Colloids and Surfaces B: Biointerfaces

Optimal functionality of native corneal stroma depends on a well-ordered arrangement of extracellular matrix (ECM). To develop an in vitro corneal model, replication of the corneal in vivo microenvironment is needed. In this study, the impact of topographic cues on keratocyte phenotype is reported. Photolithography and polymer moulding were used to fabricate microgrooves on polydimethylsiloxane (PDMS) 2 – 2.5 μ m deep and 5 μ m, 10 μ m, or 20 μ m in width. Microgrooves constrained the cells body, compressed nuclei and led to cytoskeletal reorganization. It also in ﬂ uenced the concentration of actin ﬁ laments, condensation of chromatin and cell proliferation. Cells became more spread and actin ﬁ lament concentration decreased as the microgroove width increased. Relationships were also demonstrated between microgroove width and cellular processes such as adhesion, migration and gene expression. Immunocytochemistry and gene expression (RT-PCR) analysis showed that microgroove width upregulated keratocyte speci ﬁ c genes. A microgroove with 5 μ m width led to a pro-nounced alignment of cells along the edges of the microchannels and better supported cell polarization and migration compared with other microgroove widths or planar substrates. These ﬁ ndings provide important fundamental knowledge that could serve as a basis for better-controlled tissue growth and cell-engineering applications for corneal stroma regeneration through topographical patterns.


Introduction
Cell behavior can be affected by their surrounding physical environment. Understanding the response of cells to different structural and topographical cues needs to be considered when developing materials that directly interact with cells for applications such as medical implants, cell culture systems or scaffolds for tissue engineering. Several cell types have previously been shown to modulate their behavior in response to changes in surface topography in vitro [1][2][3][4]. As an example, topographical patterns consisting of microscale grooves and ridges have been shown to encourage cell alignment [5]. The spatial and temporal patterns shown by cell colonies during tissue growth are a result of cell-cell interaction, influence of extracellular forces, cell movement and coordinated cell growth. Hence, these patterns play an important role in organ development [6].
Our understanding of how cells imbibe topographic signals remains incomplete. Cell responses can vary depending on the specific topographical pattern. Within organs and tissue, the microenvironment surrounding cells also presents structural constraints that influence the regulation of cellular functions [7][8][9][10]. Cell geometry affects nuclear deformation, cytoskeleton organization, gene expression, cell growth and apoptosis, cell division and chromatin compaction [11][12][13]. Similarly, other biophysical cues in the cell's microenvironment (e.g., fluid flow, substrate stretching and rigidity) affect gene expression and cellular and nuclear architecture [14][15][16]. While the impact such parameters have on cellular function remains obscure, multiple studies unequivocally confirm that the topography of the cells microenvironment is critical in regulating their behavior [17][18][19]. Several in vitro studies have shown that patterns on the substrate impact cell adhesion, differentiation, and migration [20]. The biophysical and biochemical stimuli influencing cell behavior have been linked to substrate-cytoskeleton crosstalk [21].
The corneal stroma is dependent on a high degree of structural organization to maintain its transparency and mechanical characteristics. The extracellular matrix (ECM) structure and composition are maintained by resident corneal stromal cell called keratocytes [22]. These cells interact with each other, the surrounding ECM and nerve fibers [5]. The ECM primarily consists of collagen aligned fibrils 22.5 nm-35 nm in diameter with a center-to-center gap of approximately 62 nm. These fibrils form lamella of thickness 2 μm-200 μm [23] in between which the keratocytes reside. Nerve fibers are mostly limited to the anterior stroma where they are in the form of bundles of up to 20 μm diameter [5] and run parallel to the collagen fibrils. The combination of these different microstructural features in the stroma likely influence how keratocytes behave.
Substrate topography, through contact guidance can strongly impact keratocytes. Arrays of parallel grooves and ridges, in nano and micro dimensions, have previously been used to models in vivo ECM architecture of the corneal stroma. Substrate topographic cues have been known to affect orientation of stromal cells and orientation and elongation of keratocytes. A study by Teixeira et al. [5] on silicon substrates with groove pitches between 400 and 4000 nm showed that for pitches over 800 nm, 70 % of human keratocytes were elongated and aligned along the grooves. Studies have also shown ridges and grooves affecting keratocyte migration and morphology in rabbit cornea [22]. Irrespective of surface chemistry, human corneal keratocytes were shown to proliferate more as ridge width increased from 0.35 to 10 μm [24]. Corneal keratocytes have been shown to slow down in their proliferation rate on nano scaled grooves, compared to microscale grooves and smooth substrates [25]. Koo et al. [24] also observed, primary human keratocytes proliferated slower with decreased groove width. Zhang et al. [26] used a biomimetic 3D corneal model, made up of patterned silk films and collagen gel to show that the topography of the silk films affected keratocyte alignment and ECM arrangement. The 3D model showed higher expression of keratocyte marker and expression of ECM compared to the 2D culture and the dome shaped model, with 3% strain showed higher expression of keratocyte marker.
To better understand how keratocytes respond to different topographical cues, microstructured substrates with microgrooves and microridges were fabricated using polydimethylsiloxane (PDMS). During tissue development and growth, cell colonies exhibited a variety of spatial and temporal patterns, as an outcome of coordinated cell growth, movement, and cell-cell communications. Here we show that geometrical confinement, induced by topographically patterned microgroove substrates, modulated cell and nuclear morphology and cellular behaviors like alignment, orientation, and migration. These factors are critical for biological phenomenon like embryogenesis, wound healing, metastasis, inflammation, and colonization of biomaterial scaffolding [20]. The results indicate that the topographic constraints significantly squeezed the cells and nuclei, increased chromatin condensationleading to changed cell proliferation rateand cytoskeleton reorganization and anisotropic growth of the cultures. The role played by filopodial probing (adhesion establishment, focal adhesion orientation), cell environmental sensing, and growth in perception of and reaction to surface topographic features were also investigated.

Fabrication of micro-patterned substrates
Micro-patterned silicon master molds were fabricated under cleanroom conditions as described earlier [27]. Briefly, molds were fabricated by standard soft lithography using a laser mask writer (Heidelberg DWL66), reactive ion etching and low-pressure chemical vapor deposition for silicon oxide coating. The molds had straight, parallel microgrooves made of photoresistant material. There were three master molds, each 1 × 1 cm in dimension: a) 5 μm microchannels with 5 μm spacing b) 10 μm microchannels with 10 μm spacing and c) 20 μm microchannels with 20 μm spacing. Dry plasma etching (OIPT Plasma lab System100 ICP180) was used to etch the master molds. 15 min of etching was utilized to achieve the depth of microchannels. Micropatterned polydimethylsiloxane (PDMS) substrates were fabricated by mixing elastomer base and curing agent (10:1 ratio by weight). The elastomer solution was degassed and poured on the master molds. The molds were then cured for 24 h, at 60°C. Control (non-patterned) substrates were fabricated by casting the elastomer solution on a 60 mm polystyrene Petri dish (Corning, Corning, NY, USA), followed by the same curing. A schematic overview of the fabrication process has been provided Supplementary Fig. 1.
The PDMS patterned substrates (1 cm × 1 cm) were characterized by scanning electron microscopy (SEM, SUPRA 35 V P, Carl Zeiss) and white light interferometry (Omniscan MicroXam-non-contact optical method) to measure the surface roughness (RMS: root mean square) and the depth of microgrooves channels.

Cell culture
Human corneal stromal cells at passage 3 were used in this study. The cells were isolated for use in this study as previously describe [28] in accordance with the declaration of Helsinki and with ethical approval from Trinity College Dublin School of Medicine Research Ethic Committee. The culture medium used to expand cells was comprised of Dulbecco's Low Glucose Modified Eagles Medium (DMEM) (Hyclone; Thermo Fisher Scientific) supplemented with 10 % fetal calf serum (Hyclone) and 1 % penicillin/streptomycin. The culture took place in a controlled, humidified environment (5 % CO 2 , 37°C) until cells reached 90 % confluency. At this point, cells were trypsinized and centrifuged to be counted while suspended in the media. Cells were stained (Trypan Blue) and counted using a Haemocytometer. The patterned substrates (1 × 1 cm 2 ) were sterilized by UV exposure (30 min) and soaked in low glucose DMEM medium for 4 h. The matrices were then partially dried before cell seeding.
10 μL of cell suspension in media, containing 10 3 cells, were seeded dropwise onto each matrix. The matrices were maintained in a humidified environment, at 37°C and 5 % CO 2 for 1 h after seeding. Culture medium was added to the cell seeded substrates consisting of DMEM/ F12 (Hyclone; Thermo Fisher Scientific) supplemented with L-ascorbic acid (0.1 mM), Insulin-Transferrin-Selenium (Gibco) (1 μl/ml), and 1 % penicillin/streptomycin. Media was replaced on every alternate day.

Cellular proliferation
Cell proliferation was assessed at multiple time-points over the 21 day culture period using PrestoBlue reagent (Molecular Probes; Invitrogen, Carlsbad, CA), following the manufacturers protocol.

Cytoskeletal organization
To observe cytoskeletal organization, following 3 days of culture, cell laden substrates were fixed with 4% paraformaldehyde for 15 min at room temperature. Cells were permeabilized with 0.1 % Triton X-100 solution in 1 % bovine serum albumin (BSA) and then blocked with 2 % BSA for 1 h. Phalloidin-TRITC and 4′,6 -diamidino-2-phenylindole (DAPI) were used to stain actin filaments (F-actin) and cell nuclei, respectively. For both stains, respective manufacturer's protocols were followed. Leica SP8 confocal microscope was used to examine the cytoskeletal organization on the matrices with LAS X Advanced software being used for image post-processing.
F-actin alignment and density distribution were analyzed. A pseudo color scale was used to present the F-actin distribution fluorescence and intensity [11].
F-actin and DAPI staining were used to outline the cell perimeter and to calculate the cell and nucleus areas [29]. Analyze Particles, from Fiji [30] was used to quantify these areas. At each time point, for each substrate, 30 cells were analyzed.

Analysis of elongation and degree of orientation
The elongation and orientation of TRITC-phalloidin stained cells were measured 4, 12, and 24 h after initial seeding as previously described [13]. Cell elongation was defined as the ratio of the maximum to the minimum principal moment of inertia ( Supplementary Fig. 2). Cell orientation was defined as the angle (θ) between the principal axis of inertia of the elliptical form and a reference axis. θ varied between 0 and 180°. For reference axis, the pattern direction was used in the micro-grooved surfaces while the horizontal axis was used for the control surfaces (no grooves) -Schematic in Fig. 3. When θ exceeded 10°, cells were considered to have random orientation and were otherwise considered to be aligned.

Focal adhesions (FAs) and morphometric analysis
Vinculin staining was used to analyze focal adhesion. 12 h after cell seeding, cell-laden constructs were fixed and permealized as described earlier and then incubated with mouse monoclonal anti-vinculin antibody (Abcam) (1:400 dilution) overnight at 4°C. The following day, substrates were washed thrice with PBS, 5 min per wash. Washed substrates were incubated in Alexa fluor-488-conjugated rabbit antimouse IgG (Abcam)incubation in 1:200 dilution, for two hours at 37°C . Actin and nucleus was counter stained as described previously. Substrates were then again thoroughly washed with PBS followed by confocal microscopy: Leica SP8 for imaging and LAS X Advanced for image post-processing.
The methodology from Maruoka et al. was used for FA morphometric analysis (length vs orientation) [31]. Only FAs with area over 0.6 mm 2 were part of the statistical analysis. ImageJ command "Measure" was used to measure FA length. FA lengths (n = 60) were analyzed with respect to pattern orientation.

Apparent chromatin condensation
DAPI staining was used to visualize chromatin and to analyze the spatial chromatin organization. Nuclear images at different focal positions along the z-axis, every 0.2 mm interval, were acquired using confocal scanning microscopy (Leica SP8) and stacked. Images were sharpened with ImageJ plugins (shading correction, dark image subtraction, Deconvolution Lab) to extract quantitative information from them. As the sum of the intensity of each pixel, the integrated fluorescence intensity was quantified. The average spatial density is the ratio of total fluorescence intensity to nuclear volume and correlates with average chromatin packing ratio [13]. It is indicative of chromatin condensation.

Filopodia feature analysis through SEM
Cell laden substrates were fixed (4 % PFA, 15 min, room temperature) and filopodia examined 24 h after seeding. Cells were subjected to graded dehydration steps using ethanol (30-100 % v/v H 2 O, steps of 10 %, 20 min. in each). Dehydrated samples were exposed to isoamyl acetate for 5 min, followed by vacuum drying and gold coating for scanning electron microscopy (SEM, Zeiss Sigma 300, operating voltage of 10 kV).

Cell migration analysis
4 h after initial seeding, cell migration was monitored over a further 6 h. The study focused on early intervals following seeding since the cell density is low during this period, helping us avoid the impact of cell-cell interactions. Six representative areas of each substrate were selected. DIC images were then acquired of each area using an EC Plan-Neofluar 10x (N.A. 0.3) objective every 15 min on a Carl Zeiss AxioImager Z1 (Jena, Germany) equipped with environmental enclosure and CO 2 and hardware autofocus. Time-lapse videos were manually analyzed with Track mate Plugin of Fiji. This analysis yielded cell trajectories. Mean migration rate was defined as the total movement of the call per unit time and directionality was assessed as movement across vs movement along pattern direction [18].

Evaluation of Extracellular Matrix (ECM) characterization and matrix stiffness
ECM formation on the cell laden substrates and the culture media was quantified using biochemical analysis after 14 and 28 days of culture period. Bisbenzimide Hoechst 33258 DNA assay with calf thymus DNA was used as the reference standard for quantifying DNA content [33]. Cells on substrates were macerated by a papain solution (125 μg/ml). Dimethyl methylene blue (DMMB) binding assay (Blyscan; Biocolor Ltd, Antrim, UK) was performed to determine levels of sGAG (acid sulfated glycosaminoglycans), standard used was bovine chondroitin sulfate. Matrix stiffness of all substrates after 7 days of culture was evaluated by contact mode atomic force microscopy (AFM-Park NX10) using force-displacement curve (Park Systems XEI software) [34]. Standard triangular silicon nitride cantilevers from Veeco (DNP) with a nominal spring constant of 0.06 N/m and a nominal tip radius of curvature of 20 nm were used. The XEI software calibration tool was utilized to measure the cantilever spring constant. The AFM was run in the force-volume mode at frequency of 7 Hz to obtain force distance curve.

Immunocytochemistry
Following 14 days in culture, cell-laden substrates were pre-processed for immunocytochemistry analysis using the same method as for F-actin staining. After permeabilization and blocking, cells were incubated at 4°C in anti-ALDH3A1 1:50 (ab76976; Abcam, Cambridge, United Kingdom), anti-keratocan 1:50 (sc-66941; Santa Cruz, Heidelberg, Germany), and anti-α-smooth muscle actin (αSMA) 1:50 (ab7817; Abcam) for 18 h. The cells were then thoroughly washed with PBS. They were then incubated for one hour, at room temperature, in a dark room, in fluorescently labelled antibodies. To detect Keratocan and ALDH3A1, Donkey anti-rabbit AlexaFluor 488 (ab150073; Abcam) was used. For αSMA, goat anti-mouse biotin was used, succeeded by Extr Avidin-FITC (B7151 and E2761). For nuclear staining, the substrates were then washed with PBS and incubated with DAPI (1 mg/ mL,1:500 dilution). Confocal microscopy was carried out using the Leica SP8 and post-processing used LAS X Advanced software.

Statistical analysis
The R statistical platform was used for all statistical analysis. Oneway ANOVA, followed by Tukey's post-hoc HSD test was used for comparing the results from different substrates. Significant differences have been marked as ***p < 0.001; **p < 0.01; *p < 0.05. Data presentation is in the form mean ± standard deviation (SD). Unless otherwise specify, sample size was 3.

Substrate characterization
To study the cell response to aligned topography, PDMS substrates with aligned micro-grooves were fabricated. To verify pattern transfer, SEM was used to inspect the groove and channel width ( Fig. 1(a)). The  grooves were of expected size (5 μm, 10 μm and 20 μm) and no defects were detected on the surface. The tops 10 μm and 20 μm ridges were also of expected size, although the 5 μm were slighty smaller (≈ 4.6 μm) due to the sides of the groove not being completely vetical. White light interferometry was used to evaluate depth of the channels (15 min plasma etching, ∼2 μm to 2.5 μm) and their average surface roughness (RMS ∼ 275−305 nm).

Cell proliferation and size changes in response to the substrate topography
The cells growth kinetics were studied over 21 days (Fig. 1(b)). Cell numbers increased linearly with time on all substrates. After 3 days, cell growth kinetics did not differ significantly between patterned substrates and control substrates. On days 7, 14 and 21 the patterned substrates had significantly more cells than the control. There was no significant difference in proliferation between cells on the different sized microchannels over 21 days.
Cell size affects vital cellular processes such as growth, morphology, differentiation and death [13]. Cells grown on micro-patterned substrates were significantly (*p < 0.05) smaller than those on the control. The nuclei of cells on patterned substrates were also significantly smaller (*p < 0.05) than the nuclei of cells grown on the control substrates. This finding is relevant in the context of the proposed mechanistic coordination between cell size, nuclear size [13] and cell cycle times [35].

F-actin network remodelling
Modulation of the F-actin network (e.g., cross-linking density, length) will have an impact on cell shape and size [36][37][38][39]. It was observed that for the patterned substrates, the density of the actin filaments was affected, indicating a change in the cytoskeletal tension of the cultured cells. The representation in Fig. 2 uses colour coding (from blue to red) to show increasing density of F-actin. The actin density was noticeably high for the 5 μm patterned substrates compared to the other substrates. The length of actin filaments on the patterned substrates was also higher than on the plain substrates. Topographic cues present on microgroove substrates altered the cells shape that in turn influenced the stress fibre redistribution.

Chromatin condensation affected by substrate topography
DAPI staining was used to investigate effect of topographical cues on cell size and shape [11,40]. DAPI uptake levels were correlated to total DNA and its condensation level [41]. The ratio between integrated fluorescence intensity and nucleus volume may be considered an average spatial density that can be used as a reliable indicator for average chromatin condensation [9,37]. Fluorescence intensity increased in proportion to chromatin condensation level. Chromatin distribution reorganization was found to be associated with decreasing microgroove separation, least for the 5 μm patterned substrates. Nuclear reshaping was found to be associated with intense levels of chromatin condensation. The topographical cues led to the remodelling of cell as well as nuclear shape and resulted in reduced nuclear area.

Impact of substrate topography on cellular degree of orientation and alignment
Previous works have shown that cell organelles like centrosome, nucleus, and Golgi apparatus can orient themselves responding to external, physical stimuli [42,43]. At 4, 12, and 24 h following cell seeding, the orientation of the cell (nucleus) major axis was measured with respect to the pattern orientation on the substrate (Fig. 3a). The cell major axis is defined as the principal axis of the moment of area of the cell's two-dimensional projection. The cells on the patterned substrates significantly changed their nuclear and actin orientation, compared to those on the plain surface that lacked any specific orientation ( Supplementary Fig. 3a). 4 h after seeding, cells on 5 μm (∼36.2 %) and 10 μm (∼23.3 %) grooved substrates had more readily oriented themselves along the pattern, compared to the 20 μm (∼16.6 %) pattern substrates (Fig. 3b). Over time the cells on microgroove substrates became more orientated along the direction of the grooves. Nuclei orientation corresponded to the substrate topography's orientation. Cell The patterned surfaces provided topographic cues that could prompt cell elongation and over shorter time duration than they would occur on a plain substrate.

Focal adhesion
Vinculin staining was used to identify FAs (Fig. 4a). There was a clear difference in the FAs between cells on the patterned and plain substrates. However, not all cells on the patterned surfaces had the same amount of FAs ( Supplementary Fig. 3b). 12 h after seeding, vinculin was diffuse in the cell cytoplasm. FAs on the plain surfaces lacked any defined orientation while those on patterned surfaces were longer, located along the top of ridges and directed in the groove direction. This was more noticeable for the 5 μm and 10 μm substrates when compared to the 20 μm substrates (Fig.4a).
FA lengths, distributed for the different pattern orientations have been presented in Fig. 4b. For the different substrates, the FA length and their spatial distribution was different. After 12 h in culture, cells on all substrates exhibited FAs. Cells on 5 μm and 10 μm substrates displayed the most oriented and longest FAs.

Characteristics of filopodia on different patterned substrates
Filopodial probing, FA establishment and growth are the precursors to cell spreading and elongation. On the plain substrates, the filopodia were straight, while they were more bent on the patterned substrates. The bending was to allow the cells to follow the pattern direction (Fig. 5a). For the different microgroove substrates, the average filopodia length was not significantly different (Fig. 5b). The average length did not exceed 7.8 ± 1.4 μm while most of the filopodia were under 5 μm in length. This seems to imply that length scales of filopodial probing may be independent of surface micropatterns.

Direction of cell migration by topographic features
As topographic cues affected FA maturation, spatial distribution, cell orientation and elongation, it would be logical to assume that the surface microgrooves may also affect cell migration. Cell migration rates were analyzed through time-lapse images of the surfaces. For each substrate examined, 6 cells were tracked as they moved over the 6 h period. The starting point of each cell was mapped onto the origin (0,0) of the coordinate system used for plotting the cell positions (Fig. 6a). Cell migration trajectories on the plain substrate were isotopically distributed. For the patterned substrates, there was a correlation between cell trajectory and pattern orientation. This co-alignment was most conspicuous for the cells on the 5 μm pattern. Mean cellular migration rate was significantly higher on the patterned substrates, than on the plain substrate (p < 0.001, Fig. 6b). Among the patterned substrates, the 5 μm patterned substrate's mean cellular migration rate was remarkably high (∼24.64 μm/h). The values for the 10 μm and 20 μm substrates were not significantly different.

Characterization of ECM
Corneal tissue generation requires that cells synthesize and secrete ECM that is specific to cornea. The most common, negatively charged macromolecules found in the corneal-stromal ECM are glycosaminoglycan (GAG) polysaccharides. Under normal circumstances, they are attached to proteoglycan core proteins by a covalent bond. Their contribution towards corneal transparency, nerve growth cone guidance, and cell adhesion can be critical [44]. GAG synthesis from the cells cultured on the different substrates were significantly higher for the patterned substrates, compared to the plain control, at day point (Fig. 7a). GAGs like chondroitin sulphate and keratan sulphate have a necessary role on controlling inter-fibril distances and transparency of the cornea [45]. After 7 days of culture, it was noted that the GAG deposits had stiffened the substrates (Fig. 7b) and this was significantly more for the patterned substrates than for the plain substrates. Stiffness for all substrates was recorded before seeding cells and it was similar (∼25 ± 3.85 MPa) for all substrates.

Immunofluorescent staining and gene expression
After 21 days of culture, RT-PCR was used to quantify the gene expression (Fig. 8). Presence of micropatterns on the substrate led to increased expression of the keratocyte markers ALDH3A1, Keratocan, Decorin and Lumican, compared to the plain substrates. Collagen types I, III, and V are the most common collagen components of corneal stroma and all three had significantly higher levels of expression for the patterned substrates after 21 days in culture. Negative expression of αSMA (ACTA2) was recorded for all of the substrates at day point 21. This, in addition to the results from the immunocytochemical straining, implies that myofibroblastic differentiation was inhibited on all the substrates. Immunofluorescent staining was used to detect molecules associated with native keratocyte and myofibroblastic cell types (Supplemental Fig. 4). Cells on all the substrates exhibited the keratocyte markers ALDH3A1 and keratocan. None of the substrates showed any detectable levels of αSMA, a myofibrotic marker.

Discussions
In this study, the presence of microgrooves on a substrate, ranging in size from 5 to 20 μm, were shown to influence the behaviour of corneal keratocytes. The micropatterned substrates mimic the parallel orientation of fibrils in each lamella of the stroma. A high percentage of keratocytes in the study aligned along the microgroove patterns implying that the cells could recognize orthotropic topographic stimuli [5]. Focal adhesion, stress fibers, orientation, migration and phenotype were all influenced by the surface topography.
The cells stress fibers and FAs were mostly aligned with the patterned topography in this study. It has previously been shown that nanoscale features in the topography partially inhibited FAs and stress fibers while they are abundant on substrates presenting microscale topography [5]. The current work shows that geometrical confinement, induced due to topographically patterned microgroove substrates modulated cell and nuclear morphology, along with such cellular behavior as alignment, orientation and migration. These factors are important for several biological phenomenon, including, embryogenesis, wound healing, metastasis, inflammation, and colonization of biomaterial scaffolding [22]. Topographic constraints led to significant squeezing of cells and nuclei, chromatin condensationleading to changed cell proliferation rateand cytoskeleton reorganization. Smaller groves (5 μm) on substrates had the most F-actin expression in cells compared to cells cultured on substrates with larger grooves (10 or 20 μm).
Cells grown on all the micropatterned substrates had significantly smaller nuclei and cell size, compared to the controls, which were devoid of any patterns. Previous studies have suggested that deformation of nuclei results from change to the cells morphology [11,46]. In addition, we observed that cells on micropatterned substrates expanded more quickly than the cells on the control substrate. Cell cycle times have previously been shown to be shorter in smaller sized cells than the average cells [35], which is in line with our findings.  Despite their limited motility, keratocytes can actively probe their surroundings using filopodial. The cells can "sense" their environment through lamellipodial and filopodial extensions along the surface. Through interconnections among them, they form a cellular syncytium [47,48]. The filopodial extensions were mostly aligned along the pattern on the substrates. This in part explain why 5 μm substrates had the highest cellular migration rates. Filopodial probing is closely related to filopodia length. As features at a greater distance than filopodia length cannot be reached, new FAs cannot be formed. Herein comes the role of topographical cues. If topographic features' spacing is smaller than filopodial lengths, establishing fresh points of focal adhesion becomes easier for the cell. Growth of FAs, in turn, influences cell adhesion and polarization. The microgrooves have the potential to affect filopodial sensing and adhesion. Existing literature provides average filopodial dimensions to be 1-5 μm [20]. From the results in the current study, it may be observed that the dimensions of filopodia are similar on micropatterned surfaces.
Substrate topography can effectively influence cell proliferation, adhesion, and migration [49,26]. A barrier to moving towards cell control via topography is our limited understanding of the fundamentals behind cell-topography interaction. For better comprehension of how cells interact with topographical cues, microgroove substrates were used, with grooves at a scale that could interfere with cell filopodial probing and the establishment and growth of focal adhesions.
FA formation follows filopodial probing. The subsequent extension of the adhesions may be obstructed by the topographic patterns [50]. It may be observed here that filopodial dimensions are long enough to reach the valleys formed by the grooves. This would not preclude the formation of new FAs inside the grooves. However, FAs were primarily formed on the ridges. As observed by Albuschies & Vogel, for high angles between filopodia and adhesion surface (exceeding 12°on glass) FA formation is not favoured [51]. This could be due to the higher normal stress on a filopodial extension when the angle increases, thus disrupting integrin-ligand complex. It is conceivable that filopodia trying to reach the bottom of grooves would be subjected to similar high angles with respect to the adhesion surface, implying the preferred spots of FAs would be limited to ridge tops. The topographic cues could also ensure their presence in specific areas on the substrate.
Among the fabricated patterns, those with the 5 μm grooves were more favourable for formation of FAs on the ridges. This is likely due to the smaller aspect ratio of the patterns in this particular substrate. In concurrence with previous reports, the ordered topographical patterns altered cell migration and spread [52,53]. The elongation was highest for the 5 μm pattern substrates and cells displayed better alignment with the pattern direction. Migration direction was in close correspondence with the pattern direction. Lengthened FAs were aligned along the patterns. For the 5 and 10 μm patterned substrates, a fraction of the FAs were directed normal to the pattern instead of being parallel. FA behavior was remarkably different for the 20 μm pattern substrate. They were isolated, occupied ridge or groove patterns only partially, and the connected actin fibers were more like uncontracted dorsal stress fibers 53].
Transverse adhesions were possible on the substrate with 5 μm dimension patterns. When an FA is constrained due to geometry and is not able to grow in proportion to the tensile forces, the resistance reduces in the transverse direction to compensate [54]. This implies that adhesions along the transverse direction are more likely to unravel. Adhesions that can elongate along the ridge are more likely to be stable. Hence, this leads to cell elongation and increased migration along the direction of the patterns. Thus, the results suggest that surface topography can be used to direct cell culture migration, i.e., patterns at the micro-scale influenced the whole cell sheet cell growth [55].
Increased ALDH3A1, keratocan, decorin and lumican expression by cells on the substrates with the microgrooves supports the idea that aligned topographical cues promote a keratocyte phenotype. This agrees with previous studies that show alignment can up-regulate the expression of keratocyte related genes [56,57]. Increases were also noted for the gene expression of collagen types I, III and V. These collagens make up a significant proportion of the corneal ECM. Collagen type I is the most abundant collagen found in the stroma, mostly in the fibrils. Interactions between collagens I and V result in heteropolymeric fibril formation and regulation of fibril diameter [24]. While collagen III is present in the healthy cornea, it is normally up regulated in keratocytes after injury [58]. This would seem to imply that keratocytes were partially activated to promote wound healing by the microgrooves but were still exhibiting markers associated with a quiescent keratocyte.
Corneal transparency is the results of the highly structured stromal ECM and the presence of intracellular crystallins in the keratocytes [47]. ALDH3A1 is one of the more prominent crystallin proteins expressed by keratocytes. Its expression was up regulated by cells on micropatterned surfaces, with expression levels increasing as pattern size reduced. In the non-patterned substrates, only low levels of mRNA expression were noted. The topographic cues played a role in regulating ALDH3A1 expression. The cytoskeletal reorganization caused by the surface microstructure likely led to the alteration in gene expression [24].

Conclusion
The current work was able to demonstrate that fundamental cell behaviors of keratocytesshape, alignment, and migrationcould be guided by just the environmental topography. The topographical cues also affected cell and nuclear morphology as well as collective cell growth. Cell morphology and nucleus deformation as a result of topographical cues was accompanied by significant cytoskeletal reorganization and chromatin condensation. As well as providing insight into how keratocytes are regulated by their microenvironment, these findings could be applied to assist in controlling how the cells behave when engineering corneal stromal tissue in vitro.

Author contribution statement
• Promita Bhattacharjee conducted experiments, analyzed data and prepared the manuscript.
• Brenton L Cavanagh conducted live cell imaging. • Mark Ahearne planned and supervised the project and corrected the manuscript for publication.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fig. 8. Fold change gene expression analysis of ECM markers COLI, COLIII, COLV, corneal keratocyte-specific markers KERA, ALDH3A1, LUM and DCN and myofibroblasts specific marker αSMA (ACTA2). Quantification done by rtPCR (n(= 3) ± SD). ***p < 0.001, **p < 0.01 and *p < 0.05, One-way ANOVA, followed by Tukey's Honest significant difference test.