Estimation of the Angles Migrating Cells Turn on Three-Dimensional Micro-Patterned Scaffolds by Live Cell Imaging with an Inverted Microscope

To determine how the three-dimensional (3D) shapes of scaffolds influence cell migration, 3D micro-patterned scaffolds with various shapes were fabricated on a silicon substrate (725 μm thick, 10 mm×10 mm quadrate) by using photolithography. We imaged living cells on a silicon substrate over 72 h using a novel simple method. NIH3T3 cells were adhered to the silicon substrate, which was then placed face-down or face-up into culture medium in a 35 mm (12φ) glass-bottomed dish. In this method, there is a sufficient gap (1.3 mm) between the downward-facing cells and the bottom of the plate for the culture medium to diffuse over the cells. Cell growth over 72 h was similar in both conditions. NIH-3T3 cells were adhered to three kinds of 3D micro-patterned scaffolds, placed face-down into culture medium in glass-bottomed dishes, and cell migration and the scaffolds were observed over 72 h. The three scaffolds differed only in terms of the unit shape of the repetitive pattern, namely a scale structure with equilateral triangular pores, a check structure with regular tetragonal pores, or a stripe structure with rectangular grooves. These scaffolds had a constant pore ratio (50%), pore depth (22 μm), and subcellular pattern size. The angle at which cells turned correlated with the unit shape of the scaffold: the interior turning angles were multiples of 60◦ on a scale structure with equilateral triangular pores, multiples of 45◦ on a check structure with regular tetragonal pores, and close to 0◦ or 180◦ on a stripe structure with rectangular grooves. Therefore, the angle that cells turn is influenced by the unit shape of the 3D patterned scaffold on which they are cultured. Furthermore, when the angles at which the migrating cells turned were investigated in detail, it was found that a cell turns in one of two directions that correlate with the unit shape of the scaffold; one corresponding to the edge of the pattern, and the other corresponding to the upper surface of the pattern. These differences in the angles that migrating cells turned correlated with differences in the angles they extended protrusions. The angles of cell protrusions markedly differed between the three different scaffolds, which partly underlies why migrating cells turned at different angles. In summary, the unit shape of the micro-patterned scaffold affects the angle at which cells extend, which in turn affects the angle at which migrating cells turn. [DOI: 10.1380/ejssnt.2014.289]


I. INTRODUCTION
Cells adhere to three-dimensional (3D) extracellular matrices in vivo.Various 3D scaffolds, ranging in size from the nanometer [1,2] to micrometer scale [3][4][5], have been used to reproduce such cell adhesion in vitro [6], and many studies report that the topologies of these scaffolds affect cell functions.
The effect of the 3D shape of the scaffold on cell functions involves many unidentified factors, and it is not known whether 3D and 2D scaffolds have different effects of cell functions.
Although the responses of cells grown on matrices with various 3D shapes have been investigated, a clear explanation of why cell functions differ when cells are cultured on different 3D micro-patterned scaffolds has not been reported.Furthermore, the 3D micro-patterned scaffolds used to date vary widely in terms of pattern shape, chemical composition, electric charge, and viscoelasticity.To determine the behavior of cells cultured on patterned scaffolds, the effect of changing one property of the scaffold at a time, such as the shape of the pattern, must be investigated.
Cell migration is important for tissue formation, maintenance of homeostasis, and immunity in vivo, and it is assumed to be the function that is most affected by the 3D cell matrix shape.Therefore, in this study, we focused on the unit of a repetitive pattern of a 3D scaffold and investigated whether the shape of this unit influences cell migration.To this end, we produced three 3D micro-patterned scaffolds on a silicon substrate that differed only in terms of the unit shape of the repetitive pattern, namely a scale structure with equilateral triangular pores, a check structure with regular tetragonal pores, and a stripe structure with rectangular grooves.The unit shape of each patterned scaffold was of a subcellular size.Changes in the pore ratios of scaffolds may affect cell functions since cell functions are markedly changed when cells are cultured on two-dimensional scaffolds [24][25][26][27].Therefore, the 3D micro-patterned scaffolds needed to have a constant pore ratio.Since patterns fabricated on a single crystal silicon substrate are not elastically deformed, elasticity did not vary between the different scaffolds.This meant that the influence of the scaffold pattern, and no other factors, on cell migration could be investigated.
It is difficult to observe an opaque scaffold, such as a silicon substrate, with an inverted microscope, which is generally used for live cell imaging, because the side of the scaffold to which cells are adhered must face-down.However, to prevent contamination and evaporation of the culture medium over an extended period of time, an inverted microscope is recommended because cells can be observed in a culture dish with the lid closed.Moreover, to observe living cells for an extended period of time, there must be a sufficient gap between the downward-facing cells and the bottom of the plate to allow the culture medium to diffuse over the cells.Therefore, we also investigated how the substrate could be placed face-down into the culture medium.

A. Fabrication of 3D micro-patterned scaffolds by photolithography
The 3D micro-patterned scaffolds were fabricated on a silicon substrate (725 µm thick, 10 mm×10 mm quadrate) by using photolithography.Deep etching of the silicon substrate was performed using the Bosch process for deep reactive ion etching.Patterns were fabricated in various regions (1000 µm×1000 µm), and each patterned area was located 500 µm from the adjacent patterned areas.A scale structure of equilateral triangles with 10 µm long sides, a check structure of regular tetragons with 10 µm long sides, or a stripe structure of 6 µm wide stripes that were 6 µm apart were drawn onto the photomask in each regions.The generation of 3D micro-patterned scaffolds on the substrate was confirmed by field emission scanning electron microscopy (JSM-6700FT, Jeol Ltd., Japan) and color 3D laser scanning microscopy (VK-9700, Keyence Co., Japan).The cutting planes of the patterns were observed by FE-SEM, and the depths of five points of each pattern were measured and the mean depth was calculated.

B. Determination of the pore ratios of the 3D micro-patterned scaffolds
Cell migration was observed on five patterns of each shape.3D images of each pattern were obtained using a 150×0.95NA objective (Nikon Co., Japan) and a color laser 3D microscope VK-9700 (Keyence Co., Japan).These images were analysed using Metamorph Software ver.7.7 (Molecular Devices, Inc., USA) to determine the total pore area per unit area (the pore ratio) of each pattern.Shape analysis of the patterns was performed in an area of 71.0 µm×94.6 µm.The pore ratios as determined by shape analysis were compared with the theoretical pore values, which were calculated on the basis of blueprints of the patterns.

C. Hydrophilic treatment, sterilization, and degassing of 3D micro-patterned scaffolds
The surfaces of the 3D micro-patterned scaffolds were treated with oxygen plasma (oxygen cleaning; O 2 , 50 sccm, 3 Pa, 100 W, 8 min) using reactive ion etching (RIE-10NRV, Samco, Inc., Japan), which greatly enhanced the hydrophilicity of the surfaces.The contact angle of a water droplet (5 µl) on a flat silicon substrate was 72.6± 1.9 • (mean ±S.D., n = 26) prior to oxygen cleaning and 11.2±3.2• (mean ±S.D., n = 22) after oxygen cleaning.This value is lower than the contact angle (approximately 20 • ) of a commercial cover glass used for cell culture that has been ultrasonically cleaned with pure water; therefore, the oxygen-cleaned silicon substrates were more hydrophilic than a cover glass.The contact angle on the 3D micro-patterned scaffolds could not be measured because the pattern area was too small to apply a water droplet.
The 3D micro-patterned scaffolds were sterilized at 120 • C for 20 min, immersed in a 35 mm cell culture dish containing culture medium (DMEM containing 10% foetal bovine serum (FBS) and 1% penicillin-streptomycin), and degassed with gentle agitation at −0.08 to −0.09 MPa for 1 min.Light microscopy confirmed that air bubbles were not present in the reflected images of the 3D micropatterned scaffolds.

E. Cell growth on downward-and upward-facing substrates
Two flat silicon substrates were prepared that were the same size as the 3D micro-patterned scaffolds.These scaffolds were subjected to 50% hydrofluoric acid cleaning and hydrophilic treatment and were degassed as described for the 3D micro-patterned scaffolds.The culture medium was replaced with fresh culture medium and the flat substrates were incubated for 12 h at 37 • C in 5% CO 2 /95% air.NIH-3T3 cells were cultured in the same conditions for 76 h and were then seeded onto the sterile flat substrates at a density of 1×10 4 cells/cm 2 .After seeding, cells were cultured for 4 h in the same conditions and then one scaffold was placed face-down into culture medium in a 35 mm (12ϕ) glass-bottomed dish.The four corners of the scaffold were placed onto the frame of the glass base with sufficient space to allow the culture medium to diffuse between the downward-facing cells and the bottom of the plate.The other scaffold was placed face-up into a 35 mm cell culture dish containing culture medium.The scaffolds were then incubated for 76 h in the same conditions and were fixed with PBS containing 4% paraformaldehyde.The fixed cells were washed three times with PBS and three times with PBS containing 0.01% Tween 20, and were then stained with phalloidin-Alexa 488 (1:250, 6 h) and DAPI (1:2000, 20 min) to label F-actin and nuclei, respectively.The cells were then washed three times with PBS containing 0.01% Tween 20, and micro cover glasses were applied with EUKITT mounting medium.Fluorescence images of the cells and reflection images of the scaffolds were simultaneously captured in 18 fields of 871 µm×690 µm using a 10×0.45NA objective (Zeiss, Germany) and a CCD camera.
The number of DAPI-labelled cells within each defined field was counted manually, and the number of DAPIlabelled cells per unit area was designated the 'cell density'.The mean and standard deviation of the cell densities in the 18 defined fields of each scaffold were calculated.As the same number of cells was seeded onto both scaffolds, the cell density is representative of the proliferation rate.As a reference, cells were also seeded onto a haemocytometer cover glass and placed face-up into a culture dish.This cover glass was chosen as it has a similar thickness to the silicon substrate.

F. Staining of living cells with PKH26
Cells were seeded onto scaffolds as described above, cultured for 2 h in culture medium at 37 • C in 5% CO 2 /95% air, and then immersed in PKH26 (1:500) for 4 min.The PKH26 reaction was stopped by the addition of FBS and the scaffold was washed with culture medium.The staining and washing procedures took approximately 15 min and were based on the manufacturer's recommendations.
The scaffold was placed face-down into culture medium in a 35 mm (12ϕ) glass-bottomed dish as described above.Cells were incubated for approximately 1.75 h in the aforementioned culture conditions, and cell proliferation and migration were then analyzed.

G. Time-lapse observation using an inverted microscope and analysis of cell migration
Cells were seeded onto the various 3D micro-patterned scaffolds, which were then placed face-down into culture dishes and cultured as described above.At 4 h after seeding, time-lapse microscopy was commenced for 72 h and images were acquired every 30 min with an inverted fluorescence microscope (Observer Z1, Zeiss, Germany) and a Stage Top Incubator INU-ZILCS (Tokai hit, Japan).PKH26-labelled cells were observed on flat and patterned surfaces.Fluorescence images of the cells and reflection images of the substrate in fields of 871 µm×690 µm were obtained using a 10×0.45NA objective (Zeiss, Germany) and a CCD camera.Scaffolds with five patterns of each shape and five flat surfaces were imaged.
Individual cell movements were manually tracked using image analysis software (AxioVision4.8,Zeiss, Germany) and the coordinates of each cell were measured at each time-point.To exclude the influence of cell-cell interactions, cell migration was analysed over the first 24 h of the time-course, during which cell density was comparatively low.On each scaffold and flat surface, 11-20 cells were tracked.The trajectory of each cell was graphed based on its coordinates.On each surface, 172 interior angles of the turns made by the tracked migrating cells were manually measured using image analysis software, and were displayed in histograms.To prevent deformation of a cell being analyzed as a migration, these interior angles were only measured when a cell changed direction and then migrated for 30 µm or more, which is approximately the length of an extended NIH-3T3 cell.Each of the histograms had 17 unit intervals of 0 • -170 • in 10 • bins.Interior angles of the turns of more than 170 • were not evaluated because it was difficult to distinguish between cells that were turning and those that were migrating in a straight line.

H. Observation of the morphology of cells extending protrusions and the angles of these protrusions on 3D micro-patterned scaffolds using confocal laser scanning microscopy
Following time-lapse microscopy, NIH-3T3 cells cultured on the 3D micro-patterned scaffolds were fixed with PBS containing 4% paraformaldehyde.Micro cover glasses were applied using EUKITT mounting medium.
The 3D morphologies of the cells were observed using confocal laser scanning microscopy (FV-1000D, Olympus Co., Japan).A reflection image of the scaffold and cells was observed using a 60×1.35NA objective (Olympus Co., Japan).To observe regions of cells that had entered the pores, imaging was performed from the level of the upper surface of the scaffold to a depth of 30 µm at 0.5µm intervals.
Next, the 2D morphologies of the cells were observed by superimposing images taken at the level of the substrate and of the upper surface of the scaffold.To exclude the influence of cell-cell interactions on cell morphology, cells that were well separated from surrounding cells were selected on each patterned scaffold and flat surface.The angles of cell protrusions, clockwise from the vertical di- rection, were estimated using Metamorph Software ver.7.7.

I. Statistics
Statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc., USA).The mean ± standard deviation is shown.One-way factorial ANOVA and Tukey's multiple comparison test were used to compare each patterned scaffold with the flat surfaces and all of the patterned scaffolds to each other.P < 0.05 was considered significant.

A. Fabrication of 3D scaffolds with a constant pore ratio
The 3D shapes of cell matrices reportedly greatly influence cell functions.To investigate this, we used photolithography to fabricate 3D micro-patterned scaffolds with a scale pattern, a check pattern, or a stripe pattern on a silicon substrate, which does not deform elastically.The 3D micro-patterned surface has a silicon crystal face and is smoother than a cover glass.Figure 1(A)-(C) shows color laser 3D microscopy images of scaffolds with these three patterns.The triangular pores had 10 µm long sides and were densely arranged in a scale pattern (Fig. 1(A)).The tetragonal pores had 10 µm long sides and were densely arranged in a check pattern (Fig. 1(B)).In the stripe pattern, the bands were 6 µm wide and 6 µm apart (Fig. 1(C)).In each of the scaffolds, the shapes were oriented perpendicular to the surface and the vertices of the pores were slightly rounded as a result of the processing method.
The cutting planes of the patterns were observed by field emission scanning electron microscopy, and the depths of five pores or grooves of each pattern were measured and the mean depth was calculated.In all of the patterns, the depth (mean ±S.D.) was approximately constant at 22.1±0.6 µm for equilateral triangular pores, 21.9±0.6 µm for regular tetragonal pores, and 21.6±0.3µm for striped grooves.Therefore, the patterned scaffolds had an almost constant depth of approximately 22 µm, which is larger than the thickness of a NIH-3T3 cell, which has a maximum thickness of 4 µm [28,29].The hydrophilicity of these silicon surfaces was enhanced by oxygen cleaning, and cell adhesion on these surfaces was comparable to that on a cleaned cover glass (data not shown).For all experiments, the patterned substrates were processed in a single batch.
To investigate the influence of the unit shape of these scaffolds on cell functions, scaffolds with a constant pore ratio were required.This is because changes in the pore ratio may affect cell functions since such functions are markedly changed when cells are cultured on twodimensional scaffolds [24][25][26][27].To determine the pore ratio of each patterned scaffold, color laser 3D microscopy images of the shapes of the patterns were analyzed.The actual and theoretical pore ratios are shown in Fig. 1(D).The actual pore ratios were nearly constant in each of the patterns.The actual pore ratios tended to be almost equal to the theoretical values (50%).This confirmed that the patterns fabricated were almost the same as the blueprints, in which the pore ratio was constant.
In  FIG.3: Densities of NIH-3T3 cells grown on a flat silicon substrate orientated face-up or face-down.NIH-3T3 cells (seeded at a density of 1×10 4 cells/cm 2 ) were cultured for 72 h on a flat silicon substrate orientated face-up or face-down, or on a haemocytometer cover glass.Cell density was then determined by manually counting the number of cells in a 871 µm×690 µm field.This area was observed using a 10× objective and was located approximately at the centre of the scaffold (width: 1000 µm×1000 µm).Values represent mean ± standard deviation (n = 18).A one-way factorial ANOVA and a Tukey's multiple comparison test confirmed that the cell densities in each of these conditions did not significantly differ after 72 h of culture.
over a wide area could be fabricated using photolithography.These patterns had a similar depth and a constant pore ratio; therefore, the porosities of these scaffolds were very similar.This allowed the effect of the pattern shape of the scaffold, and no other factors, on cell functions to be investigated.

B. Comparison of the proliferation of cells on downward-and upward-facing scaffolds
To observe cells with an inverted microscope, the side of the scaffold on which cells are adhered must be facing down.Moreover, to observe living cells for an extended period of time, there must be a sufficient gap between the downward-facing cells and the bottom of the plate so that the culture medium can diffuse over the cells.To achieve this, the substrate was placed face-down in a 35 mm (12ϕ) glass-bottomed dish and the four corners of the substrate were positioned onto the frame of the glass coverslip in the base (Fig. 2(A)).
To examine whether cell proliferation differs between upward-and downward-facing substrates, NIH-3T3 cells were seeded onto two sterile flat substrates (725 µm thick, 10 mm×10 mm hydrophilic quadrate) at a density of 1×10 4 cells/cm 2 .After 4 h, one substrate was placed face-down into culture medium in a 35 mm (12ϕ) glassbottomed dish (Fig. 2(B)), whereas the other was placed face-up (Fig. 2(C)).After a further 72 h of culture, there was no significant difference in cell density between the two substrates (Fig. 3).Moreover, the cells densities on these two silicon substrates were similar to that on a haemocytometer cover glass placed face-up into a culture dish.These data show that cells grow normally on a substrate placed face-down in culture medium in a 35 mm (12ϕ) glass-bottomed dish, and that this culture method can be used for live cell imaging of an opaque scaffold for 72 h using an inverted microscope.The movie in Supplementary materials is the result of time-lapse observation that was imaged every 30 min for a total of 72 h at four different places on a flat silicon substrate.This movie shows that the cultured cells are highly active, exhibit- ing cellular migration, process extension, and retraction for 72 h.However, the process of inverting the scaffold can cause cells to detach; therefore, the length of time between seeding the cells and placing the scaffold into the culture dish should be varied according to the adhesion properties of the cell line.

C. Trajectories and interior turning angles of migrating cells on 3D micro-patterned scaffolds
The trajectories of NIH-3T3 cells grown on 3D micropatterned scaffolds with various shapes were next investigated.At 4 h after seeding, time-lapse imaging of PKH26labeled cells at five locations on each scaffold was commenced for 72 h (Fig. 4(A)).The trajectories of 4-12 cells on each surface were plotted on the merged fluorescence reflection images of the substrates.To exclude the influence of cell-cell interactions, cell migration was only analyzed over the first 24 h of the observation period, during which cell density was comparatively low.Representative examples of cell trajectories on each surface are shown in Fig. 4(B)-(E).These data show that migrating cells turned at random angles on the flat surface but at defined angles on each of the 3D micro-patterned scaffolds.Furthermore, the angles migrating cells turned were influenced by the repeated shape of the patterned scaffold.Cells tended to turn along the edges of the equilateral triangular pores (Fig. 4(B)), along the edges and the diagonal lines of tetragonal prisms (Fig. 4(C)), and along the edges of the rectangular grooves (Fig. 4(D)).
Next, we attempted to quantify the influence of the unit shape of the 3D micro-patterned scaffold on cell migration by plotting histograms of the interior angles that migrating cells turned (Fig. 5(A)).Histograms of 172 such inte-rior angles of cells grown on each patterned scaffold and flat surfaces were plotted (Fig. 5(B)-(E)).This indicated that cells frequently turned at particular angles on each patterned scaffold.On the scale pattern with equilateral triangular pores, there were peaks at angles of 0 • -10 • , 50 • -80 • , and 110 • -130 • (Fig. 5B).On the check structure with regular tetragonal pores, there were peaks at angles of 0 • -10 • , 40 • -50 • , 80 • -90 • , and 130 • -140 • (Fig. 5(C)).On the stripe structure with rectangular grooves, there were peaks at angles of 0 • -10 • and 160 • -170 • (Fig. 5(D)).However, cells did not frequently turn at any specific angles on the flat surface (Fig. 5(E)).Moreover, the peaks at 0 • -10 • indicate that cells were more likely to turn around and migrate back in the direction they had come on the patterned scaffolds than on flat surfaces.
To explain why cells turned in defined directions on the 3D micro-patterned scaffolds, we must consider two important characteristics of how cells migrate: (I) cells cultured on patterned scaffolds with non-continuous pores primarily extend in a horizontal direction along the continuous upper surface of the 3D scaffolds [30,31], and (II) it has been postulated that cells adhere and extend along the edge of the upper surface [15,[32][33][34].The reason cells tended to turn at the defined angles of approximately 60 • and 120 • on the scale scaffold might be due to their contacting the edges of the equilateral triangles (Fig. 5(F)).In Fig. 5(C), cells mainly migrated in defined directions along the edges or diagonal lines of the regular tetragonal prisms of the check pattern.Cell migration along the diagonal lines was considered to be owing to the important characteristic outlined in point (I).Adjacent tetragonal prisms of the check pattern intersect at an angle of 45 • with the edges.Such migration results in cells turning at 45 • (Fig. 5(G)), which is in agreement with the results shown in Fig. 5

45
• .Many groups report that cells mainly extend parallel with stripe patterned scaffolds [4,12,14,15] and the migration speeds and angles of cells on such surfaces are reported [35].We confirmed that the angles that migrating cells turned on the stripe pattern scaffold were close to 0 • and 180 • (Fig. 5(H)).These data indicate that the angles migrating cells turn are influenced by the subcellular unit-shape of the 3D scaffold on which they are cultured.

D. Morphologies and protrusion angles of extending cells on 3D micro-patterned scaffolds
The top views of the cells varied greatly between the different patterns.In Fig. 6, the morphologies of cells cultured on the scale (A), check (B), or stripe (C) patterns, or on the flat surface (D) are shown.Cells grown on scale with 10 µm long sides were spread and extended protrusions along the edge of pattern.The protrusion angles were multiples of 60 • on the scale pattern (Fig. 6(A)).Cells grown on check with 10 µm long sides were spread and extended protrusions along the edges and the diagonal lines of tetragonal prisms.The protrusion angles were multiples of 45 • on a check pattern (Fig. 6(B)).Cell grown on stripe, which the bands were 6 µm wide and 6 µm apart, elongated parallel with the patterns.The protrusions were extended just in the opposite direction along the edges of rectangular grooves (Fig. 6(C)).Cells grown on the flat surface were spread and extended protrusion at random angles (Fig. 6(D)).These protrusion results in cells extending at defined angles on each pattern (Fig. 6(A)-(C)), which is in agreement with the results shown in Fig. 5(B)-(D) where there were defined interior turning angles of migrating cells on each pattern.

IV. DISCUSSION
In cells cultured on 3D micro-patterned scaffolds, the expression levels of adhesion signaling complexes [36], focal adhesion kinase [13,36], and F-actin [13,15,37], their proliferative [8,13,20,22], cell aggregation [17][18][19][20], and metabolic capacities [13,[19][20][21], and their differentiation potential [9,23], differ from those of cells cultured on flat surfaces.Although such differences have been demonstrated in many cell types, the underlying mechanisms remain unclear.It is difficult to quantitatively compare the effects of 3D scaffold shape on cell behavior on the basis of previous reports because the patterns used differ in terms of material composition, shape, pore ratio, and depth.To determine the physical properties of 3D micro-patterned scaffolds that underlie these differences in cell behaviors, we examined cells cultured on scaffolds with repetitive patterns of various shapes.We investigated whether cell migration differs when cells are grown on these patterned scaffolds, which differ only in terms of unit shape, and if so, the underlying mechanism.
On 3D scaffolds, cells primarily migrate in a horizontal direction along a long, continuous path.Cells migrate along a stripe [35,38] or lattice-shaped pattern [39]; however, on scaffolds with other patterns, it is unclear whether cells migrate along the pattern.Moreover, we were interested in whether the unit shapes of the repetitive patterns with a constant pore ratio affected cell migration.To investigate the influence of the unit shape of these scaffolds on cell functions, scaffolds with a constant pore ratio were required.This is because changes in the pore ratio may affect cell functions since such functions are markedly altered when cells are cultured on twodimensional scaffolds [24][25][26][27].There has been no report comparing cell migration on 3D scaffolds with a constant pore ratio.In this study, we examined the effect of unit shape of patterned scaffolds on cell migration by preparing various scaffolds on a silicon substrate that had a constant pore ratio and depth.Because patterns fabricated on a single crystal silicon substrate are not elastically deformed, elasticity did not vary between the different scaffolds.This meant that the influence of the scaffold pattern, and no other factors, on cell migration could be investigated.We demonstrated that the subcellular unitshape of a repetitive pattern regulates the direction that migrating cells turn.The angle at which cells turned correlated with the unit shape of the scaffold: the interior turning angles were multiples of 60 • on a scale structure with equilateral triangular pores, multiples of 45 • on a check structure with regular tetragonal pores, and close to 0 • or 180 • on a stripe structure with rectangular grooves.Therefore, the angle that cells turn is influenced by the unit shape of the 3D patterned scaffold on which they are cultured.Furthermore, when the angles at which the migrating cells turned were investigated in detail, it was found that a cell turns in one of two directions that correlate with the unit shape of a scaffold; one corresponding to the edge of the pattern, and the other corresponding to the upper surface of the pattern.In the stripe and scale patterns, the direction in which the upper surface continues is the same as the direction in which the pattern edge continues; however, these directions differ from each other by 45 • in the check pattern.By observing cell migration on these patterns, we found that cells turned in one direction on the stripe and scale patterns, where the edge and the upper surface are continuous, but turned in two directions on the check pattern, where the direction in which the pattern edge continues differs from the direction in which the upper surface of the pattern continues.Our study clearly showed that the direction cells turn can be controlled by two shape elements, namely, the continuous directions of the edge and upper surfaces of a pattern.
To discuss the mechanism that underlies this, we must consider three important properties of cell extension on 3D micro-patterned scaffolds.First, cells cultured on patterned scaffolds with non-continuous pores primarily extend in a horizontal direction along the continuous upper surface of the 3D scaffolds [30,31].Second, cells are postulated to adhere and extend along the edge of the upper surface [15,[32][33][34].Third, when cells extend along a 3D micro-pattern of parallel grooves, F-actin aligns parallel to the edges of these grooves [15,37,40].We hypothesize that the angles migrating cells turn on patterned scaffolds is determined by the orientation of the continuous upper surface, the edge of the repeated unit shape to which cells adhere and extend along, and the expression of F-actin parallel to this edge.Observing the angles at which cells extended helped us to understand why the angle that migrating cells turn markedly differed between the three micro-pattered scaffolds.The angles at which cells extended on each scaffold correlated with the angles at which migrating cells turned.Cells migrate in the direction that they extend protrusions.Therefore, the unit shape of the micro-patterned scaffold affects the angle at which cells extend, which in turn affects the angle at which migrating cells turn.To further clarify this mechanism, we plan to investigate F-actin morphology, p-FAK expression, and Cdc42/Rho activation in migrating cells cultured on each of the patterned scaffolds.
This study demonstrates that the angle at which cells extend and migrate is influenced by the subcellular unitshape of the 3D micro-patterned scaffold upon which they are cultured.These angles influence the direction and distance cells migrate, which reportedly affect cell aggregation and cluster formation, and these influence cell differentiation [9,24,[41][42][43], growth [17,44], and metabolic function [13,[19][20][21].If the direction and distance that cells migrate can be controlled by modulating the angles at which they turn, cell aggregation and cluster formation could be controlled and thus cell differentiation, growth, and metabolic function could be modulated.
In this study, it was necessary to image live cells on opaque scaffolds using an inverted microscope.However, it is difficult to observe opaque scaffolds using an inverted microscope, because the side of the scaffold to which cells adhere must face-down.This type of live cell imaging had been performed only for a short period of time (24 h at longest) in previous work, and the viability and vitality of cells in the face-down approach has not been compared with that of cells in upward-facing substrate [38,[45][46][47].To observe living cells for an extended period of time, there must be a sufficient gap between the downwardfacing cells and the bottom of the plate to allow the culture medium to diffuse over the cells.We achieved this by placing the substrate face-down into culture medium in a glass-bottomed dish.This method enabled cell mihttp://www.sssj.org/ejssnt(J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/) e-Journal of Surface Science and Nanotechnology Volume 12 (2014) gration on 3D micro-patterned scaffolds to be imaged for an extended period of time (72 h) keeping the viability and vitality of cells.The movie in the Supplementary materials is the result of time-lapse observation that was imaged every 30 min for a total of 72 h at four different places on a flat silicon substrate.This movie shows that the cell-cell interaction is influenced on the cell migration more strongly as the increase in cell density.That is, our procedure make it possible to carry out time-lapse observation of not only a cell migration but growth, differentiation, or a metabolic turnover on opaque substrates.So, we think that this simple procedure can greatly contribute to researches of cell scaffold.Furthermore, this method is labor and cost effective, is extremely simple to perform, and avoids the need for cell fixation; therefore, it can be applied to various scaffolds.We expect this method to be useful for the simultaneous observation of live cells and 3D substrates, including opaque scaffolds, optically transparent glass, and polymer substrates.Because focus image can be distorted when light go through the optical transparency 3D structure.

V. CONCLUSIONS
The goal of this study was to determine the influence of the unit shape of a 3D micro-patterned scaffold on cell migration.We fabricated 3D micro-patterned scaffolds consisting of a scale structure with equilateral triangular pores, a check structure with regular tetragonal pores, and a stripe structure with rectangular grooves.These scaffolds differed only in terms of unit shape and had a constant pore ratio and depth.As the pore depth was greater than the thickness of a cell, it did not influence cell migration.This allowed us to qualitatively describe how unit shape influences cell migration.We devised a method to simultaneously observe live cells and the surface topology of an opaque scaffold using an inverted microscope for an extended period of time.Scaffolds on which NIH-3T3 cells were adhered were placed face-down in culture medium in glass-bottomed dishes, and cell migration and the 3D shape of the scaffolds were imaged over 72 h.The angles migrating cells turned markedly differed between the three scaffolds, and correlated with the unit shape of the scaffold.The angles migrating cells turn were investigated in detail.It was found that a cell turns in one of two directions that correspond with the unit shape of a scaffold, namely, the direction in which the pattern edge continues and the direction in which the upper surface continues.This study demonstrates that the subcellular unit shape of 3D micro-patterned scaffolds affects cell migration.
We developed a novel means to perform live cell imaging for an extended period of time using an inverted microscope.We used this technique to reveal that the unit shape of the scaffold on which cells are cultured affects the angle at which migrating cells turn, which correlates with the angle at which cells extend.In conclusion, the subcellular unit shape of the scaffold on which cells are cultured affects the angle at which cells extend, which in turn affects the angle at which migrating cells turn.

FIG. 1 :
FIG.1: Confocal laser scanning microscopy images of the 3D micro-patterned scaffolds.Photolithography was used to fabricate 3D micro-patterned scaffolds on a silicon substrate.(A) Top view of the scaffold with triangular pores with 10 µm long sides and densely arranged in a scale pattern.The adjoining pores are not connected.(B) Top view of the scaffold with tetragonal pores with 10 µm long sides and densely arranged in a check pattern.The adjoining pores are not connected.(C) Top view image of the scaffold with 6 µm wide bands that are 6 µm apart arranged in a stripe pattern.(D) Comparison of the pore ratios of the 3D micro-patterned scaffolds.Actual pore ratios (total pore area per unit area) of the scale, check, and stripe patterned scaffolds were calculated using image analysis software.Values are mean ± standard deviation (n = 5).The dashed red line represents the theoretical value (50%) based on the blueprints of the photomasks.A one-way factorial ANOVA and a Tukey's multiple comparison test confirmed that the pore ratios of the three scaffolds did not significantly differ.

FIG. 2 :
FIG.2: Culture of NIH-3T3 cells adhered to a flat silicon substrate face-down in a glass-bottomed dish.(A) A 35 mm 12ϕ glass-bottomed dish has a circular hole in the bottom, over which a cover glass is applied on the outside of the dish with silicon.Therefore, there is a slightly depressed circle (1.3 mm) with a diameter of ∼12 mm at the bottom of the dish.The four corners of the 10 mm silicon substrate quadrate were placed onto the frame of this base.(B) Cells were seeded onto a flat silicon substrate, and placed face-down onto the frame of the glass base.There is a 1.3 mm gap between the silicon substrate and the glass base, and the silicon substrate does not completely cover the hollow of the glass base; therefore, the culture medium can diffuse between the downward-facing cells and the bottom of the dish.(C) As a control, the substrate was placed face-up in a 35 mm cell culture dish containing culture medium.

FIG. 4 :
FIG. 4: Trajectories of NIH-3T3 cells grown on 3D micro-patterned scaffolds or flat surfaces.(A) To observe migrating cells on scaffolds using an inverted fluorescence microscope, PKH26-labeled cells were allowed to adhere to the substrates, which were then inverted and placed face-down into glass-bottomed dishes.The cells and substrates were imaged for 72 h, during which images were acquired every 30 min.The patterned scaffolds had 25 patterned areas within an area of 1 mm×1 mm.Cell movements were manually tracked.Representative examples of the trajectories of individual cells are shown on the scale (B), check (C), and stripe (D) patterns, and on the flat surface (E).The red dots denote the starting positions of the cells and the yellow lines show their trajectories.To exclude the influence of cell-cell interactions, cells were only tracked for the first 24 h of the 72 h period, during which cell density was comparatively low.

FIG. 5 :
FIG. 5: Estimation of the angles these cells turn.(A) Schematic illustration of the turning angle and the interior angle of a migrating cell that changes direction.Histograms of 172 interior angles of cells cultured on the scale (B), check (C), and stripe (D) patterns, and on the flat surface (E).All histograms consist of 17 unit intervals of 0 • -170 • in 10 • bins.Arrows indicate peaks.(F) Cells cultured on the scaffold with a scale pattern turn at interior angles of approximately 0 • , 60 • , or 120 • (every 60 • ).(G) Cells cultured on the scaffold with a check pattern turn at interior angles of approximately 0 • , 45 • , 90 • , or 135 • (every 45 • ).(H) Cells cultured on the stripe pattern turn at interior angles of approximately 0 • or 170 • (close to 180 • ).

FIG. 6 :
FIG. 6: Morphologies of cells extending protrusions and the angles of these protrusions on 3D micro-patterned scaffolds.A reflection image of the scaffold and cells cultured on the scale (A), check (B), or stripe (C) patterns, or on the flat surface (D).The yellow arrows indicate the directions that cells are extending in.The angles that cells are extending in, clockwise from the vertical direction, are shown in yellow.