Stirred culture of cartilaginous microtissues promotes chondrogenic hypertrophy through exposure to intermittent shear stress

Abstract Cartilage microtissues are promising tissue modules for bottom up biofabrication of implants leading to bone defect regeneration. Hitherto, most of the protocols for the development of these cartilaginous microtissues have been carried out in static setups, however, for achieving higher scales, dynamic process needs to be investigated. In the present study, we explored the impact of suspension culture on the cartilage microtissues in a novel stirred microbioreactor system. To study the effect of the process shear stress, experiments with three different impeller velocities were carried out. Moreover, we used mathematical modeling to estimate the magnitude of shear stress on the individual microtissues during dynamic culture. Identification of appropriate mixing intensity allowed dynamic bioreactor culture of the microtissues for up to 14 days maintaining microtissue suspension. Dynamic culture did not affect microtissue viability, although lower proliferation was observed as opposed to the statically cultured ones. However, when assessing cell differentiation, gene expression values showed significant upregulation of both Indian Hedgehog (IHH) and collagen type X (COLX), well known markers of chondrogenic hypertrophy, for the dynamically cultured microtissues. Exometabolomics analysis revealed similarly distinct metabolic profiles between static and dynamic conditions. Dynamic cultured microtissues showed a higher glycolytic profile compared with the statically cultured ones while several amino acids such as proline and aspartate exhibited significant differences. Furthermore, in vivo implantations proved that microtissues cultured in dynamic conditions are functional and able to undergo endochondral ossification. Our work demonstrated a suspension differentiation process for the production of cartilaginous microtissues, revealing that shear stress resulted to an acceleration of differentiation towards hypertrophic cartilage.

ones while several amino acids such as proline and aspartate exhibited significant differences. Furthermore, in vivo implantations proved that microtissues cultured in dynamic conditions are functional and able to undergo endochondral ossification.
Our work demonstrated a suspension differentiation process for the production of cartilaginous microtissues, revealing that shear stress resulted to an acceleration of differentiation towards hypertrophic cartilage.  5,6 and periosteum (hPDCs). 7 This strategy has been explored in multiple tissue formats such as chondrogenically primed microaggregates, 8 hypertrophic microtissues, 4 hollow tubes 9 and cell sheets 10 which resulted in regeneration of critical size long bone defects.
In a recent study, 4 the use of planar culture technologies through the use of non-adherent microwells resulted in the formation of cartilaginous microtissue modules, able to self-assemble in larger implants and regenerate large tibial defects in murine animal models. However, suspension culture of the abovementioned functional microtissues, needs to be further explored, as this will further pave the way towards the development of scaled up bioprocesses able to produce clinically-relevant amounts of microtissue populations. Bioreactors allow for scalability, while providing capacity for real-time monitoring and control of the culture and differentiation process of stem cells. 11 Stirred tank reactors represent a universal, well-established vessel type for the production of adult progenitor cells on microcarriers 12 and human pluripotent stem cells. 13 These systems have also been used to culture articular chondrocytes seeded on microcarriers. 14 Several differentiation protocols for the production of differentiated induced pluripotent stem (iPS) cells in stirred tank bioreactors have been reported for different tissues including cardiac, 15,16 neural 17,18 and kidney 19 cell types. Few studies have focused on the static suspension culture of cartilage microtissues from iPS cells 20 or chondrocyte microtissues from bovine source in a stirred culture environment 21 with limited process characterization. We have previously explored a microcarrier-based stirred tank suspension process for expansion and differentiation of human adult periosteal derived progenitors with incomplete chondrogenic differentiation and maturation due to limited cell condensation on the microcarrier surface, 22 hence there is still room to explore the potential to culture microtissues in stirred bioreactor systems.
An inherent component in dynamic suspension culture, is the effect of fluid-generated shear stresses on the culture progenitor cells or microtissues. Currently there is lack of knowledge on the impact of process-generated shear stresses on chondrogenic maturation towards hypertrophy in suspension cultures of cartilaginous microtissues. Furthermore, to date, in vitro studies have reported mixed results related to the impact of mechanical stress on chondrogenic differentiation with diverse mechanical regimes enhancing primary chondrocyte hypertrophy 3 or inhibiting it during adult MSC culture. 5 For a thorough characterization of the process environment to be made, CFD models have been developed. 23 However, a limitation of these approaches is that they rely on bulk estimates without considering the discrete nature of these microtissue suspensions 24 and the active (adhesive) properties of cell aggregates/microtissue structures.
In this study, we aimed at investigating suspension culture of cartilaginous microtissues and the role of fluid-generated shear stresses on the process of chondrogenic differentiation towards hypertrophy.
To carry out these experiments, we produced a microbioreactor system that provided the same volume and microtissue density as to that of the static-planar setup. Mechanical stimulation during culture on single microtissues was characterized and quantified through simulations with advanced mathematical models representing individual microparticles instead of only the fluid component. Additionally, exometabolomics analysis was conducted to further characterize the cellular state due to the mechanical stimulation. Our work suggests that culture in suspension bioreactors is possible, and that shear stress alters the phenotypic and metabolic state of the microtissue with a rapid commitment to hypertrophy.
The small microtissues were assembled in an Aggrewell800 well culture plate to create a second static control with larger assembled microtissues (referred in the manuscript as "static large microtissues").
One well of the Aggrewell800 well plate contains an array of 300 μwells with 800 μm in size. The well plate was pre-treated using the Anti-Adherence Rinsing Solution (STEMCELL Technologies) to reduce surface tension and prevent adhesion. Microtissues from the agarose μ-wells (2000 microtissues) were flushed out and transferred to the AggrewellTM800 (300 microwells) 2 days after seeding. This would result in approximately seven small microtissues that fuse into a larger microtissue ( Figure 2).

| Bioreactor design and culture
The mini-bioreactor (miniBR) setup consisted of four impeller rows, each with six impellers adapted to fit onto a 24 well plate (Corning). The impeller rows and exchangeable impellers were printed using polyamide 12, Nylon (PA12) and selective laser sintering (Formando, BE). The impeller row and impellers were connected with bevel brass gears (Reely) and metal bearings (W 604-2RS1, SKF) to allow smooth stirring. Each row of six impellers was driven by a ST3518 stepper motor (Nanotec) driven by SMCI-12 motor controllers (Nanotec). In this way, 24 mini-stirred tank reactors with a volume of 2 ml were created using four impeller rows.
An in-house developed software (Windows Forms application, .NET) written in Visual Basic was used to control the four impeller rows individually. Before use, the 24 well plate was coated with Sigmacote (Sigma Aldrich) to prevent adherence of the microtissues and all components were gas sterilized (Ethylene Oxide). The bioreactor was placed in an incubator at 37 C, 5% CO 2 and 95% humidity to achieve appropriate cell culture conditions. Microtissues cultured for 2 days were carefully flushed out from their microwells and 2000 microtissues were added to one well of the coated 24 well plate which corresponds to the microtissue density in the static control. The impellers were inserted and started as fast as possible to avoid microtissue fusion and the mini-BR were run at 7, 13, and 20 rad/s corresponding to low, medium, and high speed for 14 days. Chondrogenic media was changed 2 times per week.

| Microtissue characterization
Microtissues were characterized microscopically during the differentiation process to assess microtissue size, proliferation, and viability. Cell After removal of the staining solution, the samples were imaged in their media using a fluorescence microscope (Olympus IX83).

| DNA quantification and gene expression analysis
Microtissues from one well were pooled together to represent one sample and lysed in 350 μl RLT buffer (Qiagen) and 3.5 μl β-mercaptoethanol (VWR). DNA was quantified using the Qubit dsDNA HS Assay Kit (Invitrogen). Briefly, 10 μl lysed sample was diluted in 90 μl milliQ water.
Next, 5 μl was added to 195 μl of working solution, vortexed and incubated for 5 minutes at room temperature. The DNA content was measured using a Qubit R Fluorometer. Next, RNA was isolated using RNeasy Mini Kit (Qiagen) whereafter RNA concentration and quality was assessed with NanoDrop 2000 (Thermo Scientific). Complementary DNA (cDNA) was synthesized with PrimeScript reagent kit (TaKaRa) followed by quantitative real-time polymerase chain reaction (qRT-PCR) using SYBR ® Green (Life Technologies) and StepOnePlus R Real-Time PCR System (Applied Biosystems). The heating cycle was as follows: hold at 45 C for 2 min, at 95 C for 30 s, followed by 40 cycles of 95 C for 3 s and 60 C for 20 s. Relative differences in expression were calculated using the 2 ÀΔΔCt method normalized to the housekeeping gene Hypoxanthine-guanine phosphoribosyltransferase 1 (HPRT1). 28

| Formation of microtissue-based implants
Microtissues from two wells of a 24 well plate (containing approximately $2400 microtissues) cultured statically we collected on day 14. In addition, from the miniBR after14 days microtissues were assembled from two well of a 24 well plates. In both cases microtissue suspensions were dispensed in fresh wells containing in their bottom, a layer of agarose with an inverted conical 3 mm diameter well (3 mm in diameter) developed inhouse. Microtissues were allowed to sedimented for 60 min at 37 C, 5% CO 2 and 95% humidity leading to the entrapment of the entire population of suspended microtissues in the agarose well. Subsequently chondrogenic media was added and the microtissue assemblies were let to fuse for 24 h leading to the formation of a mechanically stable mesotissue.

| In vivo implantation analysis
A subcutaneous mouse model was used to assess the microtissues' capacity to form bone after assembly. The fused mesotissues were implanted subcutaneously in immune compromised mice (Rj:NMRInu/ nu) and explants were retrieved after 4 weeks and fixed in 4% PFA.

| Sampling and metabolite extraction
Ιn order to avoid bias due to sampling, we used all the media for the exometabolomic analysis. Also samples were taken each time (D0, D7, D14) 2 days after the media changes. In that way, we had consistency for changes related to the refreshing of the medium. For every sample, 10 μl of 80% methanol with 2 μM d27 myristic acid was added to 990 μl of sample. Extracts were stored overnight at À80 C and were centrifuged.  In the CBM, individual microtissues are represented by deformable spherical particles with a size distribution base on microscopy measurements. Using a discrete element-like approach, we solved the equation of motion for each particle to simulate how the micro-tissues moved and interacted. For this, we explicitly determined the forces acting on individual particles at each time step, see supporting information (ref 2, 3m). Furthermore, based on the forces acting on the particles and local fluid velocities we estimated, for example, the magnitude of shear stress acting on the microtissues. We estimate drag and lift forces acting on the particles due to the local fluid flow by probing the local fluid velocity, pressure, and shear gradients, see supporting information. Given the low Stokes number, Stk ¼ t0vt

| Liquid chromatography-mass spectrometry (LC-MS) analysis
with v t , R p , ρ p the particle terminal velocity, radius and density, micro-tissues are expected to follow the streamlines closely and behave as stream tracers, allowing for a one-way CFD-CBM coupling. Furthermore, due to Re p ¼

| In-silico characterization of the dynamic process environment in mini-bioreactors
To allow high-throughput screening of culture in stirrer-based bioreactor culture, a mini-bioreactor (miniBR) system was developed.
Centimeter-sized impellers were 3D printed and attached in parallel to a motor to fit commercially available 24-well plates (Figure 1a).  Figure 1d. We found that the stagnant region persists in a range of relative spheroid mass density ρ p À ρ f À Á =ρ f between 0 and 0.06.

| Cartilaginous differentiation in dynamic and static culture environments
In order to generate cartilaginous microtissues, we first allowed cells to aggregate for 2 days, forming stable microtissues, before inoculat-

| Exometabolomics of chondrogenic differentiation
We  hypertrophic differentiation and endochondral ossification. 30 As observed by our mathematical modeling study, cartilaginous microtissues were intermittently exposed to regions of high shear stress and low shear stress regimes within the bioreactor system. The distinct difference in proliferation between static and bioreactor conditions could be linked to the commitment of cells towards ECM production rather than proliferation something that is known during chondrogenic differentiation 31  There are several recent studies underlying the importance of studying metabolism as a critical regulator for bone regeneration process. Chondrocyte metabolism and more specifically glutamine metabolism controls the collagen synthesis and modification 39 and targeting skeletal metabolism has been shown as advantageous for the expansion and self-renewal of skeletal progenitors. 40 In this study, we aim for exometabolomic analysis, as a first effort to illustrate changes in the metabolome in the dynamic culture and due to the mechanical stimulation. From a metabolic perspective, cartilaginous microtissues showed increased glucose consumption and lactate production during culture time, indicative of a highly glycolytic metabolism which is a characteristic of growth plate chondrocytes. 41,42 Higher glycolysis has been also reported in chondrocytes under compression. 43 Additionally, lactate accumulation that is higher in the dynamic condition can be beneficial for bone regeneration. Lactate has been observed to stimulate collagen deposition in an autocrine and paracrine manner, thereby contributing to soft callus progression, 44 although further in vivo studies will be required to corroborate this for the specific type of microtissue implants. Regarding the amino acid profiles, among the most interesting differences in the context of chondrogenic differentiation, between static and dynamic conditions was proline secretion, which can be linked with the process of extracellular matrix remodeling as seen for similar chondrogenic differentiation processes. 4 Proline, as a product of ECM degradation can be used as a precursor of other amino acids and as an energy source. 45,46 Moreover, mechanical stimulation in the bioreactor condition can stimulate the synthesis of matrix degrading enzymes. 47 Interestingly, we observed aspartate and glutamate secretion significantly higher in the dynamic conditions especially in Day 7. These secretome profiles can be linked with hypertrophic-like biology, as aspartate is shown to be elevated in synovial fluids from OA patients 48 whereas aspartate release significantly increases by 90% in IL-1β stimulated chondrocytes. 49 The serine secretome profile also showed distinct differences between static and dynamic conditions. Taken together, these metabolites provide a panel of metrics that could be measured from the supernatant providing a target for non-destructive assessment of the progression of chondrogenic differentiation towards hypertrophy. These secreted metabolites could in the future provide markers that could be monitored during bioreactor culture for non-destructive assessment of phenotypic state of the cultured microtissues, however further validation also through in vivo studies, is required.

| CONCLUSIONS
In this work, we investigated the impact of dynamic culture conditions during cartilaginous differentiation in microtissues using a novel

CONFLICT OF INTEREST
Gabriella Nilsson Hall is now an employee in Astra Zeneca.