All the experiments were approved by the Committee on Animal Use and Care of the Dalian Medical University. The rats were euthanized prior to the collection of tissue samples. Articular cartilage chondrocytes were isolated from humeral heads, femoral heads, and femoral condyles of male Sprague-Dawley rats weighing 80–120 g, as reported (18). Briefly, chondrocytes were isolated by digestion with 0.15% type II collagenase for 16 h and resuspended in the Dulbecco's modified Eagle's medium/F-12 (HyClone; GE Healthcare Life Sciences, Logan, UT, USA) containing 10% fetal bovine serum (FBS; HyClone; GE Healthcare Life Sciences), 50 mg/ml ascorbic acid-2-phosphate (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany), and 100 U/ml penicillin-streptomycin. The primary chondrocytes obtained were used in the subsequent experiments.

The primary rat MSCs were isolated from the bilateral femurs and tibias at the same time. After the distal ends of the bone were cut open, the marrow cavities were lavaged with sterile phosphate-buffered saline (PBS). The cells were then resuspended in the high-glucose Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) containing 10% FBS (HyClone; GE Healthcare Life Sciences) and 100 U/ml penicillin-streptomycin. After incubation at 37°C in 5% CO₂ for 48 h, the medium was changed to remove the nonadherent cells. After two passages, the attached MSCs were then used in the following experiments.

The microdevice was coated with 100 mg/ml fibronectin (Sigma-Aldrich; Merck KGaA) for 1 h at room temperature. Then, the chambers were washed with PBS. The suspension of osteoinduced MSCs at a density of 1×10⁵ cells/ml was injected gently from the inlet of top chamber to seed cells on the polycarbonate membrane. After 12 h, a chondroinduced MSCs suspension of 0.5×10⁵ cells/ml, a chondrocyte suspension of 1×10⁵ cells/ml and a mixture of two cells with the corresponding cell density were injected into the three microchambers, respectively. The chip was incubated at 37°C for 12 h to allow the cell attachment. In groups 2 and 4, the inlets of the bottom layer were connected to three peristaltic pumps (Longer Pump BT100-2J; Longer Precision Pump Co., Ltd., Baoding, China). The cells were exposed to 1 mmol/l cytochalasin D (CytoD; Sigma-Aldrich; Merck KGaA) by pipetting medium containing CytoD into the inlet of chip for 1 h to disrupt stress fibers before the application of the flow stimulus.

To generate chondrogenic cells, MSCs were first expanded for 5 days and then subjected to 14 days of predifferentiation in a serum-free chondrogenic medium containing DMEM-LG, 6.25 µg/ml insulin, 50 mg/l ascorbic acid, 1×10⁻⁷mol/l dexamethasone, 10 ng/ml TGF-β3 (PeproTech, Inc., Rocky Hill, NJ, USA) and 1 µM ascorbate-2-phosphate (Wako Chemicals USA, Inc., Richmond, VA, USA), 1% sodium pyruvate (Invitrogen; Thermo Fisher Scientific, Inc.). To obtain Osteogenic commitment MSCs, MSCs were cultured in a complete osteogenic medium containing high glucose DMEM, 10% (v/v) FBS, 50 mg/l ascorbic acid, 10 mM β-glycerophosphate, 1×10⁻⁸ mol/l dexamethasone. Osteoinduced MSCs were exposed to the osteogenic medium for 7 days.

The schematic diagram of the device is as shown in Fig. 1A. The device was composed of two layers of polydimethylsiloxane (PDMS) with several microchambers. The top layer of the chip has one cell culture chamber which was 500 µm in height, 6 mm in width and 11 mm in length and the bottom layer has three microchambers, each of which was 500 µm in height, 2 mm in width and 16 mm in length. Between two pieces of PDMS, there was a polycarbonate membrane (Whatman Inc., Piscataway, NJ, USA), with the pore size of 0.4 µm, which allowed the transition of cytokines and small molecular compounds (19). Each chamber has one inlet and one outlet, respectively. Perfusion system could be linked with the inlet of each chamber to provide the fluid flow.

Microfluidic chip was fabricated using a conventional microfabrication technique. The mask was designed by AutoCAD 2007 (Autodesk, Inc., San Rafael, CA, USA) and printed on transparencies with 4,000 dpi resolution. The master was fabricated by photoresist SU-8 3035 through soft lithography technique. Then the microfluidic chip was fabricated in PDMS by replicating molding the master, followed by the curing process of PDMS at 80°C in an oven for one hour. After that, PDMS layer was peeled off from the master. A polycarbonate membrane and two pieces of PDMS were sealed together after the treatment by oxygen plasma for 1 min.

The overall experimental design is as shown in Fig. 2. In this study, the top layer of the chip was seeded with or without osteoinduced MSCs, and the bottom layer of the chip was separately seeded with chondroinduced MSCs, chondrocytes and a mixture of them in three microchambers. MSCs were first expanded for 14 days in the general medium and then incubated in the osteogenic medium for another 7 days to obtain osteogenic phenotype cells. Chondroinduced MSCs were cultured for 14 days in the chondrogenic medium with TGF-β3 to obtain chondroinduced MSCs. After 14 days' culture, the chondroinduced MSCs were seeded on the bottom of the chip alone or co-cultured with chondrocytes.

To produce a certain shear stress (0.05 dyne/cm²) in the microchambers on the bottom layer of the chip, three perfusion pumps were connected with the inlets of the three microchambers. The cells cultured there were subjected to the shear stress simultaneously. We assumed the flow as a model of laminar and incompressible fluid. A finite volume method (FVM)-based CFD could simulate the 3D flow field in the microchambers with a known pressure at the inlet and outlet of the microchamber. To avoid the computational rigor required to solve Fourier series expansions, we used an approximate version in algebraic form as follows: R=[12ηL/(1–0.63(h/w)] × (1/h3w).

In this formula, R is the hydraulic resistance of the rectangular microchambers, η the dynamic viscosity of the liquid, L the channel length, h and w (always h<w) the channel height and width, respectively. With a given perfusion rate, the Δp could be obtained and used in CFD method. This result can be summarized using the Hagen-Poiseuille equation, as follows: Δp=QRH.

After some calculations, the pressure data obtained were then used as the inlet and outlet pressure conditions to simulate the 3D flow field in each culture chamber using the CFD method. The computational domain was discretized using approximately 52,000 hexahedral meshes and solved using FVM along with the aforementioned inlet/outlet pressure conditions and no-slip boundary conditions at the chamber walls. The density of the perfusion medium was 993.2 kg/m³, and the viscosity was 7.85×10⁻⁴ Pa·s at 37°C.

Cells were seeded at a density of 60 cells/cm² and cultured for 14 days. Then the cells were washed with PBS and fixed with acetone/methanol (1:1). 2% crystal violet solution was used for 20 min staining. Excess stain was removed with tap water.

After MSCs were exposed to the osteogenic medium for 7 days, the osteogenic differentiation was detected by alkaline phosphatase (ALP) activity. After a wash with PBS, the cells were fixed with 500 µl of 4% paraformaldehyde for 2 min. Fixed cells were washed with 1 ml of washing buffer (0.05% Tween-20 in PBS w/o calcium and magnesium) and incubated with BCIP^®/NBT substrate at RT for 10 min. After a wash cells staining for alkaline phosphatase activity were observed under a microscope.

The chondrogenic differentiation was detected by Alcian Blue Assay. After being fixed for 15 min with 4% glutaraldehyde at room temperature, the cells were rinsed with 0.1 N HCl to decrease the pH to 1.0, stained 30 min with 1% Alcian blue solution (Sigma-Aldrich; Merck KGaA), and rinsed twice with 0.1 N HCl to remove nonspecific staining.

The live cells on the bottom layer were stained by plasma membrane dye DiI (Invitrogen; Thermo Fisher Scientific, Inc.) and DiO (Invitrogen; Thermo Fisher Scientific, Inc.) to illustrate the distribution of cells in the three chambers. The cells were exposed to 10 µM DiI and DiO for 1 h and then cells were rinsed with fresh medium before pipetting into the inlet of chip.

The vimentin intermediate filament, collagen I, collagen II and aggrecan were stained by immunofluorescence. Chondroinduced MSCs and chondrocytes were fixed in a 4% paraformaldehyde solution for 15 min, followed by permeabilization with 0.1% Triton-X for 10 min. After being washed with PBS for 3 times, the samples were blocked with goat serum for 30 min at room temperature and incubated with anti-intermediate filament (BIOSYNTHESIS, China), anti-collagen II (Beijing Biosynthesis Biotechnology Co., Ltd., Beijing, China), anti-collagen I (Beijing Biosynthesis Biotechnology Co., Ltd.) and anti-aggrecan (Beijing Biosynthesis Biotechnology Co., Ltd.) overnight. Then they were mixed with FITC-conjugated goat anti-rabbit IgG or TRITC-conjugated goat anti-rabbit secondary antibodies (Zhongshan Golden Bridge, Beijing, China) for 1 h at room temperature, respectively. Cell nuclei were stained with 40, 6-diamidino-2-phenylindole (DAPI) for the following 10 min. For the imaging of cell shape, the intermediate filament was stained by using the same protocol. After staining, the chip was washed with PBS for 2–3 times. The cells were then observed on an Olympus fluorescence microscopy (Olympus IX 71; Olympus Corporation, Tokyo, Japan Japan). The images were analyzed using image Pro Plus 6.0 software. The angle of orientation and the fluorescence area per cell were analyzed by the intermediate filament staining with Image J to evaluate the effect of shear stress.

Living and dead cells were distinguished by different colors. Propidium iodide visualise the nucleus and more than 100 nucleus in each group were picked randomly to check cellular viability. For cell proliferation analysis, cells were fixed and stained by DAPI to count the proliferation rate in each group.

After 7 days exposure to fluid flow stimulus, qPCR was used to analyse the expression level of collagen I, collagen II and aggrecan. After rinsing with cold PBS once, Total RNA was extracted from the cultured cells with TRIzol (Invitrogen; Thermo Fisher Scientific, Inc.), and first-strand complementary cDNA was synthesized using the PrimeScript RT reagent kit (Takara Biotechnology Co., Ltd., Dalian, China). qPCR was performed using Power SYBR-Green PCR Master Mix (Applied Biosystems; Thermo Fisher Scientific, Inc.) in combination with a 7500 Real-Time PCR Detection System. β-actin was used as the internal control gene to normalize the quantities of target gene expressions. Thermocycling conditions were as follows: 95°C for 30 sec 40 cycles of denaturation (95°C, 5 sec), annealing (60°C, 30 sec) and extension (72°C, 30 sec). The primers used in this study were listed in Table I. The relative mRNA level was expressed as fold change relative to group 1 after normalization to the expression of β-actin with 2^−ΔΔCq method (20).

In this study, all the experiments were performed at least in triplicate with different batches of microfluidic chips. All the results were presented as mean ± standard deviation. One-way ANOVA method followed by Tukey's post hoc test was used to compare the statistically significant differences among different groups. P<0.05 was considered to indicate statistical significance.

In this study, peristaltic pumps were connected the inlets of the microchambers to drive the fluid flow through the bottom layer of chip. The fluid flow was Poiseuille flow and we assumed that the shear stress on the cells was equal to the shear stress of the bottom wall of the chamber. When a perfusion flow rate of 7 µl/h was applied to the microchip, a constant fluid flow stimulus could be generated on the cells. From the simulation results by FVM-based CFD (Fig. 1B), the central area of the chambers exhibited a uniform shear stress distribution within the wall (the uniform yellow color means a uniform 0.05 dyne/cm² shear stress in three chambers, 1 dyne/cm²=0.1 pascals), and the average shear stress of the bottom wall was approximately 0.05 dyne/cm², which is equal to the interstitial fluid flow in the cartilage space.

MSCs can rapidly proliferate under suitable conditions with high proportion of FBS. When grown at low density, MSCs are capable of forming colonies. Also, they can differentiate into multiple lineages when induced by special induction medium. When isolated from bone marrow, primary MSCs showed colony-forming capacity (Fig. 3A) and the morphology of MSC colonies was also checked by crystal violet staining (Fig. 3B). After passaging, MSCs showed a fibroblast-like, spindle-shaped morphology and a homogeneous phenotype (data not shown). Then we used osteogenic and chondrogenic medium to induce MSCs. Induced osteogenic commitment cells showed typical flattened morphology after one week's induction (Fig. 3C). Also the osteogenic commitment differentiation was proved by alkaline phosphatase staining (Fig. 3D). Besides, chondrogenic MSCs were induced after 14 days' culture. The morphology was seen in the bright field picture (Fig. 3E) and also proved by Alcian Blue staining (Fig. 3F).

The bottom layer of the chip had three cell culture chambers, in which chondrocytes, chondroinduced MSCs and a mixture of them were seeded inside, respectively. Through the fluorescent probe of DiI (red) and DiO (green), the distribution of cells on the bottom layer could be clearly observed, which indicated the uniform distribution of cells in 3 separated chambers (Fig. 4A). Besides, the viability of cells in different groups was evaluated (Fig. 4B). Under different experiment conditions, all these cells exhibited good state after 7 days' culture. Very few dead cells (positive propidium iodide (PI) staining) were observed, which indicated the feasibility of this microfluidic chip for characterizing the stimulation of osteoinduced MSCs and the shear stress to the cells on the bottom layer. To analyze the effects of osteoinduced MSCs and fluid shear stress on cell proliferation, the proliferation rates of cells in different groups were analyzed by counting the cells (stained by DAPI, data not shown) attached to the bottom of the chambers. As shown in Fig. 4C, the proliferation rates of cells in group 2/3/4 were higher than the proliferation rate in group 1. These findings suggested that both the given shear stress and the osteoinduced MSCs could affect the proliferation rates of chondroinduced MSCs, chondrocytes and a mixture of two cells.

To characterize the shape and cytoskeleton change of cells on the bottom layer in response to the shear stress, the morphology change of chondroinduced MSCs, chondrocytes and a mixture of them was evaluated by intermediate filament staining (Fig. 5A). The results show that chondrocytes showed significant differences on intermediate filament expression under perfusion condition (group 2 and 4), compared with that in group 1 (Fig. 5B). Besides, chondrocytes culture in group 2, 4 showed remarkable morphological changes with spread adhering area per cell on the substrate of the chips (Fig. 5C). Similar results were observed for chondroinduced MSCs and the mixture of cells (data not shown). However, the angle of the orientation induced by the change of cytoskeleton showed no significant difference compared with those in group 1 (Fig. 5D), which indicated the fluid flow may be not strong enough to change the angle of cellular orientation.

After 7 days' cultivation, collagen I, collagen II and aggrecan, the markers for the dedifferentiated and differentiated phenotype of chondrocytes and chondroinduced MSCs, respectively, were characterized by immunofluorescence staining and qPCR to evaluate the phenotypic changes of cells in different groups. The detail data of the relative fluorescence intensities of collagen I, collagen II and aggrecan can be seen in Figs. 6–8. In terms of collagen I, the cells in group 1 were nearly negative (Fig. 6A). The mechanical loading provided by perfusion system in group 2 enhanced the expression of collagen I, while osteoinduced MSCs on the top layer weakened it. However, the cells in group 4, in which the effects of osteoinduced MSCs and the shear stress were combined, showed a weaker staining intensity compared with those in group 2, but stronger than those in group 3, indicating the rescue of the hyaline cartilage phenotype of cells by the stimulation of osteoinduced MSCs. This tendency can also be seen in qPCR result (Fig. 6B). Shear stress improved the mRNA expression of collagen I and this improvement can be reduced by stimulation of osteoinduced MSCs in Group 4 when compared with Group 2.

Then for collagen II and aggrecan, the cells in group 1 were weakly positive (Figs. 7A and 8A). As for cells in group 2 and 3, where there are perfusion systems on the bottom layer and osteoinduced MSCs on the top layer, the expression of collagen II and aggrecan was significantly higher than that in group 1. Cells in group 4, which had perfusion systems and osteoinduced MSCs simultaneously, the fluorescent intensity of collagen II and aggrecan was the strongest compared with those in other groups, indicating that osteoinduced MSCs on the top layer and the shear stress provided by the perfusion system acted synergistically to the cells on the bottom layer. Noncontact co-culture with osteoinduced MSCs neutralized the negative effect of the shear stress and even reversed the dedifferentiation process. Figs. 7B and 8B showed the changes of mRNA levels of collagen II and aggrecan in 4 groups. The collagen II and aggrecan mRNA expression were significantly up-regulated by shear stress and osteoinduced MSCs and the effects were synergistic.