Effects of nano-grooved gelatin films on neural induction of human adipose-derived stem cells

The extra cellular matrix (ECM) and cell–cell interactions facilitate the survival, self-renewing and differentiation capabilities of stem cells. Biomaterials with specific structures such as grooves, ridges, pits, or pillars can mimic the topographic landscape of the niche. Cells can “sense” the mechanical properties and surface patterns, ranging from the microto nano-scale, of the substrate; hence, different sizes of nano-grooves on a gelatin surface were designed. The design of grooves can be systematically modified and those structures can reflect the organization of the ECM. In previous studies, polystyrene (PS) was often used because it is easy to fabricate topographic structures. For better biocompatibility, gelatin was chosen and fabricated into ideal nano-groove films. On the other hand, gelatin can be crosslinked by using several crosslinking agents, which leads to a higher mechanical strength and better flexibility. It was known that stem cells can serve as a source of neurons in transplantation therapies. The differentiation of neurons is associated with directionality of stem cell. To investigate the effect of topographic cues on stem cells, groove pattern arrays were constructed onto gelatin surfaces. Human adipose-derived stem cells (hASCs) were seeded onto the patterned gelatin films to observe cell proliferation and differentiation.


Introduction
According to a neurological disorders report from the World Health Organization (WHO) in 2006, 1 the number of people who die from neurological disorders increases year by year. It is estimated that around 9 million people will die from these disorders in 2030. In general, injured nerves can regenerate by themselves. 2 However, surgery is required if the injuries are too severe. Autologous nerve graing is the most common surgery traditionally, 3 but nerves cannot recover fully. Therefore, the aim of neural tissue engineering is to support traditional surgeries, even replace them. Recently, stem cell therapy has drawn a lot of attention. 4 Stem cell therapy uses neural cells differentiated from stem cells to enhance nerve repair. There are many kinds of stem cells that can differentiate into neural cells, but some of them are hard to harvest.
To investigate the behavior of stem cells in human tissues, scientists have constructed substrates with both physical and chemical features. These stimuli can mimic stem cell niches to affect cell proliferation, morphology and differentiation. 5 Stiffness and topographic cues are commonly used to imitate physical features of the ECM. Matrix stiffness has a large effect on cell function. It regulates cell adhesion, growth, survival and motility, and even the cell phenotype. 6 The stiffness of tissues can be measured by the elastic modulus E of a solid, which ranges from 0.5 kPa (fat tissue) to 20 000 kPa (bone tissue).
Comparing these values with a tissue culture on plastic or glass, where the elastic modulus is in the GPa range, it clearly shows that cells are not usually cultured in a non-physiological environment. As a result, many researchers have studied cells in vitro under a more physiological environment. For example, McDaniel et al. revealed that the stiffness of collagen brils can inuence the phenotype of muscle cells. 7 Synthetic substrates with controllable stiffness showed the differences in cell motility and adhesion. 8 Besides, stiffness can further regulate cell lineage commitment and differentiation state, such as the differentiation of precursor cells into osteoblasts, neurites outgrowth and the striation of muscle cells. 9 On the other hand, the effect of topography on cell behavior has been investigated since 1911. 10 Cells can respond to topographic cues as small as 5 nm. Therefore, topographic features from the micro-to the nano-scale have been widely developed. In general, surface topography is affected by roughness and patterns on the surface. 11 Many types of features, for instance, grooves, dots and pillars, are developed, because these structures can be systematically modied and reect the brillar organization of the ECM.
Among these structures, ridges and grooves have been investigated extensively. Micro-or nano-grooves can guide cells to align along the patterns. However, a previous study showed that broblasts do not align with groove depths below 35 nm or ridge widths smaller than 100 nm. 12 Yang et al. investigated the inuence of nano-grooves on the morphology of osteoblast-like cells with a groove-to-ridge ratio of 1 : 1 (90-500 nm in width and 300 nm in depth). It showed an increased cell-spreading area compared to at surfaces and elongated nuclei. 13 Sciancalepore et al. fabricated bronectin (FN) micro-patterns to drive the differentiation of adult renal progenitor cells (ARPCs) in the absence of an exogenous chemical or cellular reprogramming. 14 As mentioned above, there are lots of stem cells that can differentiate into neural cells including embryonic stem cells (ESCs), 15-17 neural stem cells (NSCs) [18][19][20] and adipose-derived stem cells (ASCs). [21][22][23][24] Among these stem cells, ASCs are commonly used because they can be harvested easily and are able to differentiate into multiple lineages including adipocytes, 25 myocytes, 26 osteoblasts, 27 chondrocytes 28 and neural cells. 29 Many studies have investigated how topography affect differentiation of human adipose-derived stem cells (hASCs). Deiwick et al. fabricated a "Lotus" structure using titanium to investigate how it affected the osteogenic differentiation of hASCs. 30 In addition, Mobasseri et al. used polymers to fabricate grooves and studied neural differentiation of hASCs. 31 However, materials used to fabricate topography are too rigid for cells.
In this study, to mimic the physiological environment of normal tissue, nano-grooves were fabricated onto a gelatin substrate, which is a natural and so biomaterial. The mechanical properties of gelatin can be adjusted through crosslinking. Grooves were chosen because they can be fabricated into different sizes easily. To study the relationship between neural differentiation of stem cells and surface topography, hASCs were cultured on different groove sizes on gelatin lms.

Materials and methods
Fabrication of nano-grooved gelatin lms. Four sizes of grooved silicon substrates with a groove-to-ridge ratio of 1 : 1 (groove width/depth (nm): 400/100, 400/400, 800/100 and 800/ 400) were used to fabricate topography onto polydimethylsiloxane (PDMS, Dow Corning, USA) molds and gelatin lms ( Fig. 1). At rst, the silicone rubber molds were fabricated by mixing the elastomer with a curing agent with a ratio of 10 : 1. The mixture was cast onto silicon substrates and degassed to remove entrapped bubbles. Aer polymerizing at 80 C for 2 h, PDMS molds with negative replicas of topography on substrates could be easily removed from the silicon substrates.
To fabricate the nano-grooved gelatin lms, PDMS molds were placed and xed in a Petri dish with the patterned side upwards. An aqueous solution of 5 wt% gelatin (Sigma-Aldrich, Germany) and 1% PSA (Sigma-Aldrich, Germany) was cast into the Petri dish, and air-dried overnight at 25 C. The gelatin samples were subsequently immersed in binary solvent mixtures containing N-(3-dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride (EDC, Sigma-Aldrich, Germany) and N-hydroxysuccinimide (NHS, Sigma-Aldrich, Germany). The reaction was allowed to proceed at 25 C for 96 h. 32 Crosslinked gelatin lms were cut down and thoroughly rinsed with deionized (DI) water to remove excess EDC and the urea byproduct. Before cell culturing, gelatin lms were immersed in cell culture medium for 24 h to ensure the removal of excess crosslinking agent.
Topography measurement. To verify the delity of replication, the topography of silicon substrates was examined using a scanning electron microscope (SEM, NovaTM NanoSEM 23, FEI, USA). On the other hand, PDMS molds and nano-grooved gelatin lms were examined using an atomic force microscope (AFM, Multimode 8, Bruker, USA). Gelatin lms were immersed in DI water until fully swollen before measurement.
Crosslinking extent measurement. A previously reported assay was used to determine the number of uncrosslinked 3amino groups in the crosslinked gelatin. 33 Initially, 11 mg of gelatin was mixed with 1 mL of 4% NaHCO 3 (Sigma-Aldrich, USA) and 1 mL of 0.5% 2,4,6-trinitrobenzenesulfonic acid (TNBS, Sigma-Aldrich, USA), and heated at 40 C for 4 h. 3 mL of 6 N HCl (Honeywell, USA) was added and the mixture was autoclaved for 1 h at 120 C and 15-17 psi. The hydrolysate was diluted with 5 mL of water, and then extracted with ethyl ether.
A 5 mL aliquot of the aqueous phase was removed from each sample and heated for 15 min in a hot water bath. Aer cooling to room temperature, samples were diluted again with 15 mL of water. The absorbance was measured at 346 nm in a UV/Vis spectrophotometer (Cary 100, Agilent, USA) against a blank. Four replicates were used in each determination. Blanks were prepared in triplicate by the same procedure as above, except that HCl was added before TNBS to prohibit the reaction of amino groups with TNBS.
Mechanical test. Before the mechanical test, gelatin lms with different crosslinking times were immersed in DI water until use. Gelatin lms were cut into a round shape with an 8.1 mm diameter. The thickness of lms was measured using a thickness gauge. Round gelatin lms were compressed to half of their thickness by a mechanical test instrument (ElectroForce 3200, TA, USA). The Young's modulus was calculated from the linear region of the stress-stain curve (initial slope: 0-10%).
Swelling test. Gelatin lms were weighed aer being fully airdried. They were then immersed in DI water for different periods of time. Wet samples were wiped with lter paper to remove excess liquid and reweighed. The amount of adsorbed water was calculated as W(%) ¼ 100((W w À W d )/W d ), where W w and W d are the weights of the wet and dry samples.
Contact angle measurement. Nano-grooved and at gelatin lms were immersed in DI water before measurement. At rst, gelatin lm was xed on the glass slide, and immersed in water with the patterned side down. Then, one air bubble was blown onto the patterned surface of the gelatin lm using a microsyringe. Aer snapping clear images of the lm and air bubble, the contact angle was analyzed and calculated using ImageJ soware.
Cell culture conditions. Isolation, cultivation and identication of hASCs from the stromal vascular fraction were performed as described in our previous study. The protocols were approved and maintained by the Research Ethics Committee at National Taiwan University Hospital under the guidelines of the Human Subject Research Acts of Taiwan, R.O.C. Informed consent was obtained from human donors. In this study, hASCs (from National Taiwan University Hospital) were cultured in growth medium (DMEM/F-12 (Hyclone, USA) containing 10% FBS (Biological Industries, USA), 1% PSA and supplementary 1 ng mL À1 basic broblast growth factor (bFGF, Sigma-Aldrich, USA)) in T75 ask at 37 C in an incubator containing 5% CO 2 and saturated humidity. Once the conuence was reached, hASCs were removed from the surface by trypsin-EDTA (Biological Industries, USA), and seeded onto 1 Â 1 cm 2 gelatin lms in the 24-well plate at the concentration of 10 4 cells per cm 2 . To avoid cells dropping from the gelatin surface, PDMS pads were put under the gelatin lms. Aer 3 h, PDMS pads were removed and each well was treated with 0.6 mL medium and the medium was refreshed every 3 days. The cell morphology, viability and proliferation were analyzed.
To induce neural induction, hASCs were cultured in an induction medium (DMEM/HG (Hyclone, USA) containing 1% FBS and supplementary 100 ng mL À1 bFGF) for 7 days. Then cells were cultured in the presence of 10 mM forskolin (Sigma-Aldrich, USA) for another 7 days. The expression of neural differentiation markers was analyzed by immunouorescence and qPCR.
Cytotoxicity/cell viability. To test the cytotoxicity, the medium in each well was removed and rinsed with PBS two times. Aer being xed by 3.7% formaldehyde solution (ACROS Organics, USA) in PBS for 15 min, cells were stained by using the LIVE/DEAD™ Viability/Cytotoxicity Kit (molecular probes, USA).
Samples were immersed in solution with 2 mM calcein AM and 4 mM Ethidium homodimer-1 (EthD-1) in PBS for 30 minutes.
To test cell proliferation of hASCs on gelatin lms, we cultured 5000 cells on nano-grooved and at gelatin lms for 1, 4 and 7 days. CytoScan™ WST-1 Cell Cytotoxicity Assay (G-Biosciences, USA) was used. With a nal volume of 200 mL per well culture medium, blank wells with culture medium only were prepared. 20 mL WST-1/CEC assay dye solution was added to each well and shook gently to mix the chemicals with the medium. Then, the plates were incubated for 2 h in the cell culture incubator. Aer that, the plates were shaken for 1 min on a shaker and the absorbance at 450 nm was measured using a microplate reader. The wavelength of 630 nm was set as a reference.
Cell alignment analysis. At rst, 10Â phase contrast microscope images of hASCs cultured on nano-grooved and at gelatin lms were snapped in 20 random areas. To quantify cell alignment, these images were analyzed using ImageJ soware. At least 30 cells were analyzed in each image. Cells aligned perfectly in the direction of the grooves have an angle of 0 and cells perpendicular to the direction of grooves have an angle of 90 . The full range of angles varies between 0 and 90 . Alignment of cells was determined by measuring the angle between the long axis of cells and the direction of grooves. The experiments were repeated at least three times for each group.
Immunochemistry. Aer being xed by 3.7% formaldehyde solution in PBS for 15 min and permeabilized by 1% Triton X-100 solution (J. T. Baker, USA) in PBS for 10 min at room temperature, samples were blocked by blocking buffer (5% BSA solution in PBST, Sigma-Aldrich, USA) at 37 C for 1 h. Aer blocking, samples were immersed in primary antibodies of neural markers including Tuj-1 (2 mg mL À1 , abcam, USA) and nestin (0.8 mg mL À1 , abcam, USA) in blocking buffer at 4 C overnight. Aer this, secondary antibodies including Alexa Fluor 488 and 594 (molecular probes, USA) in PBST were used to conjugate with primary antibodies at room temperature for 1 h. Finally, the nucleus was stained with a 10 mg mL À1 DAPI solution (Sigma-Aldrich, USA) in DI water for 5 min. Samples were washed extensively by PBST in between each step.
Then, cDNA, F/R primers (Yuanying, Taiwan), Power SYBR Green PCR Master Mix (Thermo Fisher scientic, USA) and DNase/RNase Free Water (Yuanying, Taiwan) were mixed in Eppendorf PCR tubes (8-tube strips). The sequence of primers is shown in Table 1. Aer sealing using clear adhesive lm, a PCR was run using a real-time PCR instrument (StepOnePlus, Applied Biosystems, USA). The expression level was analyzed and normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for each sample. The relative quantity (RQ) of gene expression was calculated using the comparative C T method. The sequence of primers is shown in Table 1.
Statistical analysis. All of the results were expressed as means AE standard error of the mean. The error bar indicated the standard deviation. Comparison between the different groups was analyzed using Student's t-test in Microso Excel 2013. This determined statistically signicant as a p value < 0.05 and statistically highly signicant as a p value < 0.001.

Results and discussion
Topography measurement AFM was used to measure nano-grooves on gelatin lms (Fig. 2). The distance between adjacent peaks was around 800 nm on 400/100 and 400/400, and 1800 nm on 800/100 and 800/400. The vertical distance between the peak and valley was around 150 nm on 400/100 and 800/100, and 200 to 300 nm on 400/400 and 800/400. The groove depth of 400/400 was not as deep as 800/400, which may result from the groove with a 400 nm width being too narrow for gelatin to permeate. Besides, the soness of gelatin and soaring grooves led to a curvature in 2D AFM image of 400/400.

Crosslinking extent measurement
To determine the crosslinking degree of the gelatin lms, TNBS was used to bind with the 3-amino groups in gelatin. It showed yellow and had an absorbance at 346 nm. Less 3-amino groups refers to a higher extent of crosslinking. The result showed that the crosslinking extent of the gelatin lms aer crosslinking by EDC and NHS for 1 to 4 days rose from 42% to 57% (Fig. 3a).
The soness of the gelatin lms aer crosslinking for 4 days was more suitable for experimental operations. Although the crosslinking extent of gelatin crosslinked by EDC and NHS was not as high as glutaraldehyde ($90%), lower cytotoxicity and proper mechanical strength made it a better choice.

Mechanical test
The Young's modulus of crosslinked gelatin lms was used to determine whether the soness of gelatin is close to natural organs. The Young's modulus of gelatin lms aer crosslinking for 1 and 4 days was calculated from the linear region of the stress-stain curve (initial slope: 0-10%). It was found that the Young's modulus rose from 326 to 894 kPa (Fig. 3b), and was similar to human organs. 34 Hence, cells were cultured on so substrates whose stiffness was close to natural organs instead of stiff substrates such as polystyrene and polyurethane.

Swelling test
Crosslinked gelatin lms were bendable and would absorb a lot of water. To investigate when gelatin lms stop swelling, the swelling ratio of gelatin lms was measured. The results showed that weight of gelatin lms crosslinked for 4 days reached 260% of dry weight aer immersing in DI water for 30 min (Fig. 3c), which means gelatin lms had a high water content. On the other hand, it was found that gelatin lms can be dehydratedhydrated repeatedly. In this study, all of the gelatin lms were immersed in DI water or culture medium for at least 30 min until use to prevent topography change.

Contact angle measurement
To investigate the hydrophilicity of nano-grooved gelatin lms, the contact angle was measured (Fig. 3d). As gelatin lms were immersed in culture medium during cell culture, air bubble contact angle was measured rather than water contact angle. In general, gelatin is hydrophilic and has high air bubble contact angle. The results showed that the contact angle of the air bubble on the 400/100 surface was signicantly lower than the other groups, which means the surface of 400/100 was more hydrophobic. A previous study revealed that wettability affects cell adhesion, as cells adhered well on hydrophilic surfaces. 35

Cytotoxicity/cell viability
To test the cytotoxicity of gelatin lms, the live/dead assay was used for hASCs proliferated on nano-grooved and at gelatin lms for 7 days, and TCPS was used as the control (Fig. 4a). The image showed a massive green uorescence, which indicates that most of the cells are alive. The ratios of living cells in each group were higher than 90%. This proved that unreacted crosslinking agent had been removed successfully or has only little toxicity to cells.
In addition to cytotoxicity, cell viability was tested using the WST-1 assay (Fig. 4b). hASCs proliferated on nano-grooved and at gelatin lm for 1, 4 and 7 days were tested, and TCPS was used as the control. The results showed cell amounts increase steadily from day 1 to 7. There is no signicant difference between each group except TCPS, which may result from the difference of surface area between the 24 well and gelatin lms (1 Â 1 cm 2 ).

Cell alignment analysis
To observe the cell morphology of hASCs on nano-grooved and at gelatin lms for 1, 4 and 7 days, phase contrast microscope images were taken (Fig. 5a). On day 1, cells were aligned with the grooves on 400/400 and 800/400, whereas part of cells were aligned with grooves on 400/100 and 800/100 until day 7. It was assumed that groove depth has a greater inuence on cell alignment than width.
To quantify cell alignment, the angle between the cell and groove (Fig. 5b) was measured. As mentioned above, about 80% of cells aligned with grooves within 10 on 400/400 and 800/400 in the beginning. However, 40 to 55% of cells on 400/100 and 800/100 seemed to align with grooves within 10 until day 7, whereas cells on at gelatin lms still remained in random alignment.

hASCs morphology aer neural induction
Compared to cells proliferated on gelatin lms for 14 days, cells differentiated on gelatin lms for 14 days showed contraction and neurites outgrowth (Fig. 6a). At the rst stage of differentiation, cells were still elongating on day 7. Then, the cell body gradually contracted and neurites projected. There was no signicant difference observed for cell morphology between each group from the phase contrast microscope images.

The expression of neural markers
To study whether the cells successfully differentiate into neural cells and the effect of grooves on differentiation, we stained Tuj-1 and nestin for cells proliferated in growth medium and differentiated in the induction medium on 800/400 and at gelatin lms for 14 days (Fig. 6b). Tuj-1 is a neural marker expressed in immature neurons and nestin is an intermediate lament protein expressed during the early stage of neural differentiation. Confocal images showed a higher ratio of neural marker expression aer neural induction, nestin especially. However, there is no difference between 800/400 and at gelatin lms in these images.
To quantify the expression of Tuj-1 and nestin, we ran qPCR for cells differentiated on 800/400 and at gelatin lms for 14 days with Tuj-1 and nestin (Fig. 6c). The results showed that Tuj-1 expression of hASCs on 800/400 was three times higher than TCPS, whereas nestin expression was similar between each group. The increase of the Tuj-1 and nestin expression also correlated with our results from previous studies. 36,37 Li et al. demonstrated adult neural stem cells showed a signicant increase in neuronal differentiation on engineered anisotropic substrates (Si wafer) compared to the control. In their western blot analysis, upregulated Tuj-1 expression was also seen. 36 Béduer et al. demonstrated that pre-coated micropatterned PDMS surfaces can serve as effective neurite guidance surfaces for human NSCs. Immunocytochemistry analysis showed that the channel width can strongly impact development and differentiation. 37

Conclusions
Topography has shown a great inuence on cell behavior in previous studies. Cells were cultured on patterned materials to investigate how topography affects cells. However, most of the materials such as polystyrene (Young's modulus > 3 GPa) are not as so as natural tissues. In this study, nano-grooves on gelatin, whose Young's modulus is close to natural organs, was successfully fabricated. Gelatin lms are believed to benet cell adhesion due to their hydrophilic properties and high water content. In addition, gelatin crosslinked by EDC and NHS showed low cytotoxicity to hASCs. Cells adhered well on gelatin lms and aligned with grooves on 400/400 and 800/400. As for neural differentiation, hASCs contracted and neurites projected aer 14 days of induction and immunochemistry and qPCR results revealed that expression of neural markers (Tuj-1 and nestin) was higher aer neural induction on gelatin substrates, which are more physical and chemically relevant to real tissue, especially the nano-groove patterned gelatin. Engineered anisotropic topographical cues could improve neurite outgrowth and promote neural differentiation.

Conflicts of interest
There are no conicts to declare.