2.1 Cell Culture

Human MSCs were provided from PromoCell (Heidelberg, Germany) and maintained in low-glucose Dulbecco's Modified Eagle's Medium (DMEM with 1 g/L glucose) containing 10% fetal bovine serum (FBS), 100 units/ml penicillin, and 100 μg/ml streptomycin [2,3]. The cells were incubated at 37°C and 5% fully humidified co². Cells in passage 6 were used for all experiments in this study.

2.2 Substrate preparation

Substrates for cell culture were fabricated on two different materials, polydimethylsiloxane (PDMS) and SU-8. For patterning structures on PDMS substrates, PDMS were cast on a master mold which was photo-lithographically fabricated in SU-8 on a silicon wafer, or master patterns in etched silicon surfaces after e-beam lithography. Micropatterned SU-8 substrates were fabricated on tissue culture plates (TCP) by soft lithography18 using a PDMS replica mold. In detail, PDMS replica molds were placed on SU-8-coated TCP surfaces and exposed to UV-rays for 10 min., and the PDMS molds were removed and cured for 20 min. For nanoscale patterns, since soft lithography using PDMS molds cannot guarantee accurate features, a hard polyurethane acrylated polymer mold coated with an anti-adhesion PDMS layer19 was used for the nanopattern transfer process. The rest of the process is the same as the microscale SU-8 molding described above. All the substrates were prepared for cell culture by coating with 10 μg/ml fibronectin for 2 h in a 5 % co² cell culture incubator. PDMS substrates were treated with oxygen plasma before the coating.

2.3 Cell staining

After the cells were cultured on various substrates, they were fixed in 3.7 % formaldehyde for 1 h, rinsed in phosphate buffered saline (PBS), and permeabilized by treating with 0.1 % tritonX-100 for 15 min. Then, actin cytoskeletons were stained with rhodamine-conjugated phalloidin, and the nuclei were stained with 4'-6-diamidino-2- phynylindole (DAPI).

3.1 Various substrates for cell culture

To compare the cellular shapes and the organization of the actin cytoskeletons in MSCs on different elastic matrices with the same surface topography, we selected two polymers: polydimethylsiloxane (PDMS) as a soft polymer and SU8 as a rigid polymer. PDMS is a silicon elastomer whose Young’s modulus is around 1 MPa, and SU8 is a highly rigid epoxy-based polymer whose Young’s modulus is 2~5 GPa. The surface topographies were created by replicating micro- or nano-patterns from a master mold on each polymer by soft lithography or nanopattern transfer technique [19], which are described the materials and methods section. Figure 1 show all the patterned PDMS and SU8 substrates used as matrices for culture of MSCs throughout this study. The dimensions of the groove and ridge patterns are 0.5 × 0.5 × 0.5 μm (Figures 1a and 1e), 1 × 1 × 1 μm (Figures 1b, and f), 20 × 20 × 10 μm (Figures 1c and 1g), and 50 × 50 × 10 μm (Figures 1d, and 1h) (groove width × spacing width × depth). As shown in Figure 1, micron-scale patterns made of both PDMS and SU8 presented high accuracy in their structures (Figures 1 b, c, f, and g), but the nanoscale pattern (0.5 μm groove width × 0.5 μm spacing width × 0.5 μm depth) in PDMS (Figure 1e) displayed less structural accuracy than the same pattern in SU8. Thus, in the following experiments, the effects of PDMS and SU8 patterns on MSCs were compared to one another on only the micron-scale patterns. Results from the culture on nanopatterned surfaces were considered as reference data.

Figure 1: Various substrates for culture of MSCs. Various-dimensional groove and ridge patterns fabricated in SU-8 shown in (a)-(d), patterns fabricated in PDMS are shown in (e)-(h). (a) and (e) are the AFM (atomic force microscope) images of the substrates molded from a master with 500 nm × 500 nm × 500 nm (groove width × spacing width × depth) groove and ridge patterns. (b) and (f) are scanning electron microscopic images of the images1 μm × 1 μm × 1 μm. The others are microscopic images of 20 μm × 20 μm × 10 μm ((c) and (g)) and 50 μm × 50 μm × 10 μm ((d) and (h)) groove and ridge patterns.

3.2 Morphological change of MSCs grown on various substrates

To evaluate the morphological changes of MSCs on the micro and nanotopography made of two different polymers, cells were cultured on each substrate, and F-actin was stained with rhodamineconjugated phalloidin to observe the boundary shape of the cells using microscopy (Figure 2). As shown in Figures 2(a-c), the shape of MSCs were not significantly changed in the non-patterned PDMS and SU8 substrates, compared to MSCs on tissue culture plates. The cells were stretched well on all the substrates, but their orientations were irregular. In a previous report, it was suggested that MSCs showed significant changes in their morphology within a few hours when they were grown on different elastic matrices with flat surfaces11. In our study, the elasticity of the materials was between about 1 MPs and 2 GPa, whereas the stiffness of the matrices used in the previous paper was between 0.1 KPa and 40 KPa. We assumed that the cells could not sense the difference in stiffness in this range, and this might also be a reason for why changes in cell shape were not clearly observed on the different flat substrates in our results. To elucidate the correct reason for this phenomenon, further studies are needed for the identification of the threshold values of the stiffness ranges sensed by MSCs on different elastic matrices.

Figure 2: Fluorescent microscopic images presenting shapes of the MSCs on various substrates. (a)-(c) MSCs on flat substrates; (a) tissue culture plate, (b) SU8, and (c) PDMS. (d)-(k) MSCs on groove patterned substrates; (d) and (e) 500 nm, (f) and (g) 1 μm, (h) and (i) 20 μm, and (j) and (k) 50 μm. Images, on the left panel, (d), (f), (h), and (j), show cells on SU8 and images on the right side, (e), (g), (i) and (k) are cells on PDMS. The scale bars represent 20 μm and the direction of the groove patterns are represented by double-arrow headed lines above the scale bars.

When MSCs were cultured on the patterned substrates with the dimensions of 0.5, 1, 20, and 50 μm (groove and spacing width, Figures 2(d-k), the cells oriented along the direction of patterns such that the longest axes of the cells were aligned in parallel to the grooves and ridges. In addition, the cells on the PDMS groove and ridge patterns were elongated more than those on SU8 substrates with the same pattern of structures. On the PDMS substrate with 1-μm groove patterns, it was commonly observed that cells were extremely elongated more than 1 mm. In addition, some parts of the cells on that substrate showed morphological similarity to neuronal cells composed of a single cell body and highly branched neuritis, while this was not observed on the SU8 1-μm groove patterns. This result indicates that cells can sense the surface topography and matrix elasticity simultaneously. Furthermore, it is very interesting that the differences in the cell elongation and shape change on the PDMS and the SU8 substrates were observed only on the pattered substrates (especially on the 1-μm groove patterns), while no difference was observed on flat surfaces. Through this result, we propose that there are some interactions between sensing the matrix elasticity and sensing the surface topography in the cells.

3.3 Actin-Alignment of MSCs grown on various substrates

To corroborate the results (Figure 2), high-magnification images (×800) of actin cytoskeletons were taken using a confocal laser scanning microscope (CLSM). Figure 3 shows some representative images of the cells cultured on the same substrates previously used for fluorescent microscopic imaging. Actin cytoskeletons were stained with rhodamine-conjugated phalloidin, and the nuclei were countstained with DAPI. In the cells on the SU8 substrates with 0.5-μm groove patterns, actins were strongly expressed only at the edge of the cells (Figure 3a). On the SU8 substrates with the 1-μm groove patterns, actin fibers aligned in parallel to the patterns were clearly observed in both the middle and the edge of the cells (Figure 3b). In the case of the 20-μm groove made of SU-8, even though the cell bodies were aligned along the patterns, most of the actin fibers inside the cells were biased by about 60 degrees (Figure 3c). However, actins on the 50-μm patterns were not organized in a specific orientation; these actins were biased and formed meshwork. On the patterned PDMS substrates, strong actin fibers were not observed as much as on the patterned SU8 surfaces (Figures 3e-g). It was observed that the filament-like actins were aligned along the pattern, and the actins formed meshwork on all the patterned PDMS substrates.

Figure 3: Confocal laser microscopic images of actin cytoskeletons of MSCs on different substrates. Actin cytoskeleton of the cells on (a)-(d) SU8 substrates and (e)-(f) PDMS substrates. Dimensions of groove patterns in each substrate are following; (a) 500 nm, (b) and (e) 1 μm, (c) and (f) 20 μm, and (d) and (g) 50 μm. The scale bars represent 20 μm and the direction of the groove patterns are represented by double arrow headed lines beside of the cells.

Subsequently, different cell shape parameters, such as the degree of elongation, distinct form, and alignment of actin cytoskeleton, were observed more significantly on substrates with different stiffnesses compared to flat substrates. Unfortunately, sufficient data were not accumulated to explain these phenomena. Further investigation is necessary to understand the mechanism of the substrate stiffness effect on the dynamic rearrangement of cytoskeletons in MSCs.