Cell number in mesenchymal stem cell aggregates dictates cell stiffness and chondrogenesis
Although mesenchymal stem/stromal cell (MSC) chondrogenic differentiation has been thoroughly investigated, the rudiments for enhancing chondrogenesis have remained largely dependent on external cues. Focus to date has been on extrinsic variables such as soluble signals, culture conditions (bioreactors), and mechanical stimulation. However, the role of intrinsic mechanisms of MSC programming-based mechanobiology remains to be explored. Since aggregation of MSCs, a prerequisite for chondrogenesis, generates tension within the cell agglomerate, we inquired if the initial number of cells forming the aggregate (aggregate cell number (ACN)) can impact chondrogenesis.
Aggregates of varying ACN were formed using well-established centrifugation approach. Progression of chondrogenic differentiation in the aggregates was assessed over 3 weeks in presence and absence of transforming growth factor-beta 1 (TGF-β1). Mechanical properties of the cells were characterized using high-throughput real-time deformability cytometry (RT-DC), and gene expression was analyzed using Affymetrix gene array. Expression of molecular markers linked to chondrogenesis was assessed using western blot and immunofluorescence.
Reducing ACN from 500 k to 70 k lead to activation and acceleration of the chondrogenic differentiation, independent of soluble chondro-inductive factors, which involves changes to β-catenin-dependent TCF/LEF transcriptional activity and expression of anti-apoptotic protein survivin. RT-DC analysis revealed that stiffness and size of cells within aggregates are modulated by ACN. A direct correlation between progression of chondrogenesis and emergence of stiffer cell phenotype was found. Affymetrix gene array analysis revealed a downregulation of genes associated with lipid synthesis and regulation, which could account for observed changes in cell stiffness. Immunofluorescence and western blot analysis revealed that increasing ACN upregulates the expression of lipid raft protein caveolin-1, a β-catenin binding partner, and downregulates the expression of N-cadherin. As a demonstration of the relevance of these findings in MSC-based strategies for skeletal repair, it is shown that implanting aggregates within collagenous matrix not only decreases the necessity for high cell numbers but also leads to marked improvement in the quality of the deposited tissue.
This study presents a simple and donor-independent strategy to enhance the efficiency of MSC chondrogenic differentiation and identifies changes in cell mechanics coincident with MSC chondrogenesis with potential translational applications.
KeywordsMechanobiology Caveolin-1 N-cadherin Survivin Matrix metalloproteinase-2 Developmental engineering
Aggregate cell number
Argininosuccinate synthase 1
Bradykinin receptor B1
Bone morphogenetic protein 4
Complement component 3
Dickkopf-related protein 1
Dulbecco’s modified essential medium
Frizzled class receptor 4
Green fluorescent protein
hes family bHLH transcription factor 1
Hypoxia-inducible lipid droplet-associated
High mobility group box 2
Insulin-like growth factor
Interleukin-1 receptor-associated kinase 2
Integrin beta 3
Kruppel-like factor 9
Minimum essential medium
Mesenchymal stems cells
Non-essential amino acids
Nuclear factor of kappa light polypeptide gene enhancer in B cell inhibitor
Nuclear receptor subfamily 1, group D, member 1
Nuclear receptor subfamily 4, group A, member 3
Optimal cutting temperature
Partial differential equation
Parathyroid hormone-1 receptor
Real-time deformability cytometry
Salt-inducible kinase 1
Transforming growth factor-beta 1
Tumor necrosis factor receptor superfamily, member 1B
Self-aggregation of a dispersed cell population occurs during different stages of development such as embryogenesis, morphogenesis, and organogenesis, and the common notion is that it is due to intracellular adhesiveness and energy minimization [1, 2, 3]. A vital stage in endochondral ossification (EO) is condensation of mesenchymal stem/stromal cells (MSCs) [4, 5, 6, 7], which starts with the formation of dense cell-cell contacts through adhesion proteins. The initiation of the process, size, boundaries, and differentiation of closely packed MSCs is tightly regulated via transmembrane adhesion proteins such as N-cadherin (N-cad) and N-CAM . The cartilage matrix generated from differentiated MSCs lays the framework for the formation of long bones . To date, efforts to enhance MSC chondrogenesis have focused on extrinsic variables such as soluble factors (transforming growth factor-beta 1(TGF-β1) , insulin-like growth factor (IGF), bone morphogenetic proteins (BMPs)) and optimizing the culture condition using different types of bioreactors including zero gravity and low-perfusion systems [11, 12, 13]. However, in vivo cells experience stresses (hydrodynamic, mechanical deformation) and this can manifest itself by changes to cell shape, cell volume, and membrane tension . For example, it is well established that during development, cell fate and pattern formation is regulated by mechanical forces . In this scenario, cell-cell contact plays an important role in transducing the mechanical forces into intra- and extracellular biochemical cues through activation of signaling pathways. For instance, it has been shown that mechanical stimulation can release latent stores of TGF-β1 from the extracellular matrix (ECM) [16, 17, 18]. Therefore, more recent mechanical stimulation of MSCs associated with a matrix (hydrogels, decellularized tissue) has emerged as another approach to enhanced chondrogenesis [19, 20]. In the aggregation step, MSCs are yet to secrete any extracellular matrix; therefore, the mechanical forces experienced by MSCs have to be derived intrinsically by the tension imposed by cell-cell contact during condensation . Consequently, cell numbers within the MSCs aggregate represent an intrinsic variable whose impact on MSC chondrogenesis has remained unexplored. It is well appreciated that cell-cell adhesion involves cadherins, especially N-cad, which has been shown to play a vital role in MSC chondrogenesis [21, 22] and also has a mechanosensing function , and cytoplasmic anchoring of cadherins to the actin cytoskeleton is known to involve β-catenin , a transcriptional co-activator of MSC proliferation and differentiation . Therefore, we theorized that the aggregation of MSCs could activate a hitherto unknown mechanobiology program for chondrogenesis involving modulation of N-cad and other potential players in mechanotransduction. In this study, we varied the number of human marrow-derived MSCs involved in the formation of aggregates (initial aggregate cell number (ACN)) and found that there is a direct correlation between ACN, mechanical properties of cells, and TCF/LEF-dependent transcriptional activity with MSC chondrogenesis could be modulated through an interplay between N-cad and Caveolin-1 (Cav-1), a protein residing in lipid rafts with a known role in mechanotransduction .
Materials and methods
Mesenchymal stem cell isolation and in vitro culture
Human marrow-derived mesenchymal stem cells (MSCs) were obtained from patients under informed consent in accordance to the regulations of the institution’s ethical committee (University Hospital Basel; ref. number of the local ethical committee: 78/07). In this study, cells from three donors ranging in age from 25 to 50 years (one female and two males) were used. The multi-lineage differentiation (osteogenic, chondrogenic, and adipogenic) potential of MSCs from all donors was confirmed by differentiation assays prior to the commencement of the studies. MSCs were expanded for two passages with α-minimum essential medium (MEM)-based media containing 10% FBS and 5 ng/mL FGF. MSC chondrogenic differentiation was carried out in aggregate culture system . Briefly, 500,000, 350,000, 250,000, 150,000, and 70,000 MSCs were suspended in 0.5 mL serum-free chondrogenic differentiation media in 1.5 mL conical polypropylene tubes (Sarstedt, Germany) and centrifuged at 800 rpm for 3 min to form aggregates. Each condition was established in at least five replicates using independent multiple donors (n = 4–5). Aggregates were then cultured in serum-free medium for chondrogenic differentiation containing high glucose Dublecco’s MEM (DMEM) with non-essential amino acids (NEAA) and glutamine supplemented with 1 mM sodium pyruvate (Gibco, Germany), 100 mM HEPES buffer (PAN-Biotech, Germany), 100 U/mL penicillin, 100 mg/mL streptomycin (Gibco, Germany), and Insulin-Transferrin-Selenium (PAN-Biotech, Germany) supplemented with 0.1 mM ascorbic acid 2-phosphate (Sigma, Germany), 10 ng/mL TGF β-1 (R&D Systems), and 10−7 M dexamethasone (Sigma, Germany) for 3 weeks. The medium was changed twice a week, and the conditioned medium was collected, divided, and stored at 4 °C in the refrigerator for fibronectin quantification with ELISA assay and − 20 °C for zymography analysis. Aggregates were characterized using microscopy, histology, immunohistochemistry, RT-CD, and Affymetrix gene array. The study was performed with at least four biological replicates in the case of each donor and for each condition at each time point, and every biological replicate was measured at least twice. The data acquired from 26-year-old male donor have been presented in the manuscript.
Histology and immunofluorescent staining
Aggregates were fixed in 3.7% paraformaldehyde for 2 h at 4 °C in the refrigerator, washed with PBS, and soaked in 3% sucrose overnight at 4 °C in the refrigerator and then embedded in OCT and cryosectioned. The sections (8-μm thick) were stained for GAG with Safranin-O and Alcian Blue. Immunohistochemical staining for extracellular matrix molecules was carried out using the following antibodies: collagen type II (Iowa University), type X (Abcam), fibronectin (FN) (Abcam), N-cadherin (Abcam), Cav-1 (cell signaling), and β-catenin (Abcam). After washing the samples to remove mounting media, the sections were treated for enzymatic antigen retrieval for Col II and Col X. In the case of Cav-1 and FN trypsin antigen retrieval was done prior to staining, and for N-cadherin and β-catenin, visualization heat antigen retrieval was used. Negative controls were established for each antibody by omitting the primary antibodies and using isotype control. For immunofluorescent (IF) visualization after incubation with primary antibodies, samples were incubated with secondary antibodies labeled with Alexa Fluor 594 and 488 (Invitrogen) followed by DAPI nuclear staining.
A. Uniformity and darkness of Alcian Blue and Nuclear Fast Red staining
Weak staining of poorly formed matrix
Moderately even staining
Even dark stain
B. Distance between cells/amount of matrix produced
High cell densities with no matrix in between
High cell densities with little matrix in between
Moderate cell density with little matrix
Low cell density with moderate distance between cells and an extensive matrix
C. Cell morphologies represented
Mixed spindle/fibrous with rounded chondrogenic morphology
Affymetrix gene array analysis
The total RNA was extracted from three biological samples for each condition using RNeasy Mini Kit (Qiagen kit) as recommended by the manufacturer. The quality and the quantity of isolated RNA were assessed using the NanoDropTM 2000C (Thermo Fisher Scientific). The quality and purity of RNA were further ascertained by ATLAS Biolabs (http://www.atlas-biolabs.com). The cDNA synthesis, amplification, purification, fragmentation and labeling, array hybridization, and scanning were done by ATLAS Biolabs according to the Affymetrix GeneChip™ technology. ATLAS Biolabs performed the microarray gene expression data analysis. The data have been processed using the Affymetrix Power Tools and Bioconductor R. The analyses have been performed using an RMA style background adjustment and quantile normalization (rma-sketch). Intensities are in log2 space. Detection above background (DABG) values was taken into account to filter for signals that are at least present in one group. A significance level of 5% and minimum twofold change cutoff were considered in the analysis.
The spherical aggregate was modeled as a sphere with isotropic continuous material properties with diffusion coefficient D and uptake rate k. The ambient compound concentration was set as a Dirichlet boundary condition for the domain partial differential equation (PDE). Discretization refinement studies were performed until convergence was achieved. To solve the PDEs, COMSOL Multiphysics 5.2a was used.
Real-time deformability cytometry
Mechanical properties of cells are characterized by real-time deformability cytometry (RT-DC) as described elsewhere  using the Accelerator (Zellmechanik Dresden, Germany), a device that utilizes a 30 μm × 30 μm cross-section constriction within a microfluidic chip to measure the hydrodynamic deformation of cells. Cell size and deformation were quantified using high-speed image analysis of up to 1000 cells per second in real time. Prior to the experiment, cells were detached by treatment with trypsin for 5 min or via incubation with 0.3% type II collagenase (Invitrogen) for 1 h at 37 °C in an orbital shaker (in case of aggregates). Cells are centrifuged at 140×g for 5 min and resuspended in 0.6% (w/v) methylcellulose in Mg2+- and Ca2+-free PBS. The prefilled microfluidic chip was flushed with the cell suspension, and the flow was stabilized at 0.16 μl/s for 1 min before measurements. For each condition, approximately 1000 mechanical single cell measurements were acquired. Data analysis was carried out based on experimental duplicates from different days using linear mixed models [30, 31]. Briefly, by performing real-time deformability cytometry (RT-DC) on biological replicates each consisting of several hundreds of cells, the application of linear mixed models allows for separation of random and fixed effects in these large datasets. Fixed effects represent the quantity of interest while random effects summarize all systematic and random measurement bias. Practically, two models are stated, one with and one without the fixed-effect term, and statistical significance is calculated using Wilks theorem.
Western blot analysis
Aggregates were lysed in Ripa buffer, and protein concentration was determined using BCA protein assay kit (Pierce). Twenty mircograms of protein was boiled with Laemmli buffer for 5 min at 95 °C and loaded in 8% (w/v) polyacrylamide-SDS gel and transferred it onto nitrocellulose membranes and blocked with 5% (w/v) BSA for 1 h at room temperature. The membranes were then probed with an antibody against caveolin-1 (Cell Signaling), β-catenin (Abcam), and FN (Technoclone); N-cadherin (Abcam) and GAPDH (Santa Cruz Biotechnology) were used as a loading control. Blots were developed using peroxidase-conjugated secondary antibodies and chemiluminescence system (Thermo Scientific). The intensities of bands were documented by a digital gel-imaging system (Thermo Scientific), and bands were analyzed by densitometry using ImageJ.
Lentiviral production and transduction
Lentiviral particles containing shRNA constructs (7TGC lentiviral plasmid (Wnt reporter system with GFP and mCherry expression), Addgene plasmid 24304) and GIPZ Lentiviral shRNA clone: RHS4439 Open Biosystems, Germany, were produced in HEK293 cells, by co-transfecting lentiviral vector and packaging vectors using polyethylenimine (Mw 25.000, Sigma, Germany) as the transfection reagent. For transfection, 30 μg of DNA (4:3:1 of transfer vector, packaging coding vector (pCMVdR8.74) and envelope coding vector (pMD2.G)) was diluted in 250 μl Opti-MEM (Invitrogen, Germany) and 11.25 μL of polyethylenimine (1 mg/mL) was added to the solution, and the resulting mixture was incubated for 25 min at room temperature prior to adding to HEK293 cells. The medium was changed after 16 h to MSC expansion medium, and 64 h after transfection, the viral supernatants were collected and filtered through a sterile 0.45-μm syringe filter (Millipore, Germany). Viral particles were added to target cells (MSC). Three days after transduction, infected cells were selected by flow cytometry sorting for 7TGC vector and by adding 1 μg/mL puromycin (Sigma, Germany) to the culture medium in case of GIPZ.
Cells within aggregates were digested via incubation with 0.3% type II collagenase (Invitrogen) for 1 h at 37 °C in an orbital shaker. Harvested cells were centrifuged and suspended in flow cytometry buffer (PBS containing 2% FBS). A total of 10,000 events were recorded for each sample using flow cytometry (Gallios; Beckman Coulter). Samples were analyzed with Flowing software, and an average of at least five different samples was calculated.
Polyacrylamide/sodium dodecyl sulfate (SDS) gel electrophoresis (Page) was performed using 10% gels containing 1 mg/mL of gelatin (Sigma, Germany). SDS was removed by washing in 2.5% Triton X-100 for 1 h at room temperature before the enzyme reaction. The gel was incubated overnight at 37 °C in enzyme buffer containing 50 mmol/L Tris, pH 7.5, 200 mmol/L NaCl, 5 mmol/L CaCl2, and 0.02% Brij-35. The MMP-2 activity was identified by the area where the gelatin was degraded which appears as a distinct white band after staining the gel with simply blue (Invitrogen, Germany) for 2 h at room temperature followed by overnight washing with water. The imaging and analysis were done using the same methodology as for western blot analysis.
A sandwich ELISA for the detection of soluble FN was developed in our laboratory. Briefly, 96-well ELISA plates were coated with mouse anti-human monoclonal FN antibody (mAb 6 FN, Technoclone, Austria) and incubated at 4 °C overnight. The plates were then washed with PBS (0.05% Tween 20) and blocked overnight in the fridge. Conditioned media (100 μL) from different samples were added to the wells in triplicate and incubated for 2 h at room temperature. Plates were then washed three times and incubated at room temperature with detection antibody (biotinylated rabbit polyclonal anti-FN (Abcam) for 2 h. The plates were then washed and incubated with UltraAvidin-Horseradish Peroxidase coupled secondary goat anti-rabbit (Abcam) for 30 min at room temperature, and the color was developed with one step substrate (R&D Systems) and the absorbance at 570 nm was measured using a plate reader (BioTek Synergy HT). The absorbance at 450 nm was subtracted from to yield the corrected values for analysis.
Human MSCs were stably transfected with ShRNA for green fluorescence protein, and aggregates were induced using different initial cell numbers. At pre-determined time points, the aggregates were dissociated with 0.3% collagenase at 37 °C and the cell numbers were obtained using a plate reader (BioTek Synergy HT). A calibration curve was established to relate fluorescence intensity to cell numbers, and this was used to determine the cell number in the digest.
In vitro culture using collagenous matrix
One million MSCs were cultured in type I collagen meshes (Ultrafoam®, Davol, Warwick, RI, USA) with 4-mm diameter, and 4 pellets of 70 k were implanted in collagen meshes 2 days of incubation. Both conditions were cultured using serum-free medium for chondrogenic differentiation for 3 weeks. The engineered constructs were cryosectioned and histologically stained (Safranin-O staining) in order to analyze the quality of deposited tissue. The Bern score was assessed for different constructs by virtually dividing each construct to10 circular regions. Bern score was plotted against the distance of each circular region form center of constructs.
All values are reported as mean ± standard deviation and p values greater than 0.05 were set as not significant (ns). Significance is notated with asterisks as follows: *p < 0.05, **p < 0.01, and ***p < 0.001. Statistical analysis (in all measurements except values extracted from RT-DC (see the “Real-time deformability cytometry” section for details)) was performed with OriginPro 2017. For statistics, a one-way analysis of variance (ANOVA) with post-hoc Tukey pairwise comparison was used.
Results and discussion
ACN activates and modulates early phase of chondrogenic differentiation in MSCs independent of exogenous soluble TGF-β1
Lower ACN enhances matrix production during MSC chondrogenesis in a TGF-β1-dependent manner
We then proceeded to quantify the quality of the cartilaginous matrix using the Bern score, an accepted semi-quantitative metric to describe chondrogenesis . In this scoring system, three factors go into assessing the quality of the cartilage matrix: (1) cell morphology, (2) distance between cells/amount of matrix produced uniformity, and (3) intensity of Alcian Blue staining, and each factor is given a score ranging from 0 to 3. Thus, aggregates with highest chondrogenic differentiation have a score of 9, and the ones with the least have a score of 0. The histological analysis and Bern score assessment of aggregates at day 7 of chondrogenic differentiation revealed that addition of TGF-β1 does not alter the Bern score of aggregates significantly (Fig. 1c). In both conditions (± TGF-β1), cells within the low-ACN aggregates particularly in the periphery had already acquired chondrogenic morphology (rounded) and showed expression of GAGs with a Bern score in the range of ~ 3. In comparison, high-ACN conditions promoted a fibroblastic morphology and showed little-to-no GAG production throughout the aggregate cross-section and had a negligible Bern score. This is a significant observation as it suggests that in the early stages of MSCs-derived chondrogenesis, intrinsic signaling plays a dominant role. However, at day 21, although the clear trend of higher chondrogenesis at lower ACN was sustained, the Bern score was consistently higher as expected in presence of TGF-β1. This trend between ACN and GAGs expression was also confirmed in other donors (age 25–56 years old male and female) confirming the generality of the observations (Additional file 1: Figure S2).
Visual inspection of MSCs aggregates using light microscopy revealed that in presence of TGF-β1, despite the fact that the lowest ACN condition (70 k) (low-ACN) had seven times less cells at the onset compared to the highest ACN (500 k) (high-ACN), after 7 days, the volumetric size difference among aggregates was negligible, and after 21 days, the diameter of the high-ACN aggregates was only around 1.5 times of the low-ACN aggregates (Fig. 1d). However, such a dramatic change from day 7 to day 21 was not present in absence of TGF-β1, again confirming that ACN acts as a trigger for chondrogenic differentiation of MSCs; however, matrix production requires exogenous signals. This led us to postulate that either ECM production in aggregates with low-ACN was more efficient, and/or cells within this environment were able to undergo proliferation. Since the effect of ACN was preserved even in presence of TGF-β1, all further analysis was undertaken in presence of TGF-β1.
ACN impacts the expression of genes regulating chondrogenesis and endochondral ossification
Upon analysis of the top 300 differentially regulated genes (min 2.2-fold), we found out that genes involved in regulation of lipid storage (GO:0010883) such as caveolin-1 (CAV1), interleukin 6 (IL6), CD36, nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor (NFKBIA), and integrin beta 3 (ITGB3) and regulation of lipid biosynthetic process (GO:0046890) and regulation of lipid transport (GO:0032368) such as nuclear receptor subfamily 1, group D, member 1 (NR1D1), leptin (LEP), complement component 3 (C3), and salt-inducible kinase 1 (SIK1) were all downregulated in low-ACNs.
Additionally, we did not see significant differences in expression of hypoxia-associated genes such as hypoxia-inducible factors 1 and 3 (HIF1A, HIF3A), egl-9 family hypoxia-inducible factor 3 (EGLN3) , and hypoxia-inducible lipid droplet-associated (HILPDA) between MSCs from various ACNs. This was an interesting observation as it has been widely accepted that multicellular tumor spheroids experience hypoxia [35, 36] and the diffusion of nutrition and oxygen in multicellular aggregates is compromised in direct relation to the aggregates size .
Modeling of oxygen and TGF-β1 diffusion and glucose consumption in different ACNs
Oxygen availability has been frequently suggested as one of the primary factors influencing MSC fate . It has been reported that differentiating embryoid bodies experience gradients of nutrients, oxygen, and cytokines and that the concentration of oxygen at the centers of embryoid bodies with a radius of 400 μm was 50% lower than that in embryoid bodies with a radius of 200 μm . Even though in this study aggregates at day 7 of differentiation have a relatively similar size (Additional file 1: Table S1), in order to rule out the impact of oxygen and nutrient diffusion in the observed outcomes, we modeled the diffusion of oxygen, glucose, and TGF-β1 in aggregates of varying ACNs at day 7 of chondrogenic differentiation.
where c = c (x, t) is the compound concentration at location x and time t, D is the diffusion coefficient, ∆ is the Laplace operator, and k is the uptake rate. The data described in Additional file 1: Table S2 was used.
ACN programs MSCs biomechanics with increasing stiffness in cells correlating with higher chondrogenesis
Expression of mechanosensing proteins N-cadherin and caveolin-1 in MSCs aggregates is modulated by ACN
In order to ascertain the mechanism underpinning the regulation of chondrogenesis by ACN, we investigated the expression of proteins involved cell-cell contact. One of the proteins known to inhibit cell-cell contact in epithelial cells is Cav-1 . Cav-1 is the main scaffolding protein residing in the cholesterol-rich membrane micro-domains (caveolae), which has a documented role in mechanotransduction in endothelial cells  and also implicated in transduction of mechanical forces across cell-cell junctions via stretch-activated channels . Caveolae have been implicated in the compartmentalization and regulation of many signaling events such as MSC renewal and differentiation (adipogenic and osteogenic) , and its expression has been observed during chondrogenesis in the tibiotarsus (avian limb) and in chondrocytes in the vicinity of the proliferating zone within the cartilage  Furthermore, Cav-1 knockout mice show an increase in length of growth plate, number of hypertrophic cells, bone size, and stiffness [52, 53]. Notwithstanding, the relevance of Cav-1 in MSC condensation and chondrogenesis remains ill-defined.
ACN influences TCF/LEF transcriptional activity and expression of N-cad and Cav-1 is MSCs
ACN impacts expression and activation of MMP-2, a negative regulator of chondrogenesis
The progression of chondrogenesis is accompanied by ECM deposition and remodeling, where matrix metalloproteinases (MMPs) play a critical role in breaking down the ECM leading to release growth factors and unraveling of cryptic binding sites for cells. It has been observed in chick limb bud-derived MSCs that MMP-2 activation during the pre-condensation step inhibits chondrogenesis by negatively regulating cell adhesion . We therefore analyzed the conditioned media from the aggregates at day 7 and day 21 using gelatin zymography and found that secretion of pro-MMP-2 after 7 days increased dramatically (~ 4–5-fold) with increasing ACN, and high-ACN aggregates showed activation of MMP-2 while in the other aggregates, there was no obvious sign of MMP-2 activation. However, by day 21, differences between the expressions of MMP-2 between samples were almost abolished but activation of MMP-2 was still 4–5-fold higher in high-ACN aggregates (Additional file 1: Figure S5). Considering that increasing ACN negatively impacts chondrogenesis, the direct correlation between Cav-1 expression and expression and activation of MMP-2 with increasing ACN led us to further postulate that upregulation of Cav-1 most likely interferes with MSC condensation during chondrogenic differentiation by impacting ECM remodeling. Interestingly, evidence for an association between Cav-1 and MMP-2 has been found in endothelial cells where it has been shown that MMP-2 co-localizes with Cav-1 on the cell surface and that Cav-1 contains both the proposed receptor and the activator of MMP-2 .
Low-ACN promotes expression of fibronectin by MSCs
ECM secreted by cells can have a modulatory role in cell adhesion. FN, a high-molecular-weight glycoprotein, is a key component of the ECM of MSCs, and its expression is upregulated during MSC condensation both in vitro and in vivo [6, 67]. FN possesses cell adhesion domains that play a vital role in mediating cell-cell contact, and it has been suggested that the role of FN in limb bud might be to provide a scaffold for MSCs and promote the formation of pre-cartilage cellular aggregates [6, 68]. FN also supports the deposition of collagen and binding protein for latent TGF-β1 and therefore also has an instructive role in a cartilage matrix formation. WB analysis revealed differences in the expression of fibrillar FN between low-ACN and high-ACN aggregates starting at day 2, with a three- to fourfold higher expression of FN after 7 days of differentiation in low-ACN aggregates compared to high-ACN aggregates (Fig. 5a). By day 21, the pronounced difference was absent and FN expression in aggregates with low-ACN was markedly diminished suggesting a maturation/homeostasis in chondrogenesis (Additional file 1: Figure S4). Since the insoluble fibrillar form of FN that is present in the ECM is assembled from secreted soluble FN through complex cell-mediated process [69, 70, 71], we therefore quantified the amount of soluble FN in the conditioned media of the aggregates at different time points using an in-house developed ELISA (see the “Materials and methods” section for details). We made a compelling finding that the soluble FN in conditioned media from different aggregates throughout the differentiation phase was in a narrow range (11–24 μg/mL) suggesting some autoregulation; however, when normalized to DNA content within the aggregates (i.e., normalization to cell number), FN secretion in early stages of differentiation (day 2 and day 7) in low-ACN aggregates was ~ 2–3-fold higher compared to high-ACN aggregates (Additional file 1: Figure S6). This finding also correlates with WB analysis of fibrillar FN in the ECM at days 2 and 7 (Fig. 5a). This is the first report showing a direct correlation between the FN secretion capacity of MSC and its potential to undergo chondrogenic differentiation and alludes to the possible use of FN expression as a biomarker for the chondrogenic potential of human MSCs. The FN level in low-ACN aggregates also correlates with higher β-catenin transcriptional activity (Fig. 6, Additional file 1: Figure S7), and this conforms to reports in the literature that FN is a direct target of β-catenin signaling [72, 73]. This is also in agreement with our proliferation data (Fig. 3d) as it has been reported that increased FN levels during acinar development result in overproliferation of mammary epithelial cells and acinar size . Interestingly, a recent study has shown that in rat trabecular meshwork cells, stiffness correlated with the expression of FN .
Our finding alludes to a complex mechanism involving activation of the canonical Wnt pathway, changes to Cav-1 and FN expression, and cell stiffness in the regulation of chondrogenesis.
ACN exerts control over MSCs proliferation through regulation of survivin
Toward engineering more efficient constructs for bone and cartilage tissue engineering and regenerative medicine
In the bone and cartilage tissue engineering, acquiring a sufficient number of the cells for in vitro culture or in vivo implantation is costly, time-consuming, and remains the rate-limiting step for clinical translation due to the inverse correlation between MSCs differentiation potential and expansion time. To address this issue, we have investigated if implanting few low-ACN in the collagenous matrix will lead to superior outcomes in comparison with traditional methodology of dispersing cells uniformly throughout the matrix. We implanted four 70 k aggregates 2 days post-culture within the collagen matrix and compared it to the outcomes in a collagen matrix seeded with one million MSCs. After 3 weeks of chondrogenic differentiation, the cells within the aggregates were fully mature hypertrophic chondrocytes and were incorporated in the matrix and the deposited tissue was highly homogenous (Additional file 1: Figure S8a). In spite of the total cell number within the aggregate condition being almost four times lower than the conventional culture condition, the outcomes as assessed by the Bern score, which is indicative of the uniformity of the cartilage matrix, was both high and uniform in the aggregate condition (Additional file 1: Figure S8b). Furthermore, interestingly, the developmentally inspired approach inhibited the formation of the hypoxic and necrotic zone in, which occurs due to diminished oxygen and nutrition diffusion in large constructs. This finding validates the premise of our study and demonstrates the framework for potential translational applications.
Induction of stable chondrogenesis from MSCs is crucial for cartilage tissue engineering, repair of cartilage lesions using MSCs, and bone regeneration using EO paradigms. While significant effort has gone into optimizing culture conditions, serial expansion of cells in vitro for generation of adequate cell numbers for manipulation (tissue engineering and in vivo implantation) comes at the expense of loss in chondrogenic potential. In this study, a donor-independent solution is presented that addressed donor variability and loss of chondrogenic phenotype. We have demonstrated a direct relationship between the cell numbers during MSC aggregation, MSC proliferation, and chondrogenic differentiation. Furthermore, enhanced chondrogenesis correlates with the emergence of a stiffer MSC phenotype, which is accompanied by regulation of proteins involved in mechanotransduction namely Cav-1 and N-cad. Interestingly, a higher expression of survivin, an apoptosis inhibitor, is observed in the low ACN environments. To exalt the translational potential of our findings, in a proof of concept study, it is demonstrated that chondrogenesis that is superior to a conventional approach can be achieved using fourfold less cell numbers by implanting low-ACN aggregates in a collagen scaffold. Our analysis showed that the aggregate approach yields superior outcomes by inhibiting the formation of necrotic core and increasing efficiency of matrix deposition. The results of our study provide compelling evidence for a role for cellular mechanics in chondrogenic differentiation of MSCs in 3D aggregates with implication for understanding the mechanisms involved in skeletogenesis and MSC-based regenerative therapies.
The authors thank Prof. Ivan Martin and Dr. Andrea Barbero from University Hospital Basel for providing human bone marrow MSCs and constructive discussions. The authors also wish to thank Esther Kohler and Miriam Heizmann for the technical assistance with histology, Dr. Jon Christensen or the assistance with MSCs transfection, Dr. Pavel Salavei for the assistance with flow cytometry, and Doreen Biedenweg for the technical assistance with RT-DC measurements.
This work was funded by a grant from the Swiss National Foundation-Sinergia Grant (Nr. CRSII3_136179), the excellence initiative of the German Federal and State Governments Grant EXC 294 and the Helmholtz Zentrum Geesthacht through the Helmholtz Virtual Institute on Multifunctional Biomaterials for Medicine, and Bundesministerium für Bildung und Forschung (ZIK HIKE grant to Oliver Otto under grant agreement 03Z22CN11).
Availability of data and materials
All data generated or analyzed during this study are included in this published article (and its supplementary information files).
MS and VPS conceived and designed the study. analyzed the data, and wrote the paper with input from other the authors. MS performed the experiments. OO analyzed the RT-CD data. ST carried out the COMSOL modeling. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Human marrow-derived mesenchymal stem cells (MSCs) were obtained from patients under informed consent in accordance with the regulations of the institution’s ethical committee (University Hospital Basel; ref. number of the local ethical committee: 78/07).
Consent for publication
Oliver Otto is a co-founder of a company developing and distributing real-time deformability cytometry.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 12.Demarteau O, Jakob M, Schafer D, Heberer M, Martin I. Development and validation of a bioreactor for physical stimulation of engineered cartilage. Biorheology. 2003;40(1–3):331–6.Google Scholar
- 29.Otto O, Rosendahl P, Mietke A, Golfier S, Herold C, Klaue D, Girardo S, Pagliara S, Ekpenyong A, Jacobi A, Wobus M, Topfner N, Keyser UF, Mansfeld J, Fischer-Friedrich E, Guck J. Real-time deformability cytometry: on-the-fly cell mechanical phenotyping. Nat Methods. 2015;12(3):199–202 4 p following 202.CrossRefGoogle Scholar
- 32.Leonard CM, Fuld HM, Frenz DA, Downie SA, Massague J, Newman SA. Role of transforming growth factor-beta in chondrogenic pattern formation in the embryonic limb: stimulation of mesenchymal condensation and fibronectin gene expression by exogenenous TGF-beta and evidence for endogenous TGF-beta-like activity. Dev Biol. 1991;145(1):99–109.CrossRefGoogle Scholar
- 62.Howard S, Deroo T, Fujita Y, Itasaki N. A positive role of cadherin in Wnt/beta-catenin signalling during epithelial-mesenchymal transition. PLoS One. 2011;6(8). https://doi.org/10.1371/journal.pone.0023899.
- 68.Kulyk WM, Upholt WB, Kosher RA. Fibronectin gene expression during limb cartilage differentiation. Development. 1989;106(3):449–55.Google Scholar
- 69.Xu J, Mosher D. Fibronectin and other adhesive glycoproteins. In The extracellular matrix: an overview. Springer; 2011. p. 41–75.Google Scholar
- 75.Wang C, Li L, Liu ZC. Experimental research on the relationship between the stiffness and the expressions of fibronectin proteins and adaptor proteins of rat trabecular meshwork cells. BMC Ophthalmol. 2017;17. https://doi.org/10.1186/s12886-017-0662-5.
- 79.Torres VA, Tapia JC, Rodriguez DA, Lladser A, Arredondo C, Leyton L, Quest AFG. E-cadherin is required for caveolin-1-mediated down-regulation of the inhibitor of apoptosis protein survivin via reduced beta-catenin-Tcf/Lef-dependent transcription. Mol Cell Biol. 2007;27(21):7703–17.CrossRefPubMedGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.