa Company, Dalian, China). Beyotime Biotechnology Co., was the source of the cell-counting Kit-8 (CCK-8). Luciferase activity was measured with a Luciferase detection kit (Shanghai Yisheng Biotechnology Co., LTD) according to the manufacturer’s protocol. The following primary antibodies were used: GAPDH (mouse mAb, Abcam #ab8245), IL-1 (rabbit mAb, Abcam #abab9722), IL-6 (mouse mAb, Abcam #ab9324), MCP-1 (rabbit mAb, Abcam #ab214819), TNF (rabbit mAb, Abcam #ab183218). Cell lysate aliquots were separated in 10% SDS-PAGE (Beyotime, Code NO. P0012A). Rabbit secondary antibody, mouse secondary antibody, and Lipofectamine^® 3000 were purchased from Thermo Fisher Scientific. Deionized water (Millipore Milli-Q Complete) with a resistivity of 18.2 MΩ was used in all experiments. All the solutions were prepared with DEPC-treated water.

ZIF-8 nanoparticles were synthesized by a microfluidic swirl mixer chip (Supplementary Figure S1, ESI†). In a typical synthesis, 5.85 mg of Zn(NO₃)₂ was dissolved in 4 mL of DEPC-treated water as solution A. 50 μg of miRNA-200c-3p and 40 mL of dimethylimidazole solution (58.90 mg) that had been treated with DEPC were thoroughly combined and agitated for 20 min at room temperature to produce solution B. Then, a syringe pump was used to concurrently inject solutions A and B into the micromixer. Centrifugation (6,000 rpm for 5 min) was used to separate the miR-200c-3p@ZIF-8, which was then cleaned three times to get rid of extraneous reactants. The hydrothermal synthesis of ZIF-8 was based on a previously reported method. Zinc nitrate (20mg/10 mL) was mixed with 2-methylimidazole solution (400mg/10 mL) in a 25 mL bottle of cillin. The mixture was reacted in a vortex for 10 min, placed at room temperature for 30 min, centrifuged (6,000 rpm for 5 min) and washed twice to obtain ZIF-8. The size distribution, zeta potential, and stability of the ZIF-8 and miR-200c-3p@ZIF-8 were evaluated by a nanoparticle analyzer (Nano-ZS, Malvern Instruments Ltd). Transmission electron microscopy was used to analyse the morphology and particle size of ZIF-8 and miR-200c-3p@ZIF-8 (TEM, JEM-2100F). ZIF-8 and miR-200c-3p@ZIF-8 prepared on the microfluidic chip were analyzed using X-ray diffraction (XRD, Rigaku, Smart lab 9). The infrared spectra of ZIF-8 and miR-200c-3p@ZIF-8 prepared on the microfluidic chip were measured by Fourier transform infrared spectrometer (FTIR, Bruker VERTE X70). Cellular uptake efficiency was analyzed by fluorescence imaging of cells using a Nikon C2+ confocal microscope (Nikon, Japan). The transfection efficiency of miRNA-200c-3p@ZIF-8 in CHON-001 cells was quantified by RT-qPCR (ABI7500qPCR).

By using a nanodrop spectrophotometer (Nanodrop Technologies), the encapsulation efficiency (EE) of miRNA-200c-3p in the miRNA-200c-3p@ZIF-8 was measured. The formula used to determine the EE (%) was: EE (%) = 𝑊_𝐸⁄𝑊_𝑇 ×100%, where W _E were the amount of miR200c-3p contained in the miRNA-200c-3p@ZIF-8 and W_T was the amount of total miRNA-200c-3p added during preparation. We studied the miRNA-200c-3p release kinetics and degradation of miRNA-200c-3p@ZIF-8 at different pH. A specific concentration of miRNA-200c-3p@ZIF-8 (2 mg/mL) is dispersed in PBS solution with pH ranging from the pH 5.5 to 7.4. The formula used to determine the miRNA-200c-3p release percentages is release percentage (%) = M _i/M _t, where M _i is the quantity of miRNA-200c-3p which was released and M_t is the total amount of miRNA-200c-3p loaded.

To find out whether miRNA-200c-3p@ZIF-8 may condense, an agarose gel electrophoresis experiment was conducted. The complexes of miRNA-200c-3p and ZIF-8 were electrophoresed at 110 V. The identically produced polyplexes were additionally incubated for 1 h at 37°C in the presence of 10 mM DTT to test the ability of DNA to condense under reducing circumstances.

Each subject provided written informed consent, and the Nanjing Medical University Ethics Committee approved all the experimental protocols in accordance with the university’s policies on laboratory and animal care ([2020]KY004-01). Healthy human whole blood from the Affiliated Changzhou Second People’s Hospital of Nanjing Medical University was used in the hemolysis assay. We adhere to the Helsinki Declaration of 1983’s tenets. So, this approach is followed in all of the experiments presented in this study. By centrifuging blood at 3000 rpm for 5 minutes, human red blood cells (hRBCs) were extracted and diluted ten times in PBS. Thereafter, PBS and Triton X-100 dissolved in PBS were used as the corresponding positive and negative controls. Co-incubated with diluted hRBCs for 2 h at 37°C with gentle mixing, test materials made in PBS were then centrifuged at 3000 rpm for 5 minutes. Moreover, 100 μL aliquots of the resultant supernatant were put into a 96-well plate, where hemoglobin release was measured by a microplate reader using the absorbance at 540 nm (Varioskan LUX). The following calculation was used to calculate the percentage of hemolysis: R e l a t i v e   r a t e   o f   h e m o l y s i s   % = A b s 540   n m   w i t h   m i R N A − 200 c − 3 p @ Z I F − 8 − A b s 540   n m   i n   P B S   /   A b s 540   n m   w i t h   0.1   %   T r i t o n   X − 100 − A b s 540   n m   i n   P B S × 100 %

According to the instructions of Cell Counting Kit-8 (CCK-8, Sigma-Aldrich), the proliferation of CHON-001 and HUVEC cells was detected. For each group, each experiment was run in triplicate. Briefly, 1 × 10⁴ CHON-001 cells and HUVEC cells were seeded in 96-well plates and cultured for 12 h. The cells were then exposed to a succession of groups including blank, ZIF-8, PEI 25 k for 24 h 10 μL of the CCK-8 solution were added to each well, and it was then incubated at 37°C for 2 h. A microplate scanner was used to measure the absorbance of cells at 450 nm of wavelength (Varioskan LUX).

To compare the cellular uptake of miRNA-200c-3p@ZIF-8 prepared by microfluidics and Lipo3000, the fluorescence intensity of intracellular 5-carboxyfluorescein (FAM)-miRNA-200c-3p was qualitatively and quantitatively analyzed using confocal laser scanning microscope (C2+,Nikon) and flow cytometer (BD C6 PLUS). At a density of 3 × 10⁴ cells/well in a culture medium, CHON-001 and HUVEC cells were seeded in a confocal plate. Cultured cells were incubated with fluorescently labelled miRNA-200c-3p@ZIF-8 nanoparticles (blank group, miRNA-200c-3p@Lipo3000, miRNA-200c-3p@ZIF-8) in a transfection solution for 24 h before being rinsed with PBS. The cells were then cultured for an additional 24 h after being exchanged for medium that contained 10% FBS. Hoechst 33342 was used to highlight the nuclei for 30 min, after which the cells were examined under CLSM. Using flow cytometry, the cell uptake of fluorescently tagged nanoparticles was evaluated at a minimum of 1 × 10⁴ cells gated per sample. Three duplicates of each trial were carried out.

CHON-001 cell (1 × 10⁴ cells/well) was treated with FAM-labelled miRNA-200c-3p for 12 h and endolysosome escape between the Liposome 3000 and miR-200c-3p@ZIF-8 was assessed and compared by fluorescence imaging. The cells were then washed twice with PBS before being labelled with Lysotracker (red) to show endo-/lysosomes and Hoechst33342 (blue) for the nuclei. The changes of green and red fluorescence at different transfection times of 3 h and 12 h were detected by CLSM. The Pearson correlation coefficient of green fluorescence and red fluorescence was measured by the Image-processing software (ImageJ).

From the American Type Culture Collection, the human chondrocyte cell line (CHON-001) was obtained (ATCC). All cell culture additives and medium were bought from Gibco. The DMEM (Dulbecco’s Modified Eagle Medium) was used to develop the CHON-001 cell. 10% fetal bovine serum (FBS), 4 mM L-glutamine, and 1% penicillin-streptomycin were added to all medium as supplements. Following an incubation with a 0.25% trypsin digestion solution, the cells were allowed to round out and the space between them to widen. At this point, the trypsin was discarded, and the full medium containing serum was added to stop the digestion process and blow the cells into individual ones. Cells were grown at 37°C and 5% CO₂ (CLM-170B-8-NF, ESCO). CHON-001 cells were seeded in 96-well plates, 5 × 10⁴ cells per well, and stimulated with LPS (10 ng/mL) for 12 h after cultured for 6 h to simulate the OA model in vitro.

With a density of 2×10⁵ cells per well, CHON-001 cells were added to 6-well plates and left to grow overnight. ZIF-8 at a concentration of 50 μg/mL and miR-200c-3p at a concentration of 100 nM were chosen. ZIF-8, PBS, miR-200c-3p, and miR-200c-3p@ZIF-8 were all applied to CHON-001 cells during a 48-h period. The total RNA was then extracted using the Trizol reagent after cells had been washed twice with PBS (Life Technologies). FastKing RT Kit reverse-transcribed the mRNA (With gDNase). With the ABI7500qPCR System and SuperReal PreMix SYBR Green (TIANGEN), quantitative real-time PCR (qRT-PCR) was carried out (Life Technologies). Data were analyzed by using the 2^-△△Ct method. The primer sequences used in this study were as follows: upstream primer: AAT​ACT​GCC​GGG​TAA​TGA​TGG​A; downstream primer: CTC​TAC​AGC​TAT​ATT​GCC​AGC​CAC (Table S1).

With a density of 2 × 10⁵ cells per well, CHON-001 cells were planted in 6-well plates and let to incubate for 24 h. ZIF-8 at 50 μg/mL and miR-200c-3p at 100 nM were chosen. ZIF-8, PBS, miR-200c-3p, and miR-200c-3p@ZIF-8 were all applied to CHON-001 cells for 48 h. After post-transfection, CHON 001 cells were washed twice with ice-cold PBS, and Total Extraction Sample Kit was used to capture the proteins from the cells (KeyGEN BioTECH, Nanjing, China). Supernatants were then obtained after being centrifuged for 10 min at 12,000 rpm. The BCA Protein Assay Kit was used to evaluate the protein content of sample proteins (KeyGEN, China). Then, 20 μg of proteins from all the samples were analyzed using SDS-PAGE before being electroblotted onto PVDF membranes from Merck Millipore in the United States. After being blocked for 2 hours with 5% dried milk (made with TBST solution), PVDF membranes were treated with the matching primary antibodies at 4°C overnight, rinsed in TBST, and then incubated for 2 hours with the peroxidase-coupled secondary antibody. An automatic chemiluminescence imaging system was used to detect particular proteins (ImageQuant LAS 4000).

Statistical analysis of the data was performed using SPSS 16.0 (SPSS Inc.). All results of at least 3 parallel samples were expressed as mean ± standard deviation (SD). Correlation analysis using Pearson statistical method, statistical significance was present as **p < 0.01; ***p < 0.001.

The TEM images showed that the ZIF-8 had a clear polyhedron form with dimensions of 103 ± 4 nm (Figure 1A), similar to the miR-200c-3p@ZIF-8 with an average size of about 121 nm and a polyhedron form (Figure 1B). After miRNA loading, the polydispersity index (PDI) of miR-200c-3p@ZIF-8 was 0.32, slightly higher than bare ZIF-8 (0.22, Supplementary Figure S2, ESI†). ZIF-8 was prepared using a hydrothermal method as reported elsewhere with modifications. ZIF-8 can be prepared by mixing zinc nitrate solution with dimethylimidazole solution, reacting on a vortex for 10 min and leaving at room temperature for 30 min. UV-vis absorption spectroscopy was employed to measure the quantity of miR-200c-3p in the supernatant before and after encapsulation (Supplementary Figure S3, ESI†), and it was clear that most of the miR-200c-3p has been loaded into the MOFs with an encapsulation rate of 60.35% ± 4.61%, much higher than that of the hydrothermal control (36.67% ± 2.14%, Figure 1C). Change the ratio of miRNA and ZIF-8 to 1:20, 1:40, 1:60, and found that the particle size of the prepared miR-200c-3p@ZIF-8 has little change (Supplementary Figure S4, ESI†). However, different flow rate ratios have a greater impact on the particle size of miR-200c-3p@ZIF-8 (Supplementary Figure S5, ESI†). Fourier transform infrared (FT-IR) results also demonstrated the successful complexation of miR-200c-3p with ZIF-8 (Supplementary Figure S6, ESI†). The powder’s X-ray diffraction (XRD) patterns of both ZIF-8 and miR-200c-3p@ZIF-8 displayed the same distinctive peaks to that of the simulated one (Figure 1D), demonstrating the formation of the intended ZIF-8, and miRNA is loaded by physisorption. ZIF-8 has a large specific surface area of 1,266 m² g^-1 (Supplementary Figure S7, ESI†), which is beneficial to loading miR-200c-3p. In comparison with the X-ray photoelectron spectrum of ZIF-8, there were two obvious new peaks of O1s (531 eV) and P2p (131 eV) in the spectrum of miR-200c-3p@ZIF-8, confirming the successful loading of miR-200c-3p on ZIF-8 (Figure 1E). After miRNA loading, the Zeta-potential was decreased from 23.6 ± 1.4 to 10.1 ± 1.5 mV (Figure 1F), confirming the adsorption of negatively charged miR-200c-3p on positively charged ZIF-8 again. Additionally, the miR-200c-3p@ZIF-8 was stable as evidenced by the fact that, after 4-weeks storage, neither the particle size nor the PDI substantially increased (Supplementary Figure S8, ESI†). Agarose gel electrophoresis was used to examine the binding preference of miR-200c-3p in ZIF-8, and the results showed that the miR-200c-3p was fully bound when the weight ratio was greater than 1:20 (Lanes 4, 5, and 6 in Figure 1G), indicating that miR-200c-3p was trapped inside ZIF-8 without release.

After incubation at pH = 7.4 for 6 h, less than 10% of miR-200c-3p can be released, while these values were enhanced to 50%, and 82%, at pH = 6.5, and 5.5, respectively (Figure 1H). That is because the coordination bond between zinc and imidazole could be dissociated in an acidic environment, leading to the broken down of ZIF-8, and subsequent miRNA release.

As shown in Figures 2A, B, the hemolytic effects of ZIF-8 and miR-200c-3p@ZIF-8 were much lower than the “Gold-standard” transfection agent PEI 25 K. Even at a relatively high concentration of 1 mg/mL, there is no obvious blood cell lysis (less than 20%) in ZIF-8 and miR-200c-3p@ZIF-8 groups. ZIF-8, miR-200c-3p, and miR-200c-3p@ZIF-8 cytotoxicity was examined using the live/dead staining (Figure 2C) and Cell Counting Kit-8 (CCK-8) assay (Figure 2D) on the CHON-001 and HUVEC cell lines, with PEI 25 k serving as the reference. It was clear that at concentrations lower than 100 μg/mL, ZIF-8, miR-200c-3p and miR-200c-3p@ZIF-8 did not impair cell viability and encouraged cell proliferation.

The cell uptake and in vitro transfection efficiency of miR-200c-3p@ZIF-8 were quantitatively analyzed by flow cytometry and RT-qCR. The positive control for fluorescence expression was liposome 3000, while the negative control for fluorescence expression was cells that had not been transfected. Compared with the original solution, the fluorescence intensities of Lipo3000 group and miR@ZIF-8 group were only 86.57% and 90.24% of the original intensity, respectively, when CHON-001 cells were transfected with 5-carboxyfluorescein (FAM)-labeled miR-200c-3p for 24 h (Figure 2E). Following that, chondrocyte cultures were introduced to miR-200c-3p@ liposome3000 and miR-200c-3p@ZIF-8. From the confocal images (Figure 2F), it was obvious that the experimental group and the positive control group’s cells displayed a clear bright green emission in contrast to the control group. RT-qCR results showed that compared with the control group, the expression of miR-200c-3p in the transfection group miR-200c-3p@ZIF-8 and Lipo3000 was significantly increased (Supplementary Figure S9, ESI†).

As shown in Figure 3A, the green fluorescence FAM-miR-200c-3p@ZIF-8 was mostly co-located with red fluorescence endosomes/lysosomes after incubation for 3 h, indicating that miR-200c-3p@ZIF-8 was internalized into cells through the endocytic pathway. Using ImageJ software, the fluorescence colocalization between endosome/lysosome (red) and FAM-labeled miRNA (green) was quantified to determine the endosome escape efficiency (Figure 3B), and the Pearson correlation coefficients were calculated as 0.98898 and 0.95937 in miR-200c-3p@ZIF-8 and miR-200c-3p@Lipo3000 groups. When the incubation period was increased to 12 h, the green emission of FAM-labeled miR-200c-3p increased in both groups while the red fluorescence decreased due to the reduction of intracellular lysosomes caused by miR-200c-3p escape (Figure 3A).

After CHON-001 cells were exposed to various amounts of LPS (0, 1, 5, 10 ng/mL) for 24 h. According to CCK-8 experiments, cell viability declined as LPS concentration increased, which was decreased to 50% at 10 ng/mL LPS concentration (Figure 4A). In subsequent experiments, CHON cells were treated with 10 ng/mL LPS for 12 h to establish an osteoarthritis model. According to the RT-QPCR data, the expression level of miR-200c-3p decreased steadily with the increase of LPS concentration (Figure 4B), indicating that the expression level of miR-200c-3p was decreased in LPS-stimulated chondrocytes.

The expression of associated proteins was examined by Western blotting, after LPS, LPS + NC@ZIF-8, LPS + miR-200c-3p, and LPS + miR-200c-3p@ZIF-8 treatments were applied to CHON-001 cells. LPS can drive CHON-001 cells to express a lot of inflammatory factors, as seen in Figure 4C The expression of TNF protein was slightly decreased in the miR-200c-3p group, whereas the expression of IL-1, IL-6, MCP-1, and TNF protein was significantly decreased in the miR-200c-3p@ZIF-8 group.