Control of Differentiation of HumanMesenchymal Stem Cells by Altering the Geometry of Nanofibers

Effective differentiation of mesenchymal stem cells (MSCs) is required for clinical applications. To control MSC differentiation, induction media containing different types of soluble factors have been used to date; however, it remains challenging to obtain a uniformly differentiated population of an appropriate quality for clinical application by this approach. We attempted to develop nanofiber scaffolds for effective MSC differentiation by mimicking anisotropy of the extracellular matrix structure, to assess whether differentiation of these cells can be controlled by using geometrically different scaffolds. We evaluated MSC differentiation on aligned and random nanofibers, fabricated by electrospinning. We found that induction of MSCs into adipocytes was markedly more inhibited on random nanofibers than on aligned nanofibers. In addition, adipoinduction on aligned nanofibers was also inhibited in the presence of mixed adipoinduction and osteoinduction medium, although osteoinduction was not affected by a change in scaffold geometry. Thus, we have achieved localized control over the direction of differentiation through changes in the alignment of the scaffold even in the presence of a mixed medium. These findings indicate that precise control of MSC differentiation can be attained by using scaffolds with different geometry, rather than by the conventional use of soluble factors in the medium.


Introduction
Mesenchymal stem cells (MSCs) are multipotent, allowing differentiation into various types of cells, such as adipocytes, osteoblasts, and chondrocytes [1].This ability is expected to be useful in many clinical applications, including the regeneration of damaged and injured tissues, and in cosmetic surgery [2][3][4][5][6].At present, MSCs are directly administrated into affected tissues, but cells differentiated from MSCs have not yet been attempted in clinical applications, because of the difficulties posed by the precise control of induction.However, if such differentiated cells were available for clinical applications, the time to engraftment of cells onto the affected tissues could be reduced.To date, to control MSC differentiation, an induction medium, containing several types of soluble factors, has been employed and has efficiently induced differentiation [1].However, obtaining a uniformly differentiated population of cells, of a quality appropriate for clinical applications, by using only soluble factors, has been challenging.In addition, residual soluble factors also pose safety issues [7].
Recently, McBeath et al. reported that MSCs attached to a wide flat surface differentiated into osteoblasts, while cells attached to a small surface differentiated into adipocytes [8].They also reported that cell shape is a key factor in cell differentiation, because the intracellular tension triggers changes in expression of GTPases, such as RhoA and its downstream Rho kinases, which results in changes in cellular behaviors.
This knowledge can also be applied to the 3-dimensional (3D) culture of MSCs, not only to cultures grown on a flat surface.However, cells cultured in a dish show quite different morphologies to cells grown in a 3D culture.Cells in a living tissue are surrounded by an extracellular matrix (ECM), which has a fibrous network structure; such cells take on elongated shapes in a particular direction, in accordance with the fibrous structure of the ECM.In order to provide a scaffold material for use in 3D culture, hydrogels have typically been prepared from synthetic and natural polymers [9,10].Although hydrogels are easy to prepare and handle, they are structurally uniform, and not anisotropic.For this reason, cells do not keep elongating in hydrogels, as they do in living tissues.
In this study, in order to address this problem, and to control cell differentiation in a 3D environment using biomaterials that mimic living tissues, we have attempted to develop a scaffold for effective differentiation, using synthetic nanofibers.We prepared nanofiber scaffolds with a high level of anisotropy, which mimics the structure of the ECM and nanometer-sized elastic fibers, such as collagen, and compared MSC differentiation on these media.We used an electrospinning method to prepare these nanofiber scaffolds, because this approach allows easy control of the diameter, density, and geometric structure of the fiber [11][12][13].We also investigated the effect of scaffold geometry, and the alignment of nanofibers, on cell differentiation.We hypothesized that cells elongated along with aligned nanofibers receive higher differentiation signal via cytoskeletal tension than cells on random nanofibers.

Experimental
2.1.Fabrication of Nanofiber Scaffold.Polyurethane (PU; P22SRNAT, JIS hardness 82A; Nippon Polyurethane Industry, Tokyo, Japan) was purified by methanol re-precipitation from a tetrahydrofuran solution (THF; Wako Pure Chemical Industries, Osaka, Japan), and dried under reduced pressure prior to use.PU was dissolved in 95% THF and 5% N,Ndimethylformamide (DMF; Wako) at 12.5% (w/v).PU nanofibers were electrospun onto a glass coverslip (No. 5; 22 × 26 × 0.45 − 0.60 mm; Matsunami, Osaka, Japan) using an electrospinning setup (MECC, Fukuoka, Japan) as detailed in Scheme 1.The electrospinning unit consisted of a highvoltage power supply, a rotating collector, a syringe pump, and a 23 G-needle that was connected to a syringe by a tube.A high voltage, 25 kV, was applied between the syringe needle and the ground electrode.A rotating collector (diameter: 10 cm) was placed horizontally, 10 cm from the needle.The PU solution was extruded from the syringe at a rate of 0.6 mL/h.The speed and duration of rotation were 1,500 rpm and 3 min, respectively, for the aligned nanofiber scaffold, and 0 rpm (static) and 30 s, respectively, for the random nanofiber scaffold.To improve the surface hydrophilicity of the fibers, oxygen plasma treatment (100 W, 30 s, 5 Pa, chamber size: ϕ 64 mm × 160 mm) was carried out using a plasma reactor (PR300; Yamato Scientific, Tokyo, Japan).

Scanning Electron Microscopy.
The fabricated nanofiber scaffold was sputter-coated with Pt/Pd using an ion sputter (E-1030; Hitachi, Tokyo, Japan).Scanning electron microscopy (SEM) observation was carried out with an S-2600HS microscope (Hitachi).ImageJ software (ver.1.46a).Fiber orientation was quantified using the second-order parameter S as follows:

Measurement of Fiber
where θ is an angle of each fiber and cos 2 θ is the average of cos 2 θ.

Histochemical Analysis.
The presence of mineralized deposits after osteogenic differentiation was determined with von Kossa staining [15].In a brief, cultures were rinsed twice with PBS, then fixed with 10% formaldehyde for 30 min and thereafter washed 3 times with distilled water.One milliliter of 5% (w/v) silver nitrate (Nacalai Tesque, Kyoto, Japan) was added to each well.The cultures were placed in the dark for 30 min, washed with water, and then developed with 5% (w/v) sodium carbonate in 25% formaldehyde for 5 min.After development, cultures were washed, 5% sodium thiosulfate added to stop development, again washed with water, and then air-dried and imaged.The amount of calcium deposits was evaluated from image analyses using ImageJ software.For the detection of alkaline phosphatase activity, AP-Substrate Kit III (SK-5300, Vector laboratories, CA, USA) was used following the manufacturer's instructions.For the detection of adipogenic induction, cells were fixed in 4% paraformaldehyde and incubated with Oil-Red-O (Wako) to stain lipid vacuoles [15].[14][15][16].Amplified products (9 μL) were electrophoresed in 2% SeaKem GTG agarose gels containing 0.5 μg/mL ethidium bromide, at 100 V for 30 min, and the PCR products visualized on an ultraviolet illuminator.The optimal PCR cycle conditions were adjusted according to the brightness of the visualized band.

RT-PCR. Gene expression in
2.8.DNA Microarray.Gene expression in MSCs cultured in the maintaining medium for 3 days on aligned and random nanofiber scaffolds were compared by DNA microarray.In brief, the total RNA extracted from MSCs as described above was quantified using an Agilent 2100 Bioanalyzer (Agilent Technologies).cDNA was generated from 100 ng of the total RNA using Ambion's Whole Transcripts Expression Kit (Ambion); these were fragmented and end-labeled using a GeneChip WT Terminal Labeling Kit (Affymetrix), according to the manufacturer's instructions.An aliquot of the labeled DNA (5.5 μg) was hybridized to an Affymetrix Gene-Chip Human Gene 1.0 ST array (genome-wide expression profiling chip, 28,869 genes with 764,885 probes; Affymetrix) at 45 • C for 16 h.After hybridization, the array was stained with Fluidics 450 station and scanned using a GCS3000 Scanner (Affymetrix).Scanned CEL files were processed with Affymetrix Expression Console software to obtain RMA signals.Differentially expressed genes with a fold-change greater than ±1. 5 were identified with a Subio Platform (Subio Inc., Tokyo, Japan).

Results and Discussion
3.1.Fabrication of Nanofiber Scaffolds.Nanofiber scaffolds for cell culture were fabricated by electrospinning from 12.5% PU solution using either a rotating or a static collector (Scheme 1).The diameters of fibers were measured from the SEM images, such as those shown in Figures 1(a) and 1(b).There was no difference in the diameters of nanofibers collected on a rotating or on a static collector (mean 1.08±SD 0.36 μm and 1.57 ± 0.36 μm, resp., Figure 1(c)).The orientation of the fibers was quantified by calculating a secondorder parameter, S, from the fiber angles that were measured in the SEM images; a higher value (approaching 1) of S implies a higher orientation along a specific direction, while a lower value (approaching 0) implies random orientation.As shown in Figure 1(d), fibers fabricated on a rotating collector demonstrated higher alignment (S = 0.74 ± 0.03) than did fibers fabricated on a static collector (S = 0.31±0.04).Therefore, fibers fabricated on a rotating collector and on a static collector were designated as "aligned" and "random" fibers, respectively.

Cell Differentiation on a Nanofiber Scaffold.
To examine the effect of scaffold geometry on MSC differentiation, MSCs were seeded on the aligned and random nanofiber scaffolds and induced to differentiate into osteoblasts or adipocytes by using appropriate differentiation medium containing specific soluble factors.To minimize the local fluctuation of soluble factors in the culture medium, in order to evaluate only the effect of the scaffold geometry, 2 types of nanofiber scaffolds were placed on the same dish and cells on each scaffold were cultured in the way, sharing the same medium.After a 14-day osteoinduction with OIM, von Kossa staining was performed to evaluate the deposition of calcium.Some black spots of calcium deposition were observed, as shown in Figure 2(A), but there was no significant difference in calcium deposition between the 2 types of scaffolds.However, after adipoinduction using AIM, cells on random fibers appeared to be less differentiated than cells grown on aligned nanofibers, despite having been cultured in the presence of the same soluble adipogenic factors (Figures 2(A) and 2(C)).
To examine the differentiation of MSCs at a genetic level, RT-PCR was also performed.Osteoinduced MSCs cultured on the different types of nanofibers showed no significant difference in terms of the expression of osteogenic markers from cells that had been cultured on a flat dish, but adipoinduced MSCs cultured on these various substrates showed different expression levels of adipogenic markers (Figure 3).Specifically, the expression of adiponectin (ADIPOQ), [19,20] a hormone secreted specifically from adipose tissue, and lipoprotein lipase (LPL), 1 of the 5 major groups of lipoproteins, [21] were inhibited in cells grown on nanofiber scaffolds compared to those grown on the flat dish.Moreover, peroxisome proliferator-activated receptor gamma (PPARG), which regulates fatty acid storage and glucose metabolism, [19,20] and CCAAT/enhancer-binding protein alpha (CEBPA), which transactivates the promotors of several adipocyte markers, [22] were inhibited more strongly on a random than on an aligned scaffold, which was consistent with histochemical result mentioned above.

Geometric Control of the Differentiation of MSCs.
To investigate whether MSC differentiation could be controlled by the geometry of the scaffold, we induced differentiation using a mixture of OIM and AIM, which can induce cells into both adipocytes and osteoblasts.As in the previous experiment, we placed the 2 types of scaffolds into the same dish, in order that the cells on either scaffold would share the same medium.As a result of the mixed medium, the MSCs cultured on a flat dish showed differentiation into both adipocytes and osteoblasts (Figure 4).On the other hand, MSCs cultured on the different nanofiber scaffolds showed less differentiation into adipocytes but showed no difference in terms of the percentage of cells that differentiated into osteoblasts on this scaffold as compared to the percentage of osteoblasts produced on the flat dish surface in the mixed medium.More particularly, MSCs grown on random nanofibers showed mainly osteoinduction, and markedly less adipoinduction, even though the medium also contained soluble factors that promote adipoinduction.This indicated that direct control of the commitment of MSCs to differentiate into a particular cell type had been achieved through the geometry of the scaffold.

DNA Microarray Analysis.
To evaluate the effect of scaffold geometry on MSC differentiation at a genetic level, we compared gene expression in MSCs that had been cultured on the aligned and random nanofiber scaffolds in maintaining medium for 3 days, using a DNA microarray that can assess the expression of 28,869 genes.The result of the DNA microarray is summarized in Figure 5.In this figure, we listed the genes of which expression levels were 1.5-fold different between the MSCs cultured on the aligned and random nanofibers.Specifically, we identified 27 genes that showed a higher expression in MSCs cultured on aligned fibers, and 20 genes that showed higher expression in MSCs cultured on random fibers.
In particular, we found differential expression of secretable soluble proteins through the geometry of the scaffold.One of these proteins was interleukin 21 (IL-21), a cytokine which plays a role in the proliferation and maturation of natural killer (NK) cells.[23] Serpin peptidase inhibitor, which helps to control cell function by blocking the activity of peptidases, [24,25] was also enhanced in cells grown on aligned nanofibers.On the other hand, keratinocyte growth factor-like protein, a type of heparin-binding growth factor previously known as FGF-7, [26] was enhanced in cells grown on random fibers.This protein is reported to play a role in differentiation of epidermal stem/progenitors into keratinocytes, as well as in DNA synthesis, and cell proliferation.[27][28][29] Thus, MSC differentiation may be affected by the balance of these soluble factors.
We also found differential expression of transcriptional factors.The expression of POU class 5 homeobox 1B, also known as Oct-3/4, which is a key transcriptional factor in the maintenance of stem cell [30,31], was enhanced on aligned fibers.We also found that the expressions of some subfamilies of zinc-finger proteins were altered; specifically, those with the C2H2 DNA-binding motif characteristic of transcriptional factors, which are functionally important regulators of development [32].It is possible that changes in these transcriptional factors trigger an intracellular cascade resulting in differentiation.However, we did not find any significant changes in cytoskeletal factors, as had previously been reported.

Conclusion
In this study, we demonstrated that differentiation of MSCs is affected by the geometry of a nanofiber scaffold.Induction of differentiation of MSCs into adipocytes was markedly more inhibited on random than on aligned nanofibers.We also showed that specific control of differentiation is possible through manipulating geometry in the presence of a mixed medium that facilitates both adipoinduction and osteoinduction.Although osteoinduction in the mixed medium was almost equal between the geometrically different scaffolds,  adipoinduction was markedly inhibited on random nanofibers.We further examined gene expression in these cells by DNA microarray and found that several types of transcriptional factors and soluble factors were affected by the geometry of the nanofiber scaffold.Thus, commitment of MSCs to differentiation into a particular cell lineage can be controlled by scaffold geometry, and not only by using particular soluble factors in the medium.This approach is expected to be useful as it may allow precise and localized control of MSC differentiation by manufactured materials used in tissue engineering.

Figure 5 :
Figure5: Results of DNA microarray.Gene expression levels were shown in a heat map.Genes listed in a upper part were expressed higher in aligned MSC, while genes listed in a lower part were expressed higher in random MSCs.