Designing of ultrathin patterned substrate with a topographically/chemically defined region
Figure 1a presents a schematic illustration of the sequential process to produce an rGO-based patterned substrate with a topographically defined region. At the first step, abundant surface hydroxyl groups were introduced to a flat glass substrate through a hydroxylation process using piranha solution, and a self-assembled monolayer (3-aminopropyltriethoxysilane, ATPES) with terminated amine groups (-NH2) was subsequently formed onto the hydroxylated glass surface (see Figure S1 in Supporting Information). A well-dispersed GO suspension (c = 2 mg ml−1) was injected in a restricted geometry composed of the upper blade and lower substrate (i.e., APTES-modified glass substrate) positioned at a 30° angle. The capillary-held GO meniscus was naturally formed at the bridge of the fixed gap between the upper blade and the substrate (~100 µm). In this unique geometry, the motor-driven translation stage connected to the lower substrate was traveled back-and-forth repetitively along the programmed route at a constant velocity of 5 mm s−1. As the trapped GO meniscus passes over the substrate, the capillary-induced flow tends to migrate to the GO sheets toward the contact line of the meniscus, and the individual GO sheets are deposited uniformly after consecutive solvent evaporation at the contact line of the GO meniscus (inset image in Figure 1a) [34]. During this process, the planar stacking GO sheets, containing the carboxylic (-COOH) group at the edges and basal planes, were bonded covalently with the terminal amino groups of the APTES monolayer on a glass substrate [35]. Next, the thermal annealing step was performed at 200 ℃ for 8 h to strengthen the CO-NH linkage in the GO-APTES coupling by eliminating the oxygen groups of the GO basal plane. At this stage, GO was transformed to the rGO form [36]. The alternately patterned surface area was defined by conventional photolithography on the rGO film/glass substrate using a mask-aligner with a photomask of micron line/space pattern in a gradient configuration. The exposed GO regions (i.e., unprotected by photoresist) were removed completely using O2 plasma. Finally, the photoresist was stripped using a resist remover, leaving behind the patterned rGO films on the glass substrate. The boundary between the rGO and glass was measured by atomic force microscopy, which revealed a sharp contrast at the edge as presented in Figure 1b. The thickness and surface roughness were also extracted from the cross-sectional height profiles, in which the thickness of the rGO stripe was identified with the ultrathin features of ~10 nm-thick over the patterned surface areas with relatively low surface roughness of ~2.5-3.2 nm (root-mean-square, RMS value). In particular, the rGO stripes (i.e., line/space pattern) produced via flow-enabled self-assembly and the lithographic pattern-transfer process were yielded an ultrathin film of stacked configuration that topologically separated on the planar structure in a single substrate. Figure 1c shows a representative optical micrograph of a gradient and periodically patterned rGO surface with a topographically defined as line-width of 10 to 200 µm, where the clear contrast of the stripes with different local densities of the micropatterned rGO films (bright) and glass (dark) was appeared.
This design is critical in our experimental concept because the trend in the alignment of the cultured MSCs can be affected by the microenvironment provided and the intrinsic size of the individual cells in differentiation, as previously reported [37]. Therefore, this topologically gradient patterned surface as a cell-culture substrate was perfectly fit for the systematic studies in the present experimental scheme under in situ observation to uncover the focal adhesion of the cells and relevant migration responses. Prior to culture the stem cells on the rGO patterned substrate, we thoroughly examined the structural characteristics of the rGO films on a glass substrate because the physicochemical property of the substratum is one of the key parameters to control the stem cell behaviors. First, the collected Raman spectrum showed two main peaks, which were assigned to the D band (1328cm−1) and G band (1569cm−1) as presented in Figure S2. The D band denotes the sp3 structural defects in the rGO basal plane. In contrast, the G band is related to the restoration of the sp2-bonded carbon lattice by eliminating oxygen-functional groups. In the graph, the G band showed higher intensity than that of the D band, suggesting that the reduction to GO proceeded through the structural restoration of the sp2-bonded carbon lattice due to the dissociation of the oxygen in the GO domain during annealing [38]. In addition, the optical properties of the transmittance were evaluated by UV-Vis spectroscopy over the wavelength range, 400 to 800 nm, for the rGO film and patterned rGO thin film as shown in Figure S3. The patterned rGO thin film exhibited a high optical transmittance above 92% at 550 nm through the O2 plasma etched opening, compared to the rGO thin film on a glass substrate (T = 83.7% at 550 nm wavelength). This clear view field is beneficial to the observation of cell migration.
To identify the details of surface chemistry generated by the sequential fabrication process, we examined the rGO/glass regions quantitatively using X-ray photoelectron spectroscopy (XPS), as summarized in Figure 1d-f. As designed, the manipulated layered surface can be considered rGO/APTES/glass and O2 plasma-exposed APTES/glass in the molecular definition. For the rGO/APTES/glass region (Figure 1d), the C1s spectra identified a series of C-C, C–N, C-O, and COOR species at 284.8, 285.6, 286.4 eV, and 288.4 eV, respectively; this result is consistent with the chemical structure model of rGO [39]. The distinct component area of the C-N peak at BE = 285.6 eV originated from the internal bonding of the carboxylic group in the rGO basal plane, the amine groups, and the presence of the unbound terminal groups of APTES. A major component of the C-N peak similarly appeared in the APTES/glass region that could be attributed to the oxidation reaction of amine head groups unbound with GO sheets on the glass surface during the O2 plasma process, but the slightly increased C-O peak was detected as a result of the destruction of amine groups and the subsequent substitution of hydroxyl groups [40]. Indeed, the impact of O2 plasma was evaluated by the direct comparison to the survey of C1s spectrum for the originally provided APTES-modified glass surface, revealing the C-N component peak (Figure. S4a). On the other hand, for the N1s XPS spectra, HNO=C at BE = 400.3 eV was dominant in the rGO region, confirming the existence of NH2 terminal group (BE = 399.5 eV) and weak hydrogen bond/protonated amines (-NH2/NH3+, BE = 401.8 eV) as shown in Figure 1f. The major component peak represents an amide formation (HN=OC) on the APTES/glass region after the O2 plasma process. Hence, the APTES/glass regions configured with rGO/APTES/glass were only optimized with copious free amines before the O2 plasma process (Figure S4b). This supports the interpretation of the C1s spectrum associated with the amide group. For more information, as summarized in Table S1, the intensity ratios of the oxygen-containing groups to the graphite component were fully evaluated to quantify the effective charge compensation in each sample. Remarkably, the O2 plasma-exposed glass regions contain a higher hydroxyl carbon content than the photoresist-blocked rGO and as-prepared APTES-modified glass regions, which suggests that the silanolated surfaces are saturated with a high degree of oxidation. Moreover, the increased level of carboxyl carbon content (HNC=O) also indicates the silanol and oxidized amine terminated surface. Finally, physically and chemically defined cell culture substrates were characterized with rGO/glass arrays of gradient micro-widths and equal heights to study how cells form adhesions and spread through them.
MSC culture on the chemophysically defined biointerfaces
The selective surface recognition and cell-specific responses to the patterned microenvironment were examined by culturing MSCs separately on the prepared substrates: glass, rGO, and the alternately patterned rGO stripes on the glass. No significant differences in cell adhesion and proliferation were observed on the glass and rGO film/glass without patterns, as shown in Figure S5a and S5b. However, unexpected cell responses on the alternately patterned substrates were exerted as presented in Figure 2a. In the anisotropic patterned cell substrate, a highly aligned configuration of MSCs was evaluated overall in the entire area. Specifically, the morphological evolutions of MSCs was obviously appeared by the extraneous surface-initiated cues in an aligned features with gradient cell densities on the glass regions adjacent to the rGO stripes. With a careful observations presented in Figure 2b and S6, the highly aligned and bundled MSCs were densely connected by clustering in the range of 100 µm or more on the glass stripes, whereas the cells were distributed as a single by end-to-end cell configuration within the patterned area of the 40 µm. Furthermore, an effective movement of the seeded MSCs on the glass/rGO substrate was monitored in real-time and traced for the direct assessment of cell behavior in a controlled environment as captured in Figure 2c (see Supplemental Movie and Figure S7). Notably, the individual single motile cells resided only on the glass stripes and interconnected with each other, guided by the adjacent edges of the rGO stripes. From a macroscopic perspective, most of the cells were polarized, and they stretched their morphologies biaxially along with the parallel directions of the patterned glass stripes by actively sensing the exposed surfaces. This observation is critically important because the morphological changes to stem cells caused by contact guidance can alter their cellular responses associated with the differentiation, in which the alignment of the cells is essential. In mechanobiology, control of the dynamic organization of large protein complexes between actin filaments and integrins (i.e., focal adhesions) can be one of the main parameters to examine the physicochemical cues for the cell functions, related to the intercellular tension and stress in the actin cytoskeleton [41–43]. Within a given unique microenvironment, Figure 2d depicts a main conceptual schematic diagram of cell alignment and elongation, that is, the surface-mediated cell adhesion and migration of the MSCs.
In fact, at the initial stages of the experiment, we hypothesized that the isometric rGO/glass stripes might induce morphological reconfiguration of the stem cells during culture, which could be effective interfacial guidance on the discriminated cellular behavior, such as alignment, orientation, and cell-cell communication [44–47]. Recent studies reported that the cells cultured on graphene-activated substrates exert favorable cell-substrate interactions through the development of focal adhesion sites [48–50]. This suggests that the hetero-electrostatic interactions or inherent nanoscale topographies of graphene surface can be synergistic with the help of patterning strategies for the cell behavior rather than the fully covered graphene thin-film substrate by promoting migration and encouraging cell-cell communication [13, 21]. Based on these firmly established assessments of cell cultures, we assumed that a delicate control of the anisotropic microenvironments from the favorable cues from the ultrathin patterned rGO films spatially provide distinctive cell interactions on each surface region. However, in this experimental result, graphene-repellent behaviors of the MSCs contrast with the existed contact guidance of graphene surface for living cell assembly [17, 45]. This implies the predictable design of the tissue scaffold for the specified stem cells (e.g., MSC) is difficult and controversial within the limited understanding involved in the precise mechanisms of the cell-repellent behaviors, responsible for the acute cellular responses on each provided microenvironment [51]. Therefore, we investigated the levels of the surface-potential distributions on the topological culture environment to understand one possible consideration of the heterogeneous electrostatic interactions in patterned rGO/glass arrays that could trigger unprecedented cellular responses.
Figure 2e presents magnified optical micrographs for the defined surface area of rGO/glass stripes with 100 µm line-intervals. As seen in this configuration, the boundary between the rGO and glass was geometrically isolated with sharp contrast. In this structure, to scrutinize the prepared biophysical interfaces accurately the potential distribution of the surface charge across the glass/rGO stripes was evaluated by kelvin probe force microscopy (KPFM) with spatial mapping data, as visualized in Figure 2f. It should be acknowledged that the surface charge differences on the glass/rGO regions were clearly examined with a certain potential level (Figure 2g). For example, the height profile for the surface topography of the area showed a higher magnitude in the rGO regions (~10 nm), but the corresponding distribution of surface potential was ~144 mV more negative than the value on the glass region (bottom panel in Figure 2f and S8). The statistical data distributed over each segmented region (i.e., glass and rGO surface) indicates obviously separated and relatively uniform locational average charge potential levels of -263.6 mV and -392.2 mV on the glass and rGO region, respectively (Figure 2h and S9).
So far, such regional physicochemical differences in the cell culture substrate that is produced by the combining construction of the nanoscale materials and chemical modiଁcations were well suited to cell culture analysis, and the critical surface properties were precisely evaluated by XPS and KPFM analysis for the patterned rGO/glass array. Figure 3a schematically describes the possible principle of the surface-mediated assembly of the MSCs based on the directional cell guidance effect and preferential adhesions. In the sequential process to prepare the patterned rGO/glass array, a strong negative charged surface was maintained at the rGO regions (i.e., ~392.2 mV) due to oxygen-related functional groups such as hydroxyl, carbonyl, and carboxyl groups. In contrast, the glass surface was subtly complexed with oxidized amide and silanol groups when exposed to the mild O2 plasma process. On the copious silanol-functionalized glass surfaces, the oxidized amide groups mainly contribute to mediate a surface potential level, introducing a relatively less negatively charged surface (i.e., ~263.3 mV) compared to the rGO regions, which leads to a fundamental heterogeneity of the patterned surfaces. Hence, the heterogeneity in each separated potential level provided selective recognition sites to the MSCs and guide them to reside on the favorable silanol and amine functioned surfaces, which is obviously in contrast to the case of the random orientated cell configurations on the unpatterned planar cell substrate [52, 53]. Thus, it can be inferred that, in our experimental system, the delicate interactions of the cells on the patterned substrates contributed more importantly to controlling cell behavior than other parameters due to the anisotropic surface potential difference.
Consequentially, the morphological features of MSCs adhered were observed by labeling the cytoplasmic microtubule (i.e., tubulin) and stress fiber (i.e., F-actin) by immunofluorescent staining and merging the images. The cells exhibited the highly confined cell alignment with enhanced focal adhesions (Figure 3b). In particular, the microtubules were organized only within the glass stripes by the presence of the cell-repellent rGO interfaces. Clear evidence was presented by highly magnified micrographs in the marked area of i) and ii) in Figure 3b. On the 100 µm patterns, multiple cells were connected by contacting each other, and the actin filaments were fully stretched only on top of the glass stripes as contact guidance to confine the MSCs to prevent crossover of the rGO regions (Figure 3c and S10). As a result, the restricted cell migration and attachment were analyzed statistically with characteristic features, as summarized in Figure 3d-f. We focused on the cell capture capabilities for the representative pattern spacing (i.e., 40 and 100 µm), which verified the morphological evolutions of the MSCs with highly aligned cell colonies. Note that the effective local guidance for MSCs was more significant than the control culture on a flat glass substrate (i.e., absence of patterned rGO arrays), which displayed a universal random growth without a preferential orientation (Figure S11). Compared to the randomly seeded MSCs, the increased cell length was observed on the 40 µm pattern-spacing substrate up to ~500 µm with self-confined geometrical factors upon organization (Figure 3d and S12). The aspect ratio, dividing the length by the width of each cell, also increased significantly in both 40 and 100 µm pattern spacing, following the directional guidance of the glass stripes (Figure 3e). Comparable to this data set, the orientation of the cell alignment was generally close to 90° with a narrow distribution on the prepared patterned substrates (Figure 3f); 90° denotes a parallel position on the axis of the stripes, and 0° represents perpendicular alignment (Figure S13). A more extensive study of the specific features on the cell attachment and selective arrangements were surveyed in detail, as shown in Figure S14 and S15, in which a cross-patterned glass surrounded by ultrathin rGO film was used to observe the cell migration toward the active target area (i.e., only exposed glass region) in real-time. Most MSCs migrated to a cross-shaped glass region (Figure S14) and stretched their morphologies within the restricted area or aligned sharply at the circular borderline between the glass and enclosed rGO film stretching the F-actin fibers (Figure S15). This set of results also demonstrated that MSCs prefer to bind on the favorable glass surface, directly compared to the rGO film surface.
Spontaneous surface-induced differentiation of MSCs to SMCs
From a stem cell functionalization point of view, our presented work could be effectively utilized to constitute a form of sheets of stranded SMCs in a highly aligned configuration. For example, as conceptually presented in Figure 4a, the diameter of blood vessels is controlled by the highly oriented SMCs to regulate blood flow and pressure. Generally in the organ parts, SMCs are longitudinally or circumferentially organized around the internal tissue layers of the visceral organs that are mainly responsible for the proper contraction and relaxation of the organs in gastrointestinal, cardiovascular, urinary, and reproductive systems [54]. Therefore, the control of cellular orientation could be one of the important issues in engineering stem cell application, especially for SMCs. As presented earlier, the MSCs rely on anisotropic contact guidance, exhibiting an elongated cytoskeletal structure located mainly on the specified micropatterns. Notably, spontaneous “cell-to-cell contact” can be inevitable in a confined microenvironment (Figure 4b), resulting in cytoskeletal remodeling caused by F-actin assembly [55–58]. This restricted condition may cause cell differentiation by activating YAP/TAZ, which is the key transcription factor of the Hippo signaling pathway, as illustrated in Figure 4b [59, 60]. With the encouraging results thus far and accumulating evidence that MSCs can differentiate into SMCs [28, 61, 62], the effect of defined guidance cues from the rGO stripe pattern was maximized by synchronizing MSC differentiation into SMCs as a proof-of-concept model (Figure S16).
In addition, the effects of the substrate as physicochemical guidance for the differentiation of MSCs into SMCs were examined by culturing the MSCs on two different rGO pattern-spacings (i.e., 40 and 100 µm) and flat glass substrates under serum-free conditions or with a TGF-β1 treatment as an agonist inducing the differentiation of MSCs to SMCs. The serum-deprived MSCs were aligned directionally on the anisotropic patterned cell substrate until day-6 of incubation to induce the gradual development of the stress fibers (Figure 4c and S17). The MSCs were arranged strictly in a unidirectional alignment by the anisotropic chemo-physical cues, where the markedly aligned arrays of the quiescent MSCs could be observed in the overlapped configuration after the six days of culture. As shown in Figure 4d, the cells exhibited the stretched phenotypes on the pattern-spacing of 40 and 100 µm, unlike the control groups (i.e., cultured on unpatterned substrates). By initially driven acute contact guidance, the quiescent MSCs expressed spontaneous morphological features with a highly aligned configuration (Figure 4e and S18), maintaining cellular integrity during cell cultivation. The bundling of MSCs with cell-to-cell contacts may reflect the morphological characteristics of the SMCs. Furthermore, the directional guidance with unidirectional alignment increased the cell-to-cell contact area between the cells dramatically during the TGF-β1-induced differentiation of MSCs to SMCs on the patterned cell substrate (Figure 4f). Thus, cell-to-cell contacts could be induced by increasing the pattern spacing with an optimal range of 40-100 µm (Figure S19a). Accordingly, the cell alignment and orientation were unaffected by the TGF-β1 treatment during differentiation to SMCs (Figure S19b). The relevant differentiation capacity was uniformly retained, which was visualized as calponin and SM22α marker expression in Figure S19c.
The influence of the microenvironment on differentiation of the quiescent MSCs to SMCs was examined by measuring the expression levels of myogenic markers by western blot analysis, as shown in Figure 4g-j. Intriguingly, the expression of SMC markers, including α-smooth muscle actin (α-SMA), calponin, and SM22α, was more increased at the 100 µm pattern-spacing than the other cases (i.e., unpatterned or 40 µm pattern-spacing), suggesting that broader cell-clustering accelerates the differentiation of MSCs to SMCs, as shown in Figure 4g and 4h. Furthermore, smooth muscle tissue consists of multiple SMCs connected through connexins stimulated in a synchronous pattern. The transmembrane protein (i.e., N-cadherin) mediates cell-cell adhesion in multiple tissues [57, 58, 63, 64]. As shown in Figure 4i and 4j, the expression of N-cadherin and connexin 43 also increased in the 100 µm pattern spacing. The cell shape can influence Hippo signaling and mechanical tension transmitted through cell-cell junctions and cell-matrix adhesions [59, 60]. Therefore, the expression of YAP and TAZ was determined by Western blotting. As shown in Figure S20, TAZ expression in the 40 µm pattern-spacing increased more than that in the unpatterned substrate and was further augmented in 100 µm pattern-spacing consistent with an increase in α-SMA expression. In contrast, YAP was expressed at relatively low levels with no significant differences. TAZ is likely to have possible implications in the myogenic differentiation of MSCs on a chemophysically defined cell substrate because it plays a vital role in regulating myogenesis and cytoskeleton organization promoting the phenotype of quiescent SMC [62, 65]. By facilitating the arrays of ultrathin rGO patterns on a glass substrate, the present study demonstrated that the cell-culture platform, with anisotropic features in the surface structure, induced phenotypic conversion by modulating the adhesion, migration, and differentiation of human MSCs.