Effect of expansion media and fibronectin coating on growth and chondrogenic differentiation of human bone marrow-derived mesenchymal stromal cells

In the field of regenerative medicine, considerable advances have been made from the technological and biological point of view. However, there are still large gaps to be filled regarding translation and application of mesenchymal stromal cell (MSC)-based therapies into clinical practice. Indeed, variables such as cell type, unpredictable donor variation, and expansion/differentiation methods lead to inconsistencies. Most protocols use bovine serum (FBS) derivatives during MSC expansion. However, the xenogeneic risks associated with FBS limits the use of MSC-based products in clinical practice. Herein we compare a chemically defined, xenogeneic-free commercial growth medium with a conventional medium containing 10% FBS and 5 ng/ml FGF2. Furthermore, the effect of a fibronectin-coated growth surface was investigated. The effect of the different culture conditions on chondrogenic commitment was assessed by analyzing matrix deposition and gene expression of common chondrogenic markers. Chondrogenic differentiation potential was similar between the FBS-containing αMEM and the chemically defined medium with fibronectin coating. On the contrary, the use of fibronectin coating with FBS-containing medium appeared to reduce the differentiation potential of MSCs. Moreover, cells that were poorly responsive to in vitro chondrogenic stimuli were shown to improve their differentiation potential after expansion in a TGF-β1 containing medium. In conclusion, the use of a xenogeneic-free medium provides a suitable alternative for human bone marrow MSC expansion, due the capability to maintain cell characteristic and potency. To further improve chondrogenic potential of BMSCs, priming the cells with TGF-β1 during expansion is a promising strategy.


Results
Effect of different expansion protocols on BMSC growth and morphology. As expected, media composition did not affect cell viability, as assessed by Live/Dead staining. The presence or absence of a fibronectin coating also had no apparent effect, even though the combination of fibronectin coating with αMEM/FBS/ FGF2 appears to result in a lower number of cells (Fig. 1).
All cells in the presence of a chemically defined medium (xeno/serum free medium-XSFM) or αMEM exhibited a fibroblast-like and spindle-shaped morphology with long cell processes (Fig. 2).
Cell size was measured in two consecutive passages: the results showed no significant variation of the average cell diameter for BMSCs expanded in the different conditions, with a higher variability observed in αMEM/ FBS/FGF2 groups (Fig. 3).
The population doubling time (PDT) was similar for cells grown in XSFM or in αMEM/FBS/FGF2, with no statistically significant differences among the groups (Fig. 4A). The PDT was calculated in the range 24-72 h for the cells to be in the exponential phase of cell growth. The growth curve also showed a similar trend, with no difference between the groups between 24 and 72 h (Fig. 4B). As expected, the cells reached a growth plateau at day 7, with similar cell numbers for all groups.

Chondrogenic differentiation of BMSCs expanded in αMEM/FBS/FGF2 or XSFM.
After comparing cell growth in different media, we focused on the chondrogenic differentiation potential of BMSCs after expansion in different conditions. We used two different types of donors that can be classified as good responders (n = 3) or poor responders (n = 2), based on histological results of cells expanded with the standard medium (αMEM/FBS/FGF2, no fibronectin coating, Fig. 5A-C).
Chondrogenic differentiation was evaluated at 21 days by Safranin-O/Fast Green staining of pellet cryosections (Fig. 5A). The histological evaluation for all donors expanded with XSFM did not show any difference compared to αMEM/FBS/FGF2 medium, although there was a tendency towards a stronger intensity for the matrix deposition and internal pellet structure either with or without fibronectin. Pellets from all groups had comparable diameters (data not shown). However, pellets from BMSCs expanded with αMEM/FBS/FGF2 on a fibronectin coating showed in some cases (donors #1, #4 and #5) a considerable dispersion and poor matrix organization compared to the other groups, indicating a greater variability.
The use of XSFM with BMSC from non-responsive donors improved the size of differentiated pellets, but not the intensity for Safranin-O staining.    www.nature.com/scientificreports/ To determine the quality of matrix produced in hBMSC pellets, immunofluorescence staining for type II and I collagen was performed. The use of XSFM or αMEM medium did not influence the deposition of type II collagen within the donors. Donors #1 and #2, as previously observed with Saf-O/Fast-Green staining (Fig. 5), were less responsive to the differentiation induced by TGF-β1, and immunofluorescence confirmed that there was almost no production of type II collagen. The other 3 donors did not show differences in type II collagen staining after cell expansion with different media (Fig. 6A). Type I collagen staining also did not show any particular influence of the expansion medium or fibronectin coating on matrix deposition. Interestingly, the less-responsive donors showed a higher staining intensity for type I collagen compared to the other donors (Fig. 6B).

Gene expression analysis on chondrogenic markers.
Histological evaluation showed that the use of XSFM medium during in vitro expansion led to similar differentiation when compared to medium containing serum and 5 ng/ml FGF2. We further investigated the expression of common molecular markers associated with chondrogenic differentiation. Analysis of cartilage related genes expression (COL2A1, COL1A1, COL10A1, ACAN, PRG4 and SOX9, Fig. 7A-G) showed no difference between groups, although a positive trend in XSFM expanded cells compared to αMEM/FBS/FGF2, except for PRG4 (Fig. 7F), can be observed. Similarly, αMEM/  www.nature.com/scientificreports/ FBS/FGF2 tends towards an increase in COL2A1/COL10A1 ratio, but this did not reach significance (Fig. 7D). The expression of cartilage hypertrophy-associated markers, namely COL10A1 and MMP13 (Fig. 7C,H), were also similar between different groups, with no statistically significant differences. We further studied the genes associated to osteogenic differentiation, RUNX2, IBSP (Fig. 7I,K) and the balance of chondrogenesis and osteogenesis though RUNX2/SOX9 ratio (Fig. 7J). Results showed no significant differences among groups of media and the use or not of fibronectin for RUNX2 expression and RUNX2/SOX9 ratio. The expression of IBSP ( Fig. 7K) showed an increased trend for the group αMEM/FBS/FGF2 compared to XSFM, but it was not statistically significant. Overall, the results from histological evaluations, as well as from gene expression analysis, showed that cells in a chemically defined medium displayed similar subsequent chondrogenic differentiation.
Non-responsive donor priming. In a separate experiment, we tested the effect of adding TGF-β1 to the standard expansion medium on subsequent chondrogenesis, in the attempt to improve differentiation of poorly responsive cells. Specifically, poorly responsive cells were expanded in αMEM/FBS/FGF2 as previously described. However, in the final 5 days of expansion, TGF-β1 was added either at increasing concentration (1 to 10 ng/ml) or at 10 ng/ml. Cell priming with gradual increase or constant concentration of TGF-β1 during the expansion in monolayer in αMEM strongly improved chondrogenic differentiation (Fig. 8A) in terms of qualitative scoring, compared to cells expanded without the addition of TGF-β1 in monolayer cultures (n = 2). These findings show that the use of TGF-β1 during cell expansion can be beneficial for chondrogenesis and can have a pivotal role in priming non-responsive donors.
As previously observed by Saf-O/Fast-Green staining, BMSCs cultured with or without TGF-β1 differently responded to chondrogenic differentiation in pellet culture. Indeed, the use of 1-10 ng/ml and 10 ng/ml TGF-β1 significantly improved the yield of type collagen II production in non-responsive donors, specifically in the www.nature.com/scientificreports/ picture donor #1 is shown (Fig. 8B). Previously we observed (Fig. 6A,B) that the donors with a poor type II collagen deposition had a high type I collagen content; here the non-responsive donors primed using TGF-β1 did not show a massive change in the type I collagen staining intensity, with a slight reduction in fluorescence intensity in the 10 ng/ml TGF-β1 group (Fig. 8C).

Discussion
Human-derived mesenchymal stem cells (MSCs) are a promising cell source for clinical applications in regenerative medicine, due their relative ease of harvest 25 . However, one of the most difficult tasks in cell manufacture is controlling their quality. FBS has been used as a primary component of cell expansion media for decades 26,27 , principally because it promotes adhesion and cellular expansion. However, due to several problems associated with batch-to-batch variability, uncontrolled composition that may vary from supplier to supplier, as well as the risk of zoonosis and transmissions of proteins not desired at the cellular level, its use is still challenging in a Good Manufacturing process (GMP) environment. One of the reasons why the application of MSCs in regenerative medicine appears to be limited is due to the non-standardized expansion protocols used in various labs. Methods involving the use of platelet lysates have been proposed 28 , but also this alternative is affected by innumerable variables, mostly associated to the donor variation 29 . In our study we have shown that the use of a chemically defined medium without FBS (XSFM) has the same performance as the gold standard medium containing FBS and 5 ng/ml FGF2 starting with the same condition of cells isolated and then cryopreserved in the presence of FBS. This suggests XSFM medium as a clinically relevant alternative.
Fibronectin coating has been proposed as a beneficial method to enhance MSC adhesion onto plastic surfaces 30 . Fibronectin does not limit cell growth in XSFM medium; on the contrary, cells expanded in αMEM/ FBS/FGF2 were negatively affected in terms of growth.
We investigated the expression of main chondrogenic markers after 21 days of differentiation, concluding that the use of XSFM medium did not modify the cell potency, indeed no significant difference in the expression of the main chondrogenic markers was observed compared with αMEM/FBS/ 5 ng/ml FGF2.
Duan et al. previously demonstrated that TGF-β and BMP rapidly mobilizes additional TGFBRs from intracellular stores to the cell surface, increasing the abundance of cell-surface TGFBRs in HaCaT and A549 cell lines, priming the downstream TGF-β signaling 31 . Based on this report and on our previous work on the TGFBR expression profile 32 , we investigated the potential of late addition of TGF-β1 into the expansion medium as a www.nature.com/scientificreports/ means to improve chondrogenic differentiation. Poorly responsive cells, after expansion in our standard medium, were grown again in αMEM/FBS/FGF2 and in the final week of expansion either increasing concentrations of TGF-β1 or a constant 10 ng/ml TGF-β1 was added. We observed that there was a significant enhancement of chondrogenic differentiation in terms of matrix deposition. This enhancement was similar to that seen in previous studies using siRNA against TGFBR2 32 , thus offering a powerful method to modulate MSC phenotype that is easily applied. This also demonstrates that cell phenotype can be modified even in late stages of expansion. Therefore, TGF-β1 addition during the expansion process can prime chondrogenic differentiation, probably by enhancing the downstream pathways associated with Small Mother Against Decapentaplegic (SMAD) proteins. Indeed, other studies have highlighted the role of TGF-β signaling associated with recruitment of the ALK5 (TGFBR1) receptor, with the consequent activation SMAD 2/3 pathway, while excess of TGF-β can lead to activation of ALK1 (ACVRL1), promoting the activation of SMAD 1/5/8 33 . We know from previous studies that these receptors are present on MSCs and are essential for chondrogenesis 34 . Furthermore, it has been shown that increasing TGF-β concentration shifts the balance of ALK signaling from ALK5 (SMAD 2/3) at low dose to ALK1 (SMAD 1/5/8) at higher doses 35 . Further analysis is ongoing to understand the possible mechanism during the priming of MSC and the role of TGF-β1 in the expansion medium.
The main limitation of this study resides in the use of previously banked BMSCs. As described in the materials and methods section, our procedure of BMSC isolation is highly standardized to minimize manipulationdependent sources of variation. The procedure involves the use of a tested batch of MSC-qualified FBS and the cryopreservation at Passage 1 in an FBS/DMSO solution. For this reason, this study was conducted with cells that were exposed to xenoderivatives during the isolation phase and first duplications in vitro. Future studies will be focused on analyzing more in detail the effect of different media composition during the isolation phase, where the presence of a fibronectin coating might be essential, and the overall performance of cells which were never treated with any animal-derived products.
In summary, this report describes how the use of xenogeneic serum-free medium during BMSC expansion provides the same performance compared to FBS containing medium with respect to chondrogenic commitment. www.nature.com/scientificreports/ Furthermore, the use of TGF-β1 during the later stages of cell expansion strongly improves the yield of differentiation in those donors defined as less responsive to TGF-β1 stimulation. With the vision of clinical translation, the use of serum-free medium could pave the way to future application in the clinical scenario by allowing chemically defined cell culture, thus limiting variations among different FBS batches. In addition, a xenogeneic serum-free medium avoids a potential risk for developing zoonosis in patients treated with cultured cells. The use of serum-free medium that has the same performance as the gold standard represents a step further for the use of cells, tissue engineered constructs and cell related products that can be suitable for the treatment of pathological and traumatic situations.

Methods
All methods were performed in accordance with the relevant guidelines and regulations. www.nature.com/scientificreports/ Isolation and expansion of human mesenchymal stem cells from bone marrow. Bone marrow from 5 different donors was harvested from vertebral body after signed informed consent and full ethical approval (Ethic -Commission, Albert-Ludwigs-University Freiburg, Germany, AN-EK-135/14). Fresh bone marrow was diluted 1:4 and layered on top of Histopaque-1077 solution (Sigma-Aldrich), in a proportion of 2.6 ml of Histopaque per ml of undiluted marrow. After centrifugation at 800 g for 20 min, the mononuclear cellcontaining interface was recovered, and cells were counted using the Cell Scepter 2.0 Automated Cell Counter (Millipore). Isolated cells were seeded at a density of 50,000 cells/cm 2 into 300 cm 2 tissue culture flasks in Minimum Essential Medium Eagle, Alpha Modification (αMEM; Gibco, Thermo Fisher, Zürich, Switzerland) containing 10% batch-tested FBS for MSC expansion (PAN-Biotech, Aidenbach, Germany), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco), and 5 ng/ml recombinant human basic Fibroblast Growth Factor (FGF2, Fitzgerald Industries International, Acton, MA, USA). Cells were maintained at 37 °C in 5% CO 2 , 95% humidity atmosphere. After 4 days, non-adherent hematopoietic cells were removed to select the BMSC population. Medium was then refreshed every 2nd day. Cells at passage 1 were frozen in 8% dimethyl sulfoxide (DMSO) and 92% FBS. After thawing, and during the cell expansion phase, two different media were compared. One was αMEM medium containing 10% FBS and 5 ng/ml FGF2 described above for cell isolation (αMEM/FBS/FGF2), while the second was a chemically defined medium (PRIME-XV MSC Expansion XSFM, Fujifilm Irvine Scientific-hence XSFM). Where specified, tissue culture plastic was coated with a fibronectin solution (5 µg/ml fibronectin in PBS-Fujifilm Irvine Scientific, Santa Ana, CA, USA) according to the manufacturer's protocol.
Cells were cultured until passage 3 in the different conditions. At each passage, the cells were counted, and their size measured using the Cell Scepter 2.0 Automated Cell Counter (Millipore), which uses the Coulter principle for particle analysis. The average cell size (expressed as diameter) was recorded and analyzed as a quantitative morphological cell parameter.
In a separate experiment, αMEM/FBS/FGF2 medium was supplemented with the addition of increasing TGF-β1 concentrations (Fitzgerald, using 1 ng/ml for 48 h, 5 ng/ml for the following 48 h, then 10 ng/ml for the final 24 h) or constant 10 ng/ml TGF-β1 for the final 5 days before chondrogenic commitment, in order to improve chondrogenesis of BMSCs.
A schematic of the experimental design is represented in Supplementary Fig. 1 Immunofluorescence. Cryosections were washed for 10 min with dH 2 O to remove the cryocompound, transferred to absolute methanol for 20 min and washed twice in 0.01% Tween 20 in PBS (PBS-T). Enzymatic treatment by 1U/ml Hyaluronidase (Sigma-Aldrich, H3506) and 0.25 U/ml Chondroitinase ABC (Sigma-Aldrich, C2905) in PBS-T for 30 min at 37 °C allowed the digestion of matrix. After washing in PBS-T, pellets were transferred in blocking solution containing 5% horse serum (Vector laboratories, S-2000) in PBS-T for 30 min at room temperature. Primary antibody anti-type II collagen (4 μg/mL, CIICI, see acknowledgement section) and anti-type I collagen (1:400, Origene Acris, R1038) were added over night at 4 °C. Slides were washed with PBS-T, then the secondary antibodies were added, 4 μg/ml anti Rabbit Alexa Fluor 488 against type I collagen (Thermo Fisher, A-11008) and 5 μg/ml anti Mouse Alexa Fluor 660 against type II collagen (Thermo Fisher, A-21055) for 1 h at 37 °C. After washing with PBS-T, the nuclei were counterstained with 2-(4-Amidinophenyl)-1H-indole-6-carboxamidine (DAPI) 2.5 μg/mL and then cover slipped with ProLong mounting solution (Thermo Fisher, P10144).

Statistical analysis.
A total number of 5 donors were used for chondrogenesis experiments, while 4 were used for cell counting/population doubling calculations. Statistical analysis was performed using GraphPad Prism 7.03 software. Non-parametric two-way Analysis of Variance (ANOVA) with Tukey's multiple comparison test were applied, with p < 0.05 considered as statistically significant. A two-way ANOVA was used to evaluate the distributions and homogeneity variance in the groups.

Data availability
All data and materials used in the analysis are available to any researcher for purposes of reproducing or extending the analysis.