Dynamic formation of cellular aggregates of chondrocytes and mesenchymal stem cells in spinner flask

Abstract Objectives Cellular aggregates are readily applicable in cell‐based therapy. The effects of agitation and inoculation density on the aggregation of cells in spinner flask and the molecular mechanism of aggregation were investigated. Materials and methods The aggregation kinetics of cells in spinner flask was evaluated with bovine articular chondrocytes (bACs), rabbit bone marrow‐derived mesenchymal stem cells (rMSCs) and their mixture. The morphology of cellular aggregates was studied with scanning electron microscopy and gene expression of cell adhesion‐related molecules was analysed. Results It was shown that suspension culture in spinner flask induced the aggregation of bACs and rMSCs. Both cells exhibited increased aggregation rate and aggregate size with decreasing agitation rate and increasing cell inoculation density. Additionally, aggregate size increased with extended culture time. By analysing gene expression of integrin β1 and cadherin, it was indicated that these molecules were potentially involved in the aggregation process of bACs and rMSCs, respectively. Aggregates composed of both bACs and rMSCs were also prepared, showing rMSCs in the core and bACs in the periphery. Conclusions Cellular aggregates were prepared in dynamic suspension culture using spinner flask, the key parameters to the aggregation process were identified, and the molecular mechanism of aggregation was revealed. This would lay a solid foundation for the large‐scale production of cellular aggregates for cell‐based therapy, such as cartilage regeneration.

multicellular aggregates. 4 For instance, autologous nasal chondrocytes suspended in alginate hydrogel as injectable constructs for rabbit articular cartilage repair obtained superior and more hyaline-like repaired tissue, demonstrating similar mechanical properties to native cartilage 6 ; dedifferentiated chondrocytes regained a functional chondrocyte phenotype when embedded in a 3D porous scaffold made of alginate and gelatin 7 ; when transplantation of aggregates of synovial MSCs formed by hanging drop regenerated meniscus more effectively than intra-articular injection of suspension of synovial MSCs with the same number of cells in a rat massive meniscal defect. 8 Among the above approaches, cellular aggregates not only provide an in vitro culture condition mimicking in vivo microenvironment, which allow cell-cell contact and cell-ECM interactions, but also are readily applicable in cell-based therapy. 9 The exquisite communication network of mechanical and biochemical signals within 3D cellular aggregates is critical for improving cellular viability, phenotype and function that are often lost in 2D culture. 10 As early as the 1960s, it was shown that monodispersed chondrocytes spontaneously formed aggregates in suspension culture and secreted abundant collagen II and glycosaminoglycans (GAG), typical matrix components in hyaline cartilage. 11 Recently, Bhumiratana et al 12 proposed that small spheroids of MSCs could be induced to fuse and form mechanically functional human cartilage in vitro by mimicking some aspects of the pivotal stage during chondrogenesis, which has potential use for repairing cartilage defects.
Approaches to fabricate cellular aggregates including liquid overlap technique, handing drop and microfluidics as well as bioreactors have been reported. 13 Among these, the advantages of bioreactors lie in conveniently generating a large quantity of cellular aggregates, providing a controlled dynamic fluid environment that will increase the mass transfer of nutrients, oxygen and metabolites and producing the mechanical stimulation such as shear stress on cells, which can potentially influence cell phenotype. 14  In the present study, culture of bovine articular chondrocytes (bACs), rabbit mesenchymal stem cells (rMSCs) and mixed cell population of bACs and rMSCs in spider flask was established and the effect of agitation rate and cell inoculation density on the kinetics of aggregation as well as the associated molecular mechanism of aggregation were explored. The key parameters to the aggregation process were identified, and the molecular mechanism of aggregation was revealed. This would lay a solid foundation for large-scale production of cellular aggregates for cell-based therapy, such as cartilage regeneration.

| Cell isolation and monolayer culture
Under aseptic conditions, full-thickness articular cartilage tissue was obtained from bovine tibiae and femora. The tissue was minced into small slices, washed three times in phosphate-buffered saline

| Cell labelling
To distinguish the different cells in the mixture of bACs and rMSCs, the harvested P2 bACs and P5 rMSCs were labelled with CFSE (Sigma) and PKH26 (Dojindo, Kumamoto, Japan) according to the manufacturer's instructions, respectively. CFSE was capable of penetrating into cell membrane, covalently binding to intracellular proteins, and releasing green fluorescence after hydrolyzation.
PKH26 was stably incorporated and retained in plasma membrane, red fluorescence.

| Cell culture in spinner flask
Spinner flasks (Corning) were silicon-treated to prevent cells from adhering to the flask walls. Cells were seeded into spinner flasks while keeping total volume at 100 mL and cultured for 5 days. The medium was kept during the process of any operations on cellular aggregates, and the sidearm caps of spinner flasks were loosened to allow for gas exchange.
P2 bACs of 4 × 10 5 cells/mL were seeded into spinner flasks at three different agitation rates (40, 50 and 60 rpm) to investigate the effect of agitation rate on the formation of cellular aggregates. And cells were also seeded into spinner flasks in three different seeding densities (2, 4 and 8 × 10 5 cells/mL) while keeping the agitation rate of 50 rpm.
P5 rMSCs were seeded at the same densities as bACs at an agitation rate of 45 rpm to investigate the effect of cell seeding density on cell aggregates, while the three agitations rates were separately 40, 45 and 50 rpm to evaluate the effect of agitation rate.
Additionally, labelled P2 bACs and P5 rMSCs were mixed at a ratio of 1:1 and seeded into spinner flask at an agitation rate of 40 rpm with a total cell density of 2 × 10 5 cells/mL.  (1). It was assumed that total cell concentration (C 0 ), which equalled to the cell seeding density, remained constant, which was verified by determining DNA content and MTT assay between 0 and 1 day (data not shown). Based on the loss of single cells from the suspension with time, the aggregation kinetics was calculated using the following formula (1), which is based on the mathematical model proposed by Morris. 16 where k [h −1 ] is the kinetic constant of cell incorporation into cellular aggregates.

| Cell counting and image analyses
In addition, 200 μL aliquots of cell suspension in triplicate were obtained after 1, 3 and 5 days of culture and imaged under inverted microscope (Eclipse Ti, Nikon, Japan). Then, the size distribution and roundness distribution of these images were analysed by Image J software (n = 500-1000 per time point).

| Scanning electron microscopy
For scanning electron microscopy (SEM), samples taken after 1, 3 and 5 days of culture in spinner flask were washed twice with PBS, fixed in 2.5% glutaraldehyde at 4°C overnight, dehydrated in ascending grades of ethanol and air dried. The samples were fixed, sputtercoated with gold and examined under SEM (S-3400; Hi-tachi, Tokyo, Japan).

| Confocal laser scanning microscopy
To evaluate the location and distribution of cells in the aggregate composed of bACs and rMSCs, the aggregates after 1, 3 and 5 days of culture in spinner flask were observed using confocal laser scanning microscopy (CLSM; TI-LU4SU, Nikon, Japan).

| Quantitative real-time reverse transcriptionpolymerase chain reaction (qRT-PCR)
To examine gene expression of cells, total RNA was extracted by Trizol (Invitrogen) reagent. The concentration of recovered RNA was determined by Nanodrop (ND-2000; Thermo, Waltham, MA, USA). Following the manufacturer's protocol, cDNAs were synthesized from 1 µg of total RNA. Real-time PCR was performed using SYBR Green qPCR kit (Takara, Tokyo, Japan) on a thermal cycler (Mx3000P; Biorad, Hercules, CA, USA). The primer sequences for bovine and rabbit gene used were listed in Tables S1 and S2, respectively. The expression levels of mRNA were calculated and normalized to GAPDH.

| Statistical analysis
All quantitative results were represented as mean ± standard deviation. Student's t test was used to analyse the significant difference between two groups. Significance was defined as P < 0.05.

| Formation of bACs aggregates in spinner flask
As shown in Figure 1A,C, bACs incorporated in aggregates were calculated by using the Equation (1), and the result showed that in all experimental conditions the percentage of bACs in aggregates increased with culture time. And nearly all the cells were encased in the aggregates at 24 hours. It was also found that with increased agitation rate, the rate of cellular aggregates decreased ( Figure 1A).
Initially, a higher percentage of bACs in aggregates in 4 × 10 5 cells/ mL was observed at 50 rpm than those in 2 × 10 5 and 8 × 10 5 cells/ mL ( Figure 1C). This trend began to change from 6 hours, and from this time point till the end of culture, more chondrocytes aggregated at 8 × 10 5 cells/mL. Based on the disappearance of single cells from suspension with time, the kinetics of cell aggregation was calculated by Equation (3). As seen in Figure 1B, the concentration of single bACs in suspension decreased exponentially with time, and the kinetic model fitted the data at 40, 50 and 60 rpm (R 2 > 0.95). bACs exhibited an increased k value at a lower agitation rate, with about 20% and 12% higher at 40 rpm than at 50 and 60 rpm, respectively.
There were differences in the kinetics of bACs aggregation with respect to cell inoculation density. Higher k values were observed with an increase in cell inoculation density, with about 47% higher at 8 × 10 5 cells/mL and 34% high at 4 × 10 5 cells/mL than that at 2 × 10 5 cells/mL for bACs. The results showed that agitation rate and cell inoculation density modulated cell aggregation and a lower agitation and higher cell inoculation density promoted the aggregation of bACs.
To monitor bACs aggregation, SEM was used to observe the surface characteristics of cellular aggregates after 1, 3 and 5 days of culture in spinner flask ( Figure 1E). It was found that bACs formed compact and spherical aggregates on day 1, while the surfaces of aggregates were irregular. The size of cellular aggregates increased gradually with culture time, and at the same time, the surfaces became regular and smooth ( Figure S1). In addition, a higher agitation rate and lower cell inoculation density decreased the size of cellular aggregates.
Based on image analysis of aggregate size, the size of bACs aggregates mainly ranged between 20 and 40 μm after 1 day, and the size distribution gradually became broad on day 3 and 5 ( Figure 1F).
From day 1 to day 5, while smaller aggregates turned less, larger ones strikingly increased in all culture conditions, indicating that the dynamic condition might promote the fusion of cellular aggregates.
Moreover, the average diameter of bACs aggregates significantly increased with extended culture time, decreased agitation rate and increased cell inoculation density ( Figure 1G,H).

| Formation of rMSCs aggregates in spinner flask
After 24 hours culture, in all experimental groups, about 80% of rMSCs were incorporated in cellular aggregates (Figure 2A,C).
The rate of aggregates fluctuated before 6 hours, but consistently increased with extended culture time and decreased agitation rate, and increased cell seeding density afterwards. As seen in Figure 2B k was approximately 30% and 37% higher at 40 rpm than at 45 and 50 rpm. It was also found that rMSCs aggregation at different cell densities followed the simple first-order kinetic model, with R 2 > 0.95 at 2 × 10 5 and 4 × 10 5 cells/mL and R 2 > 0.85 at 8 × 10 5 cells/mL ( Figure 2D). Taken together, this indicated that a lower agitation rate and higher cell inoculation density supported faster aggregation.
To monitor the aggregation of rMSCs, SEM was used to assess the surface characteristics of the aggregates after 1, 3 and 5 days of culture in spinner flask. As shown in Figure 2E, rMSCs formed spherical cell aggregates, which were different from the fusiform in 2D monolayer culture. An irregular topology of the aggregates was also observed during the culture process. Images at higher magnifications revealed that cellular aggregates showed no distinct gap between neighbouring cells ( Figure S2). The size of rMSCs aggregates was decreased with a higher agitation rate and lower cell seeding

| Formation of mixed co-aggregates of bACs and rMSCs in spinner flask
Similar to respective culture of bACs and rMSCs, coculture of bACs and rMSCs in spinner flask also formed aggregates in 1 day. The amount of cells in aggregates increased with culture time, and over 95% were entrapped after 24 hours ( Figure 4A). Similarly, the firstorder kinetic model fitted the aggregation of cocultured cells with time (R 2 > 0.95; Figure 4B).
By using SEM, an irregular topology of aggregates was identified on day 1 and to a greater extent on day 3 and 5 ( Figure 4C). Images of higher magnification indicated that the surfaces of aggregates became progressively rough. The size of aggregates also increased with culture time. To investigate cell distribution in the aggregates, confocal microscopy was adopted to display cell morphology. As seen in Figure 4D, both bACs and rMSCs were present in all mixed coaggregates. Cellular aggregates were relatively small, irregular and loose after 1 day, and then became larger and more compact with extended culture time. Images of high magnification clearly showed that bACs overlaid around rMSCs in the aggregates.

| D ISCUSS I ON
Cellular aggregates have been used in cell-based therapy for cartilage tissue regeneration. 17,18 It has been documented that chondrocyte aggregates promote cell differentiation, secretion of ECM molecules and seeding onto scaffolds than monodispersed cells. 19 MSCs can be induced to undergo chondrogenic differentiation in vitro via cellular aggregation which essentially mimics MSCs condensation, a vital stage in the skeletal development. 20 Therefore, it is worth exploring cellular aggregates for cartilage regeneration.
Culture in spinner flask has been reported to induce the aggregation of both bACs and rMSCs readily. 4,15 In dynamic 3D environment, the agitation rate and initial cell density are implicated to be very critical in modulating cell-cell and cell-aggregate contacts and then the formation of cellular aggregates. 21 In the present study, the aggregation of bACs, rMSCs and their mixture in different culture conditions was characterized. Particularly, the use of cells from different species was mainly for conveniently distinguishing between two cell types in subsequent coculture studies, specifically for genetic analysis, albeit it is not the main focus of the present study. It was demonstrated that both cell seeding density and agitation rate could affect the aggregation kinetics and the aggregate size. In the present study, the lowest agitation rate was set was significantly lower than that at 50 rpm. 22 Others also found that the aggregation rate increased with cell density. 16 The aggregation of  Figure S3A). In fact, a previous study also implicated that the formation of chondrocytes aggregates in suspension culture was mainly through binding of collagen II with integrin β 1 . 30 Coculture of MSCs and chondrocytes has been recognized a very promising strategy for cartilage regeneration. 31 In the present study, aggregates formed with a mixture of bACs and rMSCs were also successfully prepared, and the first-order kinetic model was able to well describe the aggregation process of the mixture. Moreover, it is very interesting to find under confocal microscopy that the mixture of bACs and rMSCs (the number ratio at 1:1) generated core-shell structured cellular aggregates, wherein bACs covered rMSCs tightly.
It had been implicated that cells releasing more cadherin could help to migrate to the interior of the aggregates. 32 Steinberg guessed that cell-to-cell adhesion and cell sorting behaviour in embryonic tissues are the need of maximized adhesion and minimized energy. 33 The adhesion molecules regulate cell surface tension which cells released more cadherin forming the core of the aggregates and those cells produced high integrin forming the periphery. 34 In the present study, the molecular mechanisms of cell aggregation, the expression level of cadherin and integrin β 1 have been tested. The consistent tendency was also found which was consistent with former report. Certainly, mechanisms behind the morphology form different cell type were more complex than described, which still need more investigation.
In summary, the present study demonstrated that spinner flasks bioreactor provided dynamic 3D environment which induces the numerous aggregation of bACs, rMSCs and mixture of this two cells, respectively. These aggregates can be directly used in cell-based therapies. The key culture parameters to the aggregation process, and the kinetics and molecular mechanism of aggregation were identified. The rate of aggregation and the size could be effectively controlled by various applications. Further observation revealed that chondrocyte phenotype and stemness of MSCs ( Figure S3) were preserved in aggregates. This would serve as foundation for large-scale production of cellular aggregates for cartilage regeneration.

ACK N OWLED G EM ENTS
This work was supported by the National Natural Science

CO N FLI C T O F I NTE R E S T
The authors declare no commercial or financial conflict of interest.