Materials
Table of Contents
RIPA lysis buffer, BSA, EDTA, Tris-HCl, protease inhibitor cocktail, Coomassie blue, 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO), 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate (DiD) and DiI were purchased from Dalian Meilun Biotechnology. Lecithin was purchased from Solarbio, cholesterol was purchased from Shanghai Yuanye BioTechnology, and 2-distearoyl-sn-glycero-3-phos-phoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DSPE-PEG2000) was purchased from AVT. Ferredoxin and FNR from Spinacia oleracea were purchased from Sigma-Aldrich. All other chemicals were purchased from Sigma-Aldrich unless specifically mentioned.
NTU preparation
Thylakoids were isolated from young spinach leaves using a modified method13. The obtained thylakoids were pooled, diluted and sonicated for 2 min in a Fisher Scientific FS30D bath sonicator. This step was followed by extrusion through 100-nm polycarbonate porous membranes (Whatman) using an Avanti mini extruder. The solutions were then centrifuged for 60 min at 100,000g. The pellet was resuspended in osmotic shock buffer (10 mM HEPES-KOH, 10 mM MgCl2 and 10 mM sodium l-ascorbate). NanoSight NS300 (Malvern Instruments) was used to detect the concentration (particles per ml) of NTUs. The NTUs were flash-frozen with 10% DMSO as an osmoprotectant and stored at −80 °C until use. Before use, the NTUs were stored on ice and washed two to three times in osmotic shock buffer. A similar method was applied to encapsulate gold nanoparticles into the NTUs, and equal volumes of gold nanoparticles and thylakoids were mixed and then sonicated and extruded. The chlorophyll content of the resulting solution was determined using a chlorophyll assay kit (Acmec).
Assays of NTU activity
We tested the independent photosynthesis capacity of the NTUs to synthesize ATP and NADPH in vitro using a previous method13 to construct a reaction system in vitro. In a reaction volume of 0.7 ml, NTUs were added to a reaction buffer containing 50 mM HEPES-KOH pH 7.8, 5 μM ferredoxin, 3 mM ADP, 5 mM K2HPO4, 3 mM NADP+, 10 mM sodium l-ascorbate, 10 mM KCl, 5 mM MgCl2, 1.5 μM catalase and 52 U ml−1 bovine superoxide dismutase and illuminated with red actinic light from light-emitting diodes, peaking at 630 nm at an intensity of 80 µmol photons m−2 s−1. Samples were obtained at 0, 5, 10 and 15 min, and the ATP concentration was measured using an ATP assay kit (Beyotime). For NADPH measurement, samples were illuminated with red light for 10 min, and NADPH production was measured every 5 min using a NADPH assay kit (Colorimetric, Abcam). To test the stability of NTUs, the abundance of D1 and D2 proteins (susceptible to photooxidation damage44,45,46) in NTUs over time under light and dark conditions was detected by western blotting. The prepared NTUs were illuminated at room temperature. The changes in ATP production capacity of the NTUs over time were measured under light and dark conditions.
Membrane-coated NTU preparation
Cell membranes were collected from chondrocytes according to a previously published protocol47. The membrane was suspended at 2 mg ml−1 in water. The proteins in the membrane were subsequently assessed by western blotting to detect Na+/K+-ATPase (a membrane-specific marker) and β-tubulin (a plasma-specific marker). CM solution was then added to an equal volume of NTUs for 30 min followed by sequential extrusion through 1,000, 400 and 200 nm polycarbonate porous membranes (Whatman) using an Avanti mini extruder. CM-NTUs were isolated by centrifugation at 10,000g for 5 min and then resuspended in water for further use. For CM-NTUs used in cellular NADPH experiments, different concentrations of ferredoxin (0–250 μM) were encapsulated into CM-NTUs in the extrusion process. The estimate of the ferredoxin dilution ratio (about 21-fold) was based on the ratio of the volume of CM-NTUs delivered into the cell (around 190 μm3; particle sizes of NTUs and CM-NTUs are approximately 130 nm and 216 nm, respectively, and the thickness of the cell membrane is about 6 nm48) and the total volume of the cell cytoplasm (around 3,800 μm3; cells with a diameter of 20 μm and a nucleus volume of 10% of the cell volume48). CM vesicles were prepared by extruding purified CM through the same set of porous membranes. Other cell-derived membrane-coated NTUs were prepared in a similar manner.
For LNPs, 20 mg of DSPE-PEG2000, 100 mg of lecithin and 16 mg of cholesterol were dissolved in 10 ml of CHCl3. When the organic solvent was evaporated, a thin lipid film was generated on the inner wall of the flasks. The film was hydrated and sonicated to obtain the LNPs. To produce LNP-NTUs, the lipid film was added with an equal volume of NTUs for 30 min followed by sequential extrusion through 1,000, 400 and 200 nm polycarbonate porous membranes. The LNP-NTUs were isolated by centrifugation at 10,000g for 5 min and then resuspended in water for further use.
Nanoparticle characterization
Nanoparticle size and surface zeta potential were measured by dynamic light scattering using a Malvern Zetasizer Nano ZS49. Nanoparticle morphology was observed by cryo-TEM (200 kV, FEI Tecnai G2 F20) or TEM (H-9500, Hitachi).
Proteomics analysis
To study whether the NTUs have an independent photosynthetic function of the thylakoid organelle and to analyse the biological functions (homotypic targeting and membrane fusion47,50) of CM proteins, we analysed the NTUs and CM using proteomics. Protein from the NTUs and CM was analysed according to a previously published protocol51. In brief, consecutive fractions were collected for liquid chromatography–tandem mass spectrometry analysis. To determine the biological and functional properties of all identified proteins, the identified protein sequences were analysed on the basis of GO terms.
Cell culture
RAW 264.7 mouse macrophages, a HUVEC line and a mouse fibroblast cell line (NIH/3T3) were obtained from the China Center for Type Culture Collection. For primary chondrocytes, mouse articular cartilage was dissected from the knee joint of 4-week-old male C57BL/6 mice. Human cartilage was obtained from human participants without osteoarthritis. The study design and protocol were approved by the ethics committee of Sir Run Run Shaw Hospital. Informed consent was obtained. Chondrocytes were obtained by overnight digestion of cartilage pieces with 0.025% Col II (Roche Diagnosis). To obtain primary NPCs, nucleus pulposus tissue was macroscopically isolated from 4-week-old male Sprague–Dawley rats. Next, the tissues were diced into small pieces and treated with 0.025% Col II at 37 °C for 4 h. NPCs were obtained after resuspension and filtration. To obtain primary muscle SCs, gastrocnemius muscle from 4-week-old male C57BL/6 mice was minced and digested with 5 mg ml−1 collagenase IV (Gibco-Thermo Fisher) and 1.2 U ml−1 dispase (Gibco-Thermo Fisher) at 37 °C for 45 min. SCs were obtained after resuspension and filtration. Primary cells were maintained as a monolayer in DMEM supplemented with 10% FBS. Second-passage cells were used for the subsequent experiments.
Antibodies
For western blot analysis, the following antibodies were used: D1 (Agrisera, AS05084; 1:10,000); D2 (Agrisera, AS06146; 1:5,000); β-tubulin (Abcam, ab179511, clone EPR16778; 1:1,000); Na+/K+-ATPase (Abcam, ab76020, clone EP1845Y; 1:20,000); SIRT1 (Proteintech, 13161-1-AP; 1:1,000); PGC1a (Proteintech, 66369-1-Ig, clone 1C1B2; 1:5,000); TFAM (Proteintech, 22586-1-AP; 1:5,000); NRF1 (Proteintech, 12482-1-AP; 1:500); NRF2 (Proteintech, 16396-1-AP, 1:1,000); β-actin (Proteintech, 20536-1-AP; 1:1,000); AtpB (Agrisera, AS05085; 1:2,000); anti-rabbit IgG HRP-linked secondary antibody (FDbio science, FDR007; 1:5,000); and anti-mouse IgG HRP-linked secondary antibody (FDbio science, FDM007; 1:5,000). For immunofluorescence analysis, the following antibodies were used: Col II (Proteintech, 28459-1-AP; 1:800); aggrecan (Proteintech, 13880-1-AP; 1:200); MMP13 (Proteintech, 18165-1-AP; 1:200); ADAMTS-5 (Abcam, ab246975; 1:500); iNOS (Abcam, ab178945, clone EPR16635; 1:500); MyoD (Proteintech, 18943-1-AP; 1:200); MyoG (Abcam, ab124800, clone EPR4789; 1:500); NRF2 (Proteintech, 16396-1-AP; 1:200); and CoraLite488-conjugated goat anti-rabbit IgG (Proteintech, SA00013-2; 1:500). For immunohistochemistry analysis, the following antibodies were used: Col II (Proteintech, 28459-1-AP; 1:800); aggrecan (Proteintech, 13880-1-AP; 1:200); and goat anti-rabbit IgG secondary antibody (Thermo Fisher, 31460; 1:500).
Cell viability assay
Chondrocytes were seeded into 96-well plates at a density of 4,000–5,000 cells per well and incubated for 24 h followed by the addition of the treatments at the indicated concentrations (2 × 105 NTUs per cell). After an additional 24, 48 or 72 h of incubation, cell viability was measured using a Cell Counting Kit-8 (CCK-8) assay following the manufacturer’s instructions (Dojindo).
Cellular uptake of NTUs
We sought to characterize the cross-species impact of mammalian cell membrane coating on plant-derived photosynthetic organelle interactions with macrophages and mature tissue cells (chondrocytes). RAW 264.7 mouse macrophages and mouse chondrocytes were used for the cellular uptake experiments. Cells were incubated in 12-well plates (1 × 105 cells per well) and cultured for 1 day. The NTUs were labelled with DiI before coating with LNPs or CM. Then, DiI-labelled NTUs, LNP-NTUs and CM-NTUs were used at a concentration of 2 × 105 NTUs per cell to study the cellular internalization efficiency. RAW 264.7 mouse macrophages were incubated with the NTUs for 6 h. Chondrocytes were incubated with NTUs for 1, 3 and 6 h. The cell samples were washed three times with PBS for 5 min and fixed with 4% polyformaldehyde (PFA) for 20 min. Then, the cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) for 20 min at 25 °C to label nuclei. Finally, the cells were observed by laser confocal microscopy (LCM; Nikon) or structured illumination microscopy (Nikon). The DiI fluorescence signal was measured using a Synergy H4 hybrid microplate reader (Bio Tek). A fluorescence-based assay was performed to estimate the numbers of NTUs delivered to each cell.
Membrane fusion of CM-NTUs
Chondrocytes were incubated in a 96-well plate and labelled with DiI, and the outer membrane of the CM-NTUs was labelled using DiO. The CM-NTUs were then incubated with the chondrocytes at 37 °C for 1 h. The samples were fixed with 4% PFA and stained with DAPI. The images were captured and analysed by LCM. In a parallel experiment, we first labelled the NTUs with DiO and encapsulated them with CM (labelled with DiI). Then, the CM-NTUs were incubated with chondrocytes at 37 °C for 1 h before DAPI staining and LCM observation.
Effects of endocytosis inhibitors on the cellular uptake of NTUs
We applied endocytosis-related inhibitors to study the cellular uptake pathway. Sufficient chondrocytes were seeded in 12-well plates to reach 60–70% confluency after overnight incubation. The medium was replaced with fresh medium, and four endocytosis inhibitors (chlorpromazine, filipin III, wortmannin and cytochalasin D) were subsequently separately added to the medium at concentrations of 50, 7.5, 5 or 5 µM. In particular, chlorpromazine, filipin III, wortmannin and cytochalasin D inhibited clathrin-dependent endocytosis, caveolae-dependent endocytosis, macropinocytosis and phagocytosis, respectively. After 30 min of preincubation, the cells were treated with DiI-labelled CM-NTUs (2 × 105 NTUs per cell) in the presence of the inhibitors for an additional 6 h. Finally, the cells were trypsinized, isolated by centrifugation and resuspended in PBS. The fluorescence intensity in each well was quantitatively determined by flow cytometry (FACSCalibur). FlowJo (v.10) was used for flow cytometry analysis.
Selectivity of chondrocytes taking up CM-NTUs
Equal amounts (1 × 105 cells) of chondrocytes (Hoechst 33342-labelled nuclei and DiI-labelled cell membranes), NPCs (Hoechst 33342-labelled nuclei and DiD-labelled cell membranes), SCs (DiI-labelled cell membranes), macrophages (DiD-labelled cell membranes) and fibroblasts (Hoechst 33342-labelled cell nuclei) were cultured on Petri dishes and incubated overnight. CM-NTUs (DiO-labelled NTUs) at a concentration of 2 × 105 NTUs per cell were added and incubated with these cells for 6 h. Then, flow cytometry was performed. In another experiment, five types of cell membrane-coated NTUs in equal amounts (2 × 105 NTUs per cell) were incubated with 2 × 105 chondrocytes. Owing to the limited types of staining labels, two staining schemes were used in two parallel experiments. In the first experiment, chondrocyte nuclei were labelled with Hoechst 33342. The following staining schemes were established with five different membrane-coated NTUs, including CM-NTUs (NTUs labelled with DiO), NPCM-NTUs (NTUs labelled with DiI), MM-NTUs (macrophage membrane-NTUs; NTUs labelled with DiD), SCM-NTUs (unlabelled) and FM-NTUs (fibroblast membrane-NTUs; unlabelled). These five materials were added to the culture medium and cultured with chondrocytes for 6 h (scheme 1). In the second experiment, NPCM-NTUs and MM-NTUs were not labelled, SCM-NTUs and FM-NTUs were labelled with DiI and DiD, and the remainder remained unchanged (scheme 2). Then, flow cytometry analysis was performed (LSRFortessa).
Intracellular trafficking of NTUs in chondrocytes
To demonstrate that the CM-NTUs could avoid lysosomal elimination in mammalian cells, we stained the cells with a lysosomal marker. Chondrocytes were seeded at a density of 1.5 × 105 cells per Petri dish and incubated for 24 h. The medium was replaced with fresh medium containing LNP-NTUs or CM-NTUs (NTUs labelled with DiI) at a concentration of 2 × 105 NTUs per cell. The cells were then incubated for an additional 6 h. Then, lysosomes were labelled with LysoTracker Green (200 nM) for 0.5 h according to the manufacturer’s instructions, and the nuclei were stained with DAPI for 20 min. The images were captured and analysed by LCM.
Comparison of the penetration of the LNP-NTUs and CM-NTUs in cartilage explants
Human cartilage was obtained from patients with osteoarthritis undergoing total knee replacement. The study design and protocol were approved by the ethics committee of Sir Run Run Shaw Hospital. Informed consent was obtained. Cartilage explants were extracted as solid cylinders using a sterilized 6.4-mm perforator. The explants were washed in DMEM and placed in a 96-well plate with fresh DMEM containing LNP-NTUs or CM-NTUs (NTUs labelled with DiI). After culture for 24 h, the explants were collected, immediately frozen and sectioned in a cryostat (10 μm thick). The images were captured and analysed by LCM.
Effects of EV secretion inhibitors on CM-NTU penetration
To explore the mechanism by which the CM-NTUs can achieve deep penetration, the neutral sphingomyelinase‐targeting inhibitor GW4869 (Sigma–Aldrich), which can inhibit EV secretion, was used. The prepared cartilage explants were pretreated with 10 µM GW4869 for 24 h. Afterwards, CM-NTUs (NTUs labelled with DiI) were added to the well and incubated for 24 h. Then, the explants were removed, washed with PBS and observed by LCM.
Secretion of CM-NTUs by chondrocytes
Chondrocytes were seeded at a density of 1.5 × 105 cells per Petri dish and incubated overnight. To inhibit EV secretion, chondrocytes were pretreated with 10 µM GW4869 for 24 h. Then, LNP-NTUs or CM-NTUs (NTUs labelled with DiI) were added to the dish at a concentration of 2 × 105 NTUs per cell and incubated for 1 h. Afterwards, the cells were rinsed with PBS three times and incubated with fresh medium. The culture medium was changed every few hours for 48 h. At timed intervals, the DiI fluorescence signal in the cells and culture medium was measured using a Synergy H4 hybrid microplate reader (Bio Tek).
Transcellular delivery of CM-NTUs
Chondrocytes were seeded on coverslips (1) or (2) and incubated overnight. To inhibit EV secretion, chondrocytes were pretreated with 10 µM GW4869 for 24 h. The cells on coverslips (1) were first cultured with CM-NTUs or LNP-NTUs (NTUs labelled with DiI) at a concentration of 2 × 105 NTUs per cell for 6 h. The cells on coverslips (1) were rinsed with PBS three times and then incubated with fresh cells on coverslips (2) in fresh medium for 24 h. Afterwards, the cells were washed with PBS and stained with DAPI before imaging by LCM.
ATP and NADPH enhancement in various types of cells
The general applicability of membrane-coated NTUs was studied using several types of mammalian cells, including chondrocytes, SCs, NPCs and HUVECs. The fold-changes in ATP and NADPH levels in various cell types were measured immediately after 6 h of CM-NTU, SCM-NTU, NPCM-NTU or HUVECM-NTU (2 × 105 NTUs per cell) incubation followed by 30 min of red light irradiation (80 µmol photons m−2 s−1). The intracellular ATP and NADPH concentrations were measured using assay kits, and fold-changes in ATP and NADPH were compared. To clarify the function of the NTUs in cells over time, the changes in ATP and NADPH levels over time (0–32 h) in illuminated and non-illuminated chondrocytes were measured. The degradation of NTU-derived proteins (D1, D2 and AtpB) in chondrocytes was detected by western blotting. Changes in the capacity of the NTUs to increase cellular ATP concentrations were evaluated over time under light and dark conditions. To clarify whether NTUs cause a cellular stress response, the production of ROS in cells containing NTUs under different red light illumination conditions (8.9–320 µmol photons m−2 s−1) was tested by flow cytometry (FACSCalibur) using the membrane-permeable fluorescent probe dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime).
Membrane-coated NTUs improve cell anabolism
IL-1β is highly correlated with an increase in ECM degradation and dysregulation of mitochondrial activity and associated with increased ROS generation, reduced mitochondrial biogenesis and decreased mitochondrial ATP generation in osteoarthritis24. Chondrocytes, SCs and NPCs were stimulated with IL-1β (10 ng ml−1) for 24 h followed by corresponding cell membrane-coated or membrane-coated NTUs (2 × 105 NTUs per cell) treatment for 6 h with or without red light irradiation (80 µmol photons m−2 s−1, 30 min). For chondrocytes, the levels of ECM synthesis-related proteins (Col II and aggrecan) and ECM degradation-related proteins (MMP13 and ADAMTS-5) were measured by immunofluorescence staining. JC-1 dye (Invitrogen) and MitoSOX-Red fluorescent probes (Life Technologies) were used to determine mitochondrial membrane potential and mitochondria-associated ROS production in chondrocytes, respectively, according to the manufacturer’s instructions. SIRT1, PGC1α, TFAM, NRF1 and NRF2 protein levels were detected by western blotting. The cytoplasmic ATP/ADP ratio was detected using a genetically encoded fluorescent biosensor of adenylate nucleotides (PercevalHR, a gift from G. Yellen52). For SCs, the protein levels of myogenic markers (MyoD and MyoG) were measured by immunofluorescence staining. For NPCs, the protein levels of Col II and MMP13 were measured by immunofluorescence staining. HUVECs were stimulated with 500 μM H2O2 for 24 h to induce oxidative stress damage followed by treatment with HUVEC-NTU (2 × 105 NTUs per cell) for 6 h with or without red light irradiation (80 µmol photons m−2 s−1, 30 min). Then, the protein levels of an antioxidant marker (NRF2) were measured by immunofluorescence staining.
Quantitative PCR
Total RNA was isolated from cells using a RNA kit (Qiagen). The RNA was reverse-transcribed into cDNA with reverse transcription reagents (Promega). Here, quantitative PCR (qPCR) and qPCR with reverse transcription systems (A6002, Promega) were used according to the manufacturer’s instructions. Gapdh was used as a reference gene to normalize other genes. A list of the primer sequences used for qPCR in this study is provided in Supplementary Table 5.
Western blot analysis
RIPA lysis buffer was used for protein extraction. Then, SDS–PAGE gels were used to separate the extracted protein. After electrophoresis, polyvinylidene difluoride membranes were used for protein transfer. The proteins were then blocked with nonfat milk. After incubation with primary and secondary antibodies, a chemiluminescent signal was achieved using detection reagents (enhanced chemiluminescence; Beyotime).
Transcriptomics and metabonomics study
For the transcriptomics study, 2 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using a NEBNext Ultra RNA Library Prep kit for Illumina (E7530L, NEB) following the manufacturer’s recommendations, and index codes were added to attribute sequences to each sample. Genes with P < 0.05 and absolute log2(fold changes) ≥ 1 were identified as differentially expressed genes. GO enrichment analysis of differentially expressed genes was implemented using the hypergeometric test. To determine the level of metabolic pathway enrichment, we used gene set enrichment analysis (GSEA) to compare the pathways between different groups53. The complete transcriptome of all samples was used for GSEA, and only gene sets with nominal P < 0.05 and false discovery rate q values < 0.06 were considered significant.
For the metabonomics study, cells were collected according to the manufacturer’s instructions, and the sample extracts were analysed using an LC–ESI–MS/MS system (ultra-performance liquid chromatography, ExionLC AD System; mass spectrometry, QTRAP System)54. Metabolite quantification and further analysis were performed using a multiple reaction monitoring method. Metabolites with P < 0.05 and fold change > 10% were deemed to be significant. A previously described network analysis pipeline36 was used to construct the integrated transcriptomics and metabolomics map.
Induction of osteoarthritis and intra-articular injection of CM-NTUs
Experimental osteoarthritis was induced in C57BL/6 mice. For each experiment, sex- and age-matched mice were used and randomly allocated to each experimental group. The injections were performed by blinded investigators. All animal studies were performed according to ethical regulations and protocols approved by the Sir Run Run Shaw Hospital Committee for Animal Resources and the Institutional Animal Care and Use Committee of Zhejiang Center of Laboratory Animals. All mouse experimental procedures were performed following the Regulations for the Administration of Affairs Concerning Experimental Animals approved by the State Council of People’s Republic of China. Animals were housed in groups of 4–6 mice per individually ventilated cage in a 12-h light–dark cycle (6:30–18:30 light; 18:30–6:30 dark) with constant room temperature (21 ± 1 °C) and relative humidity (40–60%). Animals had access to food and water ad libitum. A laser power metre (Zhongxi Equipment) was used to measure the efficiency of red light penetrating the mouse skin and muscle (3 mm thickness), and 3 C57BL/6 mice (8 weeks old, male) were used in the experiment.
We determined the minimum number of animals required for a specific study based on previous experiments in our group or in the published literature. Prospective power analysis was performed using G*Power analysis. The probability values of type I and type II errors were set at 0.05 and 0.20, respectively. The power analysis showed that at least 8 mice in each group were needed. We increased the number of mice to 12 per group. Three different types of mice (12-week-old male mice, 12-week-old female mice and 12-month-old male mice) were used for the treatment studies. For the surgically induced osteoarthritis mouse model, we anaesthetized mice with ketamine and xylazine and then surgically transected the anterior cruciate ligament to induce mechanical instability-associated osteoarthritis in the right knee. Control mice were sham-operated with the anterior cruciate ligament visible but not transected. Mice were randomly divided into five groups: sham surgery (sham), transection and 20 μl vehicle (PBS) treatment every 3 days without light irradiation (ACLT + vehicle + dark); transection and 20 μl vehicle treatment every 3 days with red light irradiation for 30 min every day (ACLT + vehicle + light); transection and CM-NTUs (2 × 1010 NTUs) treatment every 3 days without light irradiation (ACLT + CM-NTUs + dark); and transection and CM-NTUs (2 × 1010 NTUs) treatment every 3 days with red light irradiation in the knee joint for 30 min every day (ACLT + CM-NTUs + light). Treatments were administered by intra-articular injection into the affected joint 10 days after surgery. By quantification of ATP content in the whole femoral and tibial articular surface isolated from mice, we estimated the total number of chondrocytes in a single mouse knee joint to be about 1 × 105, and we injected 2 × 1010 NTUs (coated by CM) into each joint based on the dose of cell experiments. In the mouse cohorts (12-week-old male mice) used for ATP and NADPH analysis, tissues from the whole femoral and tibial articular surfaces were isolated and identified with n = 10 per group based on power analysis using preliminary data. At week 8 or 12, the mice were euthanized, and the joint was collected for assessment.
In vivo micro-CT image analysis
For micro-CT analysis, samples were first fixed with 4% PFA for 48 h. The knee joints were analysed using high-resolution micro-CT (Skyscan1275). We defined the region of interest to cover the entire tibial subchondral bone medial compartment. The three-dimensional structural parameters analysed included total tissue volume (containing both trabecular and cortical bone) and trabecular pattern factor.
Histological analysis and immunostaining
Knee joint samples were fixed and decalcified before histological analysis. Subsequently, the samples were dehydrated and cleared. Joints were embedded in paraffin, and 6 μm sections were taken through the entire joint at 80-μm intervals. Slides were stained with safranin-O and fast green. Each knee produced 10–12 slides for scoring by three blinded observers. Histological scoring based on the OARSI grading system (grades 0–6)55 was performed on the medial tibial plateau. The results are expressed as the mean ± 95% confidence interval of the maximum score. Immunohistochemistry was performed to assess Col II and aggrecan levels. Inflammation of the synovial membrane is directly linked to clinical symptoms, such as joint swelling, synovitis and inflammatory pain43. The synovial membranes were stained with H&E to observe the appearance of synovitis. Additionally, joint sections were used for immunofluorescence staining of iNOS (inflammation marker). ROS production in vivo was determined using dihydroethidium following previously described protocols with modifications56. In brief, 24 h before euthanasia, each mouse received a 200 μl intravenous injection of dihydroethidium at 25 mg kg−1.
Behavioural testing
Osteoarthritis-associated pain was measured using the von Frey assay and the hot-plate assay57. The two tests were performed three times before ACLT surgery and once every 2 weeks after surgery. To measure the response latencies in the hot-plate assay, a glass cylinder was used to keep mice on the hot surface of the plate, which was maintained at a temperature of 55 ± 0.5 °C. The time (in seconds) between placement of the mouse and the onset of paw shaking, licking or jumping behaviour was recorded as the index of response latency. The development of mechanical allodynia was assessed using an electronic von Frey anaesthesiometer. The withdrawal threshold was defined as the force (g) sufficient to elicit the withdrawal response. Mouse gait was analysed using an automated gait-analysis system (MGT-PR, Zhenghua Equipment) to assess motor performance. A video camera recorded from below while each mouse walked unforced across an illuminated gate platform. The software performed statistical analysis on the basis of the footprints and body-weight distribution. We performed automated gait analysis before surgery and 8 and 12 weeks after surgery.
In vivo systemic toxicity experiments
After the mice were killed, the main organs (heart, liver, kidney, lung and spleen) were collected for H&E staining to evaluate systematic pathological changes.
Statistical analysis
Statistical comparisons of two independent groups were performed using unpaired two-tailed t-test. Multiple comparisons were performed using one-way analysis of variance (ANOVA) with post-hoc Tukey test. Data based on ordinal grading systems were analysed using nonparametric Kruskal–Wallis test followed by Dunn post-hoc test. Each n indicates the number of biologically independent samples, whether mice per group or human specimens. Statistical analysis was performed using Excel and GraphPad Prism v.9.0. Statistical tests were processed using GraphPad Prism v.9.0 unless otherwise specified, and exact P values are provided in the figures whenever available (when P values are smaller than 0.0001, P < 0.0001 is shown, as the exact P value is not available in GraphPad Prism). Significance was set at P < 0.05, and the error bars represent the standard deviation for parametric data and the calculated 95% confidence intervals for nonparametric data. Data in Figs. 1c–e,g,h,m, 2a,b,h,i,l,m and 3a–f and Extended Data Figs. 1b–h, 3e, 4a–d,f,g, 5d–l, 8a and 10 were successfully replicated in two independent experiments.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.