Objective:To construct a microfluidic organ-on-a-chip and evaluate its capability in simulating subchondral bone remodeling during the progression of osteoarthritis.Methods:The chip′s main body was designed based on the microfluidic technology and cell co-culture technique. MC3T3-E1 cells were cultured adherently within the cell seeding micro-chamber, with the culture medium perfused at a flow rate of 0.5 ml/min at the bottom of the micro-chamber. Evaluation metrics were as follows: (1) Assessment of the microfluidic organ-on-a-chip: The growth culture medium was perfused and simulation experiments were conducted to test the concentration differences and equilibrium times of the fluid inside and at the bottom of the cell seeding micro-chamber at various time points; live-dead staining was performed to observe the biocompatibility of cells cultured continuously for 3 days and 7 days at a set flow rate, which was divided into 3-day and 7-day groups. (2) Osteogenic potential of the microfluidic organ-on-a-chip: The osteogenic induction medium was perfused, and ALP staining and PCR were performed to compare the number of the black alkaline phosphatase (ALP)-positive cells and the expression levels of osteogenesis-related marker genes including osteoblast-specific transcription factor 2 (RUNX2), type I collagen (COL1A1), bone morphogenetic protein-2 (BMP-2), and osteocalcin (OCN) under static, 3-day and 7-day perfusion conditions, which was divided into static non-induced, static-induced and perfusion-induced groups. (3) Characterization of morphology and size, and biocompatibility of extracellular vesicles (EVs) of three osteoblast subtypes: Three different subtypes of osteoblasts were obtained [endothelial-type osteoblasts (EnOB)-EVs, stromal-type osteoblasts (StOB)-EVs and mineralizing-type osteoblasts (MinOB)-EVs]. Their morphology and size were obtained through transmission electron microscopy and particle size analysis. Growth medium containing EVs of three different cell subtypes was perfused, and cell proliferation/apoptosis assay was performed to compare the biocompatibility of the addition of different EVs concentrations (1, 1.25, 2.5, and 5 μg/ml) for 24 hours, which was categorized into the EnOB-EVs group, StOB-EVs group and MinOB-EVs group. (4) Osteogenic effect of EVs from three subtypes of osteoblasts: Osteogenic induction media containing EVs from three different osteoblast subtypes were perfused for 3 days, and ALP staining and PCR were performed to compare the number of black ALP-positive cells and the expression levels of osteogenesis-related marker genes including RUNX2, COL1A1, BMP-2, and OCN, which was divided into non-EVs group, EnOB-EVs group, StOB-EVs group and MinOB-EVs group.Results:(1) Evaluation of the microfluidic organ-on-a-chip: Simulation results showed that the concentration in the top layer of the upper chamber reached more than 95% of that in the lower chamber and that the concentration in the bottom layer was about 96.5% of that in the lower chamber after 12 hours of continuous perfusion, reaching an equilibrium state of the concentration difference between the upper and lower chambers. The results of live-dead staining showed that the chip was biocompatible at a flow rate of 0.5 ml/min, and the cell survival rate at 3 and 7 days of perfusion was (99.48±0.12)% and (97.07±1.05)% ( P<0.01). (2) ALP staining results showed that at 3 days, the perfusion-induced group showed the highest number of black ALP-positive cells, followed by the static-induced group, and the least in the static non-induced group. At 7 days, the static-induced group had the highest number of black ALP-positive cells, followed by the perfusion-induced group, and the least in the static non-induced group. PCR results indicated that at 3 days, the expression levels of RUNX2, COL1A1, BMP-2, and OCN were 1.00±0.03, 1.00±0.12, 1.00±0.01, and 1.00±0.02 respectively in the static non-induced group; 1.80±0.04, 4.05±0.37, 9.80±1.94, and 4.38±0.89 respectively in the static-induced group, and 2.45±0.23, 5.48±0.42, 91.50±4.56, and 10.82±4.96 respectively in the perfusion-induced group ( P<0.01). At 7 days, the expression levels of RUNX2 was 1.00±0.01 in the static non-induced group, 1.46±0.46 in the static-induced group, and 1.11±0.08 in the perfusion-induced group ( P>0.05); the expression levels of COL1A1, BMP-2, and OCN were 1.00±0.03, 1.00±0.13, and 1.00±0.09 respectively in the static non-induced group, 9.38±0.25, 14.27±4.35, and 84.01±4.02 respectviely in the static-induced group, and 2.39±0.08, 133.64±8.87, and 86.64±8.36 respectively in the perfusion-induced group ( P<0.01). When comparing the static non-induced, static-induced, and perfusion-induced groups at both 3 and 7 days, the perfusion-induced group demonstrated the strongest osteogenic capability. (3) Characterization of morphology and size and biocompatibility of EVs from three osteoblast subtypes: Under the transmission electron microscope, EVs from EnOB-EVs, StOB-EVs, and MinOB-EVs all exhibited a typical saucer-shaped morphology. The particle sizes of EnOB-EVs, StOB-EVs, and MinOB-EVs were (91.3±14.7)nm, (106.0±16.0)nm, and (68.1±10.7)nm, respectively. Cell proliferation/apoptosis assay results indicated that the optimal administration concentration of EnOB-EVs, StOB-EVs, and MinOB-EVs was all 1.25 μg/mL. (4) Validation of osteogenic effect of the microfluidic organ-on-a-chip on three types of EVs: ALP staining results showed that the non-EVs group had the fewest black ALP-positive cells, followed by the EnOB-EVs group, then the StOB-EVs group, and the MinOB-EVs group had the most. PCR results showed that the expression levels of RUNX2, COL1A1, BMP-2, and OCN were 1.00±0.01, 1.00±0.03, 1.00±0.02, and 1.00±0.02 respectively in the non-EVs group, 1.95±0.11, 6.78±2.04, 7.99±0.57, and 6.93±3.83 repectively in the EnOB-EVs group, 0.79±0.12, 5.68±1.53, 12.59±3.15, and 25.59±0.95 respectively in the StOB-EVs group, and 0.68±0.10, 4.36±0.69, 18.75±3.21, and 34.74±3.98 repectively in the MinOB-EVs group ( P<0.01). Compared with the non-EVs group, EnOB-EVs group, StOB-EVs group, and MinOB-EVs group, the MinOB-EVs group showed the most significant osteogenic effect. Conclusions:The microfluidic organ-on-a-chip constructed using microfluidic technology and cell co-culture techniques is capable of maintaining the normal growth of MC3T3-E1 cells, enhancing their proliferation and osteogenic induction differentiation. EVs released by osteoblasts at different stages possess osteogenic effects and can accelerate the bone sclerosis in the remodeling of subchondral bone during the progression of osteoarthritis.