Three-dimensional cartilage tissue regeneration system harnessing goblet-shaped microwells containing biocompatible hydrogel

Differentiation of stem cells into chondrocytes has been studied for the engineering of cartilage tissue. However, stem cells cultured two-dimensionally have limited ability to differentiate into chondrocytes, which led to the development of three-dimensional culture systems. A recently developed microtechnological method uses microwells as a tool to form uniformly sized spheroids. In this study, we fabricated an array (10 × 10) of goblet-shaped microwells based on polydimethylsiloxane for spheroid culture. A central processing unit (CPU) was used to form holes, and metallic beads were used to form hemispherical microwell geometry. The holes were filled with Pluronic F-127 to prevent cells from sinking through the holes and allowing the cells to form spheroids. Viability and chondrogenic differentiation of human adipose-derived stem cells were assessed. The fabrication method using a micro-pin mold and metallic beads is easy and cost-effective. Our three-dimensional spheroid culture system optimizes the efficient differentiation of cells and has various applications, such as drug delivery, cell therapy, and tissue engineering.


Introduction
Osteochondral tissue is a complex tissue that includes both chondrogenic and osteogenic cells. Defective osteochondral tissue causes osteoarthritis, affecting millions of people worldwide [1]. Osteochondral defects refer to degeneration of both the articular cartilage and a piece of underlying subchondral bone [2]. To repair the defect, the complexity of the bone and cartilage must be considered, as the development of osteochondral regeneration has been slowed by technical obstacles related to its complex and hierarchical architecture [3]. In osteochondral tissue, the bone and cartilage exist in different matrices and biophysical environments, which are associated with the transport of molecules. Moreover, the three-dimensional (3D) arrangement of chondrogenic cells in the native osteochondral tissue is challenging to imitate in vitro [4]. To achieve this, scaffolds should mimic the 3D in vivo condition of osteochondral tissue [5][6][7]. The 3D environment can induce the chondrogenic phenotype by promoting cell-matrix interactions with the collagen fiber network and the rest of the intercellular protein matrix [5]. As osteochondral tissue is composed of a 3D structure of collagen matrix and proteoglycans, various 3D culture methods have been applied to the regeneration of osteochondral tissue [8][9][10][11].
Spheroid culture is one of the most specialized methods for the 3D cultivation of cells. It has been explored for various applications in tissue engineering [6,[12][13][14]. Spheroids are spherical cell aggregates that can self-organize through interaction with surrounding cells or extracellular matrix. Spheroid technology has been explored with adipose stem cells or human mesenchymal stem cells (hMSCs) to test chondrogenic differentiation and tissue function, and to reveal cellular functions [15][16][17][18]. In addition, spheroid preparation methods, such as the use of microwells or micropatterns, have been introduced [6,16,19]. These studies have shown that cells cultured in a micro-sized system form dense spheroids that remain relatively constant in size.
Although microwell arrays allow the spontaneous formation of spheroids (figure 1(a)), additional elements are required to mimic specific organisms and tissues. In particular, the supply of various cytokines and nutrients and interaction with other cell types that determine cell differentiation can be provided on advanced platforms [20]. The microfluidic system allows selective cell seeding in microwells so that different cells can be seeded into the desired array in the microwells [21], which enables spheroid co-culture by connecting a microwell to another cell culture chamber [22]. With this system, hepatocyte and hepatic stellate cells were cultured to mimic the environment of liver tissue. However, these microfluidic systems require sophisticated cell seeding controls and additional experiments to fabricate microchannels. Studies have mimicked various biological environments through co-culture in microwells without using microfluidic systems. Hair follicle stem cells, fibroblasts, and keratinocytes were co-cultured to mimic the hair follicle environment and thus produce new hair in microwells [23]. However, spheroids in the microwell arrays introduced in the earlier studies were not exposed to multiple cytokines. Considering that the exposure of cells to various biochemical components in the 3D microenvironment is a biological norm, it is important to develop systems capable of mimicking this exposure.
Here, we have developed a novel goblet-shaped microwell (goblet-well) system that can expose spheroids to two different types of biochemical stimuli/ nutrients ( figure 1(b)) mimicking the physiological environment of osteochondral tissue. The system permits the easy replacement of the two different media at the desired time. Furthermore, the diffusion time can be controlled according to the type of hydrogel. The biomimetic benefits of this system include 3D cell culture (spheroid), multi-stimuli capability, and the ability to expand to accommodate co-culture systems. The goblet-well consists of a microscale hemispherical well in which a spheroid is cultured. A through-hole filled with hydrogel (Pluronic F-127) helps diffusive exchange of the biochemical molecules of the medium in the bottom layer, and supports spheroid formation by reducing cell adhesion to the surface and preventing the cells sinking through the hole [25][26][27]. As a demonstration, we cultured human adipose-derived stem cells (hASCs) for 14 d in this goblet-well array and observed the differentiation of the hASCs into chondrocytes.
Goblet-wells filled with hydrogel are relevant for drug research because they provide an in vivo-like chondrogenic tissue model (figure 1(c)) that features the intercellular interaction and the transfer of multiple cytokines, such as interleukins, fibroblast growth factor, and bone morphogenic proteins [28]. This system can provide a higher-level simulation of in vivo -like conditions, and thus can be useful for drug screening, stem cell differentiation research, and other fundamental physiology related studies.

Fabrication of polydimethylsiloxane (PDMS) goblet-wells
The goblet-well substrate was made of PDMS (Dow Corning Inc., USA). Each well has a hemispherical microstructure with a diameter of 600 μm that is connected to the bottom through a 300 μm diameter hole (figure 2). To fabricate the 3D goblet-shaped microstructure, the micro-pin array and conventional central processing units (CPUs) with a pin diameter of 0.3 mm and height of 1.8 mm obtained from a CPU processor (AMD Athlon, USA) were used as a micropillar array mold. The CPU pin array was placed in a The goblet-well can be exposed to two different media by adding a hole filled with hydrogel to the existing microwell. (c) The goblet-well compared with chondrogenic cells in hyaline cartilage illustration [24]. Abbreviations are: PDMS: polydimethylsiloxane, and ECM: extracellular matrix. 10 mm diameter petri dish (figure 2(a)). PDMS solution mixed with base and curing agent (SYLGARD 184; Dow Corning) in a ratio of 10:1 was added. The volume of liquid PDMS added was adjusted so the level of the solution came to the top-end of the CPU pins without immersing them (figure 2(b)). After curing in an oven at 80°C for 2 h, the PDMS substrate was removed from the CPU pins to produce a flat rectangular PDMS substrate with an array of microholes. The bottom of the obtained substrate was sealed with plastic tape, and 600 μm diameter metal beads were aligned on the upper surface of each hole to prepare the goblet shape of the well (figure 2(c)). To immobilize the bead on the top of a hole, thermal expansion/contraction was used as follows. When the system was heated at a high temperature (approximately 100°C), the air in the holes expanded. After 10 min, the temperature was allowed to decrease to room temperature. The air inside each hole contracted, which secured the beads in place by the negative pressure created in the hole. This thermal expansion/contraction of trapped air in the hole follows Charles's law (V T =V 0 (1+T/273)), where V T is the volume of gas at temperature T and V 0 is the volume at 0°C. After that, additional PDMS solution was poured to half the height of the beads and cured at room temperature for one day (figure 2(d)). The PDMS thickness has been controlled in other studies using spin-coating [29,30]. However, this method cannot be applied in this system since the beads are fixed in place. Instead, we controlled the thickness of the poured PDMS by tilting the substrate and allowing excess PDMS to flow off. The beads were then removed and collected using a neodymium magnet (figure 2(e)) to generate the goblet-wells (figure 2(f)). By removing the plastic tape, a goblet-shaped microwell array was obtained. The uniform goblet-wells were characterized using SEM (S3400N, Hitachi, Japan). For cell culture, 10×10 goblet-well arrays were cut and attached to the insert well (figure 2(g)). A six-well plate was typically used for co-culture (figure 2(h)).

Filling goblet-well holes with Pluronic F-127
To ensure stable formation of the spheroids, the holes connected to the goblet-wells were filled with a solution of Pluronic F-127 (P2443; Sigma-Aldrich, USA). To prepare the solution, Pluronic F-127 powder was added slowly to cold Dulbecco's phosphatebuffered saline (DPBS) at 4°C (using an ice bath) and stirred at 900 rpm for 1 d. The Pluronic gel F-127 concentration of 30% w/v was poured into the 10×10 goblet-wells and allowed to drain through the hole, still at 4°C. When the Pluronic F-127 solution had completely drained, it was dried in a dry oven at 40°C for 1 d. Wells were confirmed to be filled with gel by optical microscopy using an ECLIPSE TS100 microscope (Nikon, Japan) at ×4 and ×10 magnification. Moreover, the presence of gel was validated by safranin-O dye. Briefly, safranin-O solution, 1%w/v was added to the goblet-wells. After drying the gel, both side of the hole was exposed with DPBS. Then images were captured using a microscope (Olympus 110AL2X-2, Japan) mounted with a DSLR camera (Canon EOS 600D, Japan). For better visualization, the goblet-well was observed in a vertically cut crosssection view.

Diffusion through the holes
The spheroids cultured in the goblet-wells are exposed to two different media supplied from the top and bottom layers. The bottom layer medium is transported by diffusion through the Pluronic gel that fills the hole (figure 4). It was necessary to quantitatively confirm the diffusion rate of the bottom layer medium The solidified PDMS is separated from the CPU pins, and attached to a piece of plastic tape to cover the bottom of the hole. The 600 μm metal beads are placed on the tape. (d) After heating the PDMS with the metal beads at 100°C for 10 min and bringing it to room temperature, the PDMS solution is poured to fill the middle part of the bead and is solidified at room temperature for one day. (e) The metal beads are removed from the solidified PDMS using a magnet. (f) The completed goblet-well array. The size of the hemispherical well is determined by the bead size; the diameter of the hole is 300 μm and the total height is 2 mm. (g) The goblet-well array is adhered to the insert well using PDMS glue (PDMS solution). (h) Using the prepared insert wells and six-well plates, spheroid cells are cultured using two different culture media.
(biochemical) to the spheroids. To visualize the diffusion inside the hole, Trypan blue was used to fill the bottom layer as a dye reagent. The movement of the dye in the hole was observed in real-time and photographed with a microscope. The goblet-well was cut vertically from top to bottom, and the diffusion that had occurred on the central axis of the hole was observed. Experiments were carried out in the presence of 30% Pluronic F-127 in the hole.

Diffusion simulation model
Computational geometry of a goblet-well was constructed and a 3D diffusion simulation was performed. The dimensions used in the analysis were the same as in the fabricated goblet-wells. The microwell and hole diameter were 600 μm and 300 μm, respectively, and the distance from the bottom to the goblet-well was 2 mm. A 100% mass fraction of the bottom medium was set as the source term for the bottom surface of the hole. A 1 mm square columnar domain above the microwell was set to serve as the culture medium for which the diffusion of cytokines from the bottom of the hole to the outside was calculated. The software used in this simulation was ANSYS Fluent 15 (Ansys Inc., USA). The goblet-well was assumed to be filled with water (a homogeneous, incompressible Newtonian fluid) at 36°C and the bottom of the hole was fixed with another water medium. Two different types of water media mixed through diffusion, where the diffusion coefficient (1×10 −10 m 2 s −1 ) [31] of a cytokine was applied. There were 300 000 grids, and the time step size was 1 s.

hASC spheroid culture
The hASCs were isolated from adipose tissue obtained during surgeries conducted at the Korea Cancer Center Hospital. The hASCs were expanded up to passage 5 using low-glucose Dulbecco's modified Eagle's medium (DMEM-LG; HyClone, USA) containing 1% antibiotic/antimycotic solution (HyClone) and 10% fetal bovine serum (FBS; HyClone). The medium was replenished every 2 d. The cultured cells were harvested using 0.25% trypsin (1×) solution (HyClone). The harvested cells were seeded at a concentration of 0.5×10 6 cells ml −1 on the PDMS substrate (5000 cells in each goblet-well). Cells that did not enter the goblet-wells were removed by carefully aspirating the medium. Then the substrate was refilled with the fresh medium. The hASCs from passage 5 were incubated in a humidified incubator at 37°C and 5% CO 2 . The formation of spheroids was confirmed by optical microscopy using the aforementioned ECLIPSE TS100 microscope. The diameters of the spheroids in the micrographs were measured using the ImageJ software (v. 1.48, National Institutes of Health, USA).

Cell viability analysis
The viability of hASC spheroids cultured in gobletwells was analyzed using the LIVE/DEAD Viability/ Cytotoxicity kit for mammalian cells (Molecular Probes, USA) at days 1, 3, 7, and 14. Spheroids were rinsed using DPBS for 20 min to remove the medium and the kit solution (4 mM calcein AM and 2 mM ethidium homodimer in DPBS) was added to each goblet-well and incubated at 37°C and 5% CO 2 for 30 min. The stained cells were observed by fluorescence microscopy using a model OX.2053-PLPH microscope (Euromax, Netherlands). Moreover, anoikis propelled cell death was confirmed by anoikis assay detection kit (Abcam, UK) following fluorometric detection method in kit manual. Briefly, 1 μl of ethidium homodimer (500×) was added to 96-well plate having 4 d cultured spheroid samples. The plate was then incubate at 37°C for 30 min and the signals were measured using microplate reader (Biotek Instruments, USA) at 525 nm excitation wavelength and 590 nm emission wavelength.

Cell spheroid proliferation tests
The proliferation of hASC spheroids cultured in the goblet-wells was measured using the CCK-8 assay (Dojindo Molecular Technologies, USA) compared to conventional microwells, after culture for 1, 3, and 7 d. CCK-8 solution was added to the medium and adjusted to 10% of the total medium volume in each goblet-well. After incubation for 3 h, 100 μl of the medium was collected from each well to check the absorbance at 450 nm using a Synergy H1 multi-mode microplate reader (Biotek Instruments, USA). The obtained absorbances were used to calculate cell quantity by comparison with the standard curve. The quantities obtained with time allowed determination of cell proliferation.

Chondrogenic differentiation and gene expression analysis
The hASCs were cultured in the goblet-well array with general medium (GM) including DMEM-LG containing 1% antibiotic/antimycotic solution (HyClone) and 10% FBS for 3 d to allow spheroid formation. Then, spheroids were supplied with chondrogenic differentiation medium (CDM) that contained 50 μg ml −1 ascorbic acid (Sigma-Aldrich), 10 mM Insulin-Transferrin-Selenium-A (100×; Gibco, USA), 0.01 μM dexamethasone (Sigma-Aldrich), and 10 μg ml −1 transforming growth factor-beta 1 (TGF-β1; ProSpec, Israel). The hASCs were cultured for up to 14 d. For gene expression analysis, hASC spheroids in each goblet cell were rinsed using DPBS and the total mRNA was isolated using Trizol reagent (Thermo Fisher Scientific, USA). The isolated total mRNA concentration was quantified using a microplate reader by determining the ratio of the absorbances at 260 nm and 280 nm. An equal amount of cDNA was synthesized in 10 μl of RT-PCR mixture containing RNA template, random hexamer, 5× PrimeScript TM Buffer, PrimeScript TM RT Enzyme Mix I, and oligo dT primer using the protocols of the PrimeScriptTM RT Perfect Real-Time Reagent Kit (TaKaRa Bio, Japan). Subsequently, RT-qPCR was conducted using the SYBR ® Green PCR Master Mix Kit (Applied Biosystems, USA), 10 ng of cDNA template, and 10 μM target gene primers. The cartilage related genes, including COL2, AGG, and SOX9, were analyzed by normalizing to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Sequences of the primers used are listed in table 1. Primer sequences used for RT-qPCR are summarized in table 1. The RT-qPCR conditions were 95°C for 10 min and 40 cycles of 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s, with a final step of 60°C for 1 min and 95°C for 15 s. The results were analyzed using the Step One Plus™ Real-Time PCR system (Thermo Fisher Scientific). The melting curves were obtained to confirm the specificity of the amplified products, including the GAPDH housekeeping gene.

Statistical analyses
For the determination of statistical significance, the results were analyzed using a two-way analysis of variance (ANOVA) followed by Tukey's multiple comparison tests with every other sample group. A P-value<0.05 was determined to be statistically significant. For quantitative data, the values are expressed as mean standard deviation.

Results and discussion
3.1. Fabrication of PDMS goblet-wells filled with Pluronic F-127 We successfully fabricated a 10×10 goblet-well substrate (well diameter 600 μm, hole diameter 300 μm) on PDMS. Using a uniform size of micro-pin and microbeads, goblet-wells were obtained (figure 3) and the uniform shapes were confirmed with SEM ( figure S5). The metal beads were placed on the prepatterned holes of the PDMS substrate to produce the goblet shape of each microwell. Holes of the gobletwells were filled with Pluronic F-127 to stabilize the formation of spheroids (figures 3(d)-(f)). The presence of the Pluronic F-127 gel was confirmed by the safranin-O staining method. The red color dye was trapped inside the Pluronic F-127, as shown in figure S6. The filling was performed at a low temperature due to the sol-gel properties of Pluronic F-127. At low temperature, the gel-like Pluronic F-127 can reversibly change to a liquid state and thus flow throughout the hole in each well. When the temperature is subsequently increased, micelles of Pluronic F-127 form and the gel-like preparation is sequestered in the holes [32][33][34]. Temperature-responsive Pluronic hydrogels have been studied as mediators for the delivery of proteins or injectable cell therapeutic agents in the biopharmaceutical field. Pluronic hydrogels have characteristics that include minimal cell adhesion, minimal mechanical properties, and rapid erosion, which can be applied to tissue engineering or drug delivery [32,35]. The attribute of minimal cell adhesion has been studied recently. Cell adhesion and protein adsorption are reportedly reduced on surfaces treated with Pluronic hydrogel [25,36,37]. This function supports spheroid formation at the initial stage, after which, the hydrogel degrades over time [38]. The goblet-wells also provide a better environment for biomolecule delivery.

Diffusion test
To visualize diffusion inside the holes connected to the goblet-well, we performed a computational simulation and measured the cross-section of the goblet-well interior by optical microscopy. In the experiment, the diffusion of Trypan blue at the bottom was observed in real-time, and simulation data were obtained under the same conditions ( figure 4). The simulation data were set to display an area of 5% to 100% based on the mass ratio of the water (or cytokine) spread on the bottom surface ( figure 4(a)), and the legend was set as blue so that it could be compared with the water-filled hole experimental data. In each experiment, the area of the hole was divided into 10 equal zones and the time at which Trypan blue reached each zone from the bottom of the hole was determined. When the inside of the hole was filled with water ( figure 4(b)), it took approximately 2 h for the Trypan blue to diffuse from the bottom to the top layer. The results of the simulation were similar to those of actual water diffusion (figure 4(d)) with an error rate <10%. When the inside of the hole was filled with Pluronic F-127 ( figure 4(c)), Trypan blue diffusion from the bottom to the top layer was five times longer, being completed within 10 h. Even though the diffusion rate of the hydrogel was slower than the diffusion rate of water, it was still applicable to a variety of spheroid co-culture and mass transfer systems, since diffusion was completed within 10 h. The diffusion time through the hole varied from 2 to up to 10 h, and could be adjusted according to the type and characteristics of the hydrogel. Since Pluronic F-127 can be degraded in nature [38], the diffusion rate after one point would become normal, similar to water diffusion, which would accelerate growth factor delivery when spheroids need more than initial state due to differentiation process.

Cell spheroid formation
We confirmed that the hASC spheroids formed stably depending on the Pluronic F-127 in goblet-wells. The  hASCs were cultured in sterilized goblet-wells and the formation of spheroids was confirmed over time.
The hASC spheroids formed stably and did not fall through the holes when the goblet-wells contained Pluronic F-127 gel ( figure 5). The results at day 0 indicated that more cells were maintained in gobletwells with Pluronic F-127 than in wells without gel. The comparison with goblet-wells without gel revealed that hASCs had not adhered to PDMS in the presence of Pluronic F-127 on day 1 [25][26][27]. Optical microscopy showed that the stable morphology of spheroids was maintained for a week, which was confirmed by Live/Dead imaging and staining with 4′, 6-diamidino-2-phenylindole and Rhodamine (see figures S1 and S2 in the supplementary material which is available online at stacks.iop.org/BF/12/015019/mmedia). Moreover, the proliferation rate of the spheroids was confirmed over 7 d culture, and we found the cells had proliferated (see figure S3 in the supplementary material).

Viability of hASC spheroids
The viability of hASC spheroids cultured in gobletwells was determined. Most of the spheroids were alive (green) and only a few cells had died (red color in figure 6(a)). The validation of anoikis, an apoptosis at initial stage, also showed lower Ethidium homodimer signals compared to conventional control (figure S3). These results indicated that spheroids cultured in goblet-wells did not develop the necrosis that can occur in the central part of spheroids [39]. The necrotic core forms because of the limitation of the nutrient supply to the core and is commonly reported in spheroids with a diameter >500 μm [40,41]. In our study, the average size of spheroids was <300 μm because the height of the well was 300 μm and, therefore, necrotic core formation was restricted. The average size of spheroids in goblet-wells containing 30% Pluronic F-127 was 295±68 μm on day 1, 260±35 μm on day 3, 251±21 μm on day 7, and 257±28 μm on day 14 ( figure 6(b)). Spheroid size has been related to the aggregation and proliferation of the cells [16,42]. The size of the spheroids in the presence of 30% Pluronic F-127 initially decreased (figure 6(b)) because cell aggregation was more pronounced than proliferation. After culturing for 3 d, spheroids entered the second phase, which featured excessive proliferation, followed by the third phase in  which the volume of the spheroid biomass stabilized after 10 d due to quiescent cells [42]. Corresponding with the 4′,6-diamidino-2-phenylindole/Rhodaminestained spheroids (figure S2), actin (which is stained red by Rhodamine) seemed to be more centrally localized in the spheroids, which may indicate quiescent cells. For goblet-wells without gel, cells were flush away through the holes after day 1, therefore spheroid cannot be formed properly, corresponding with Live/ Dead assay results obtained after day 7 when only separated cells remained. Due to the pronounced hydrophobicity of Pluronic F-127, cell assembly was more extensive and spheroids were more stable than the cells in goblet-wells without gel [16,43,44] 3.5. Chondrogenic differentiation of hASCs in goblet-wells Because they can differentiate into various mesenchymal-type cells with therapeutic potential, hASCs can be used in tissue engineering and immune-modulatory applications [7,[45][46][47]. Based on this background, hASCs were cultured in 2D (6-well plates) and 3D (goblet-wells) conditions. To confirm the quantitative analysis of chondrogenic differentiation, RT-qPCR was performed to evaluate gene expression levels of AGG, COL2, and SOX-9 ( figure S4). The expression of the three genes in 3D culture was higher than in 2D culture at days 7 and 14. It has been shown that spheroid cultures have the ability to enhance chondrogenic differentiation compared to 2D cell cultures [48][49][50]. To show the novelty of goblet-wells, the conventional microwells were used as a control to compare the chondrogenic differentiation ( figure 7). The AGG gene, which is related to cell aggregation, was up-regulated in goblet-wells because the system provides more area for spheroids, allowing them to absorb more of the surrounding nutrient for differentiation. The all-around contact system of the goblet-well provides an in vivo-like environment, supporting cell interaction and aggregation [51]. The increased SOX9 gene expression in spheroids reflected the activation of HF-1α induced by the cell response to low oxygen diffusion in spheroids. This activation stimulates the production of SOX9 [15]. The SOX9 gene has a regulatory role in chondrogenesis, and it is the regulator of collagen type II (COL2), which is a key regulatory protein of cartilage [52]. Therefore, the up-regulation of COL2 is due to the high expression of SOX9. These results indicated that the chondrogenic differentiation levels of hASCs in goblet-wells were enhanced during 3D growth compared to those in conventional microwells. Several factors could affect the activation of differentiation, including the morphology of the in vivo cellular environment and the interaction between cells within a spherical shape [19,[53][54][55].
3.6. Chondrogenic differentiation of hASCs in goblet-wells with different treatments using chondrogenic medium Biological signals in the form of growth factors are important in inducing cell differentiation toward the chondrogenic lineage [56]. We used different treatments with CDM for spheroids by diffusion through the bottom hole of each goblet-well. RT-qPCR examination of the expression of the AGG, COL2, and SOX9 target genes revealed increased expression of all three genes in all four conditions, with high expression at day 14 ( figure 8). Expression of AGG and COL2 was highest in the CDM/CDM condition. In the GM/CDM condition, the gene expression was lower than in CDM/ CDM because the growth factors in CDM at the bottom slowly diffuse to spheroids, and longer time is needed to reach the sufficient concentration for cell differentiation. By contrast, the CDM/GM condition displayed lower gene expression compared to CDM/CDM because of the initial high concentration of differentiation media, although it was diluted gradually by GM. Of note, when the concentration of ascorbic acid increases in media, the secretion of glycosaminoglycan (GAG) can decrease, implying a lower differentiation rate [57]. In addition, in the CDM/GM condition, the expression of every gene was higher compared to that in GM/CDM because the spheroids in the CDM/GM condition were exposed directly to the high concentration of CDM and thus had access to more supplements required for differentiation than in the GM/CDM condition. These results indicated that the growth rate of spheroids could be controlled by the diffusion rate of growth factors through the hole in a goblet-well. The presence of the growth factors can upregulate chondrogenic marker expression during 3D culture of cells [58]. This experiment using two different kinds of media, GM and CDM, demonstrated different growth rates, indicating the availability of a two-media phase for the co-culture of osteogenic and chondrogenic cells, which can be used as an in vivo-like system, mimicking osteochondral tissue. These gene expression results imply that our goblet-well system has the potential for in vivo-like spheroid cell co-culture with osteogenic cells by the biomolecule diffusion pathway. This system may be the basis of further advanced research on osteochondral tissue regeneration.

Conclusion
Using goblet-wells and Pluronic F-127, we devised a system capable of exposing spheroids to two different media/cytokines. Prolonged culture of the spheroids was successful. The spheroid culture system is a useful 3D cell culture system that results in similar functional and morphological features with cells in vivo. There are various ways to cultivate spheroids. The present system using micro-pin mold and beads is easy and costeffective. In addition, goblet-wells having a uniform size prepared through micro-pin mold and beads were successfully used to cultivate hASC spheroids. Pluronic F-127 was used to reduce cell adhesion to the PDMS surface and to form hASC spheroids by filling the holes of the goblet-wells. Spheroids with a diameter <300 μm were cultured with very little cell death. The level of chondrogenic differentiation in the goblet-wells was improved compared to the 2D culture. Moreover, we checked the effect of diffusion and found that gene expression was controllable by varying the treatment pattern. This diffusion ability in the 3D cell co-culture platform will be useful for future investigations of 3D spheroid cell co-culture systems of chondrogenic cells (in which differentiation is enhanced; figure 7) and osteogenic cells for osteochondral tissue engineering. The results demonstrate the production of spheroids that can effectively regenerate cartilage tissue and have various applications, such as in drug or growth factor delivery systems and cell therapy.