Recent advances in nanotechnology present synthetic bio-inspired materials to create new controllable microenvironments for stem cell growth, which have allowed directed differentiation into specific lineages.^[1,2] Carbon nanotubes (CNTs), one of the most extensively studied nanomaterials, can provide a favorable extracellular environment for intimate cell adhesion due to their similar dimension to collagen. It has been shown that CNTs support the attachment and growth of adult stem cells^[3-6] and progenitor cells including osteoblasts and myoblasts.^[7,8] In addition, surface-functionalized CNTs provide new opportunities in controlling cell growth. Surface functionalization improves the attachment of biomolecules, such as proteins, DNA, and aptamers, to CNTs.^[9] Zanello et al. cultured osteoblasts on CNTs with various functional groups and showed reduced cell growth on positively charged CNTs.^[10] Recent reports have shown that human mesenchymal stem cells (hMSCs) formed focal adhesions and grew well on single-walled CNTs (swCNTs).^[5,6] However, the effect of naïve swCNT substrates on the differentiation of stem cells has not been reported before. Herein, we report the osteogenic differentiation of hMSCs induced by swCNT monolayer cues without any chemical treatments. Interestingly, the surface treatment of swCNTs via oxygen plasma showed synergistic effects on the differentiation as well as the adhesion of hMSCs. The stress due to the enhanced cell spreading on swCNT layers was proposed as a possible explanation for the enhanced osteogenesis of hMSCs on the swCNT monolayers. Previous reports showed that the stress to stretch stem cells on microscale molecular patterns generated the tension on actin filaments, which eventually enhanced the osteogenesis.^[11,12] Since our method relies on monolayer coating of swCNTs, it can be applied to a wide range of substrates including conventional scaffolds without any complicated fabrication processes.

Figure 1 shows a schematic diagram depicting our experimental procedure. In this experiment, three substrates were used: glass as a control, a pristine swCNT monolayer adsorbed on a glass surface, and an oxygen-plasma-treated swCNT (O-swCNT) monolayer on a glass surface (Figure S1a in the Supporting Information (SI)).^[13] swCNTs had the average diameter of approximately 1–2 nm, and their average length was 1.5 μm. When the cleaned cover slips were dipped in a dispersed swCNT solution, swCNTs were adsorbed onto the glass surface to form a monolayer. Oxygen plasma treatment was performed to modify the surface chemistry of swCNTs, which is known to generate hydroxyl or carboxyl groups on the surface.^[14,15] Atomic force microscopy (AFM) analysis revealed that the swCNT layer maintained its surface roughness even after 40 s of the oxygen plasma treatment, while its contact angle decreased abruptly after 20 s of exposure to oxygen plasma (Figure S1b in the SI). Our control experiments show that, as the time for plasma treatment increased, the abrupt change of contact angle or surface roughness occurred at around 20 or 40 s, respectively. Due to such an abrupt transition, the properties of swCNT layers treated with oxygen plasma for 20 or 40 s usually exhibited rather large variations, which reduced the reproducibility of the following stem cell growth experiments. On the other hand, we could reproducibly obtain similar surface properties for the swCNT layers after 30 s of plasma treatment, thus enabling reliable stem cell growth experiments. Consequently, in our experiment, we utilized swCNT monolayers treated with oxygen plasma for 30 s, which reproducibly exhibited hydrophilic properties while maintaining the roughness of pristine swCNT monolayers. The hMSCs from bone marrow were seeded on these substrates, and then their adhesion, proliferation, and differentiation were examined.

Figure 2 shows the adhesion and proliferation of hMSCs cultured on glass (Figure 2a), a swCNT monolayer (Figure 2b), and an O-swCNT monolayer (Figure 2c). Their actin filaments and nuclei were visualized after 24 h from seeding. It is notable that hMSCs spread wider, and their actin fibers look thicker on swCNT monolayers than on glass substrates. For quantitative analysis of cell adhesion, the area of 200 individual cells on each substrate was estimated from that of stained actin fibers (see Figure S2 in the SI) and utilized to calculate the average area per cell (Figure 2d). The results show that the adherent area of individual hMSCs is higher on swCNT or O-swCNT monolayers compared with that on the glass substrate. At the same time, the ratio of long and short axial lengths of the hMSCs is smaller on swCNT monolayers and much smaller on O-swCNT monolayers than on glass substrates (Figure 2e). The MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) assay results show that the proliferation of hMSCs was enhanced on the O-swCNT substrate (Figure 2f).

One possible explanation regarding the enhanced cell area on the swCNT substrate can be the nanoscale roughness of the swCNT monolayers. Zhang et al. reported that nanoscale surface roughness of swCNTs can deform the cell membrane and hinder the motion of vesicles inside cells.^[16] This membrane deformation may affect the distribution and diffusion of membrane proteins including focal adhesion proteins, which are critical factors in cell adhesion. Tay et al. showed increased number of focal adhesion proteins in hMSCs cultured on an innate swCNT film.^[6] Other researchers also have observed wider adhesion of various cells on swCNT substrates than on conventional substrates such as glass. Osteoblast-like cells and hMSCs spread better on a swCNT film than on glass substrates, resulting in the larger cell area and higher occurrence of filopodia at the cell boundaries.^[6,17]

Besides the effects of nanoscale surface roughness,^[16] our results also indicate that surface chemistry plays a role in cellular interaction between cells and swCNTs. Note that even though our swCNT monolayers with or without oxygen plasma treatment had similar surface roughness (Figure S1 in the SI), hMSCs on O-swCNT substrates exhibited an enhanced cell area and proliferation compared with those on pristine swCNT substrates (Figure 2). Presumably, the chemical changes of the O-swCNT layer, such as enhanced hydrophilicity and surface oxygen content, increased the proliferation and adhesion of hMSCs on it.^[18].

The morphological change of hMSCs has been reported to be related to their capacity for multipotentiality. For example, spindle-shaped hMSCs have high potential for adipogenesis, while flat hMSCs have high potential for osteogenesis.^[19,20] Similarly, the actin cytoskeleton changed from thin and parallel microfilament bundles to thick and crisscrossed bundles under the osteogenic differentiation process, which is consistent with Figure 2b,c.^[21,22] This implies that the swCNT substrates could enhance the osteogenesis of hMSCs.

In order to check the osteogenic induction by the swCNT monolayer, the osteogenic proteins and corresponding genes were tested. Core binding factor alpha1 (CBFA1), osteocalcin (OCN), and alkaline phosphatase (ALP) were used as osteogenic markers. CBFA1 is the main transcription factor for committing hMSCs to the osteoblastic lineage; OCN is a differentiated osteoblast-specific gene for mineralizing the bone matrix, and ALP is an early osteoblastic marker.^[23,24] hMSCs were cultured on three different substrates with and without osteogenic induction media for 17 days. Immunostaining was performed at day 12, and the quantitative gene analysis was performed at day 7 and 14 from seeding.

Figure 3a shows that the hMSCs filled the whole area on all substrates at day 12. Also, note that the cells cultured in osteogenic induction media were slightly detached from the glass due to strong cell–cell interactions. The detachment was retarded on swCNT substrates (data not shown). The osteogenic protein OCN was detected and visualized with fluorescent dyes (Figure 3b). The hMSCs grown on swCNT substrates exhibited brighter OCN immunofluorescence than those on glass substrates, but it was not statistically significant (Figure S3 in the SI). However, the hMSCs grown on O-swCNT substrates definitely exhibited brighter immunofluorescence than those on glass substrates, and quantitative statistical analysis showed much more brightness on O-swCNT substrates. This high protein expression was confirmed in mRNA expression. The mRNA of beta-actin was incorporated as an endogenous housekeeping gene for all the test substrates, and the relative transcription level to that of the hMSCs cultured on glass is shown in Figure 3c. After 14 days of culture, the osteogenic genes such as CBFA1 and OCN were upregulated on both swCNT and O-swCNT substrates.^[24] In the case of ALP, the expression was enhanced only on the O-swCNT substrates in the early days of culture (Figure 3c; Figure S4 in the SI).

The brighter fluorescence and enhanced mRNA expression indicate the enhanced commitment of hMSCs on O-swCNT monolayers for osteogenesis without any differentiation-inducing chemicals. Although the enhanced osteogenic function of osteoblast on CNTs have been reported previously,^[25,26] this is the first report regarding the enhanced commitment of hMSCs to osteoblast lineage on a CNT monolayer.

A possible explanation for the enhanced osteogenesis can be the stress on cells due to the enhanced cell spreading on the swCNT monolayers. It was reported that the stress to stretch a stem cell generated the tension on actin filaments and eventually enhanced the osteogenesis.^[11,12,27] In order to verify this hypothesis, we fabricated differently sized swCNT square patterns. The size of the patterns was determined considering a previous report utilizing 12 μm extracellular matrix (ECM) molecular patterns to enhance adipogenesis and 100 μm patterns to osteogenesis.^[11] We fabricated patterns larger than 100 μm to see only the effect of enhanced spreading on osteogenesis. Figure 4a shows the bright field images of hMSCs cultured for 12 days. The hydrophobic molecules efficiently blocked the hMSC adhesion, resulting in the confined growth of hMSCs only in the swCNT regions.^[3] A closer look at cell morphologies showed that the nuclei were located at the center of the pattern, and cytoplasm was extended to the boundaries. Actin fibers were anchored and stretched to the focal adhesions located at the boundaries (Figure S5 in the SI). This indicates that the hMSCs tended to spread out to the boundaries of swCNT patterns using focal adhesion, and thus actin fibers were stretched as the square size increased.

The DAPI stained nuclei count 2, 13, 17, and 17 for 100, 200, 300, and 400 μm square patterns, respectively (Figure 4b). The spreading area per cell was calculated by dividing the actin stained area by the cell number. Data from 20 square swCNT patterns were plotted in Figure 4d. As the pattern size increased, the cell density decreased and the averaged area per cell increased. This implies that the hMSCs in larger swCNT patterns were spread wider. The OCN immunofluorescence value per single cell was averaged from nine of each swCNT pattern size (Figure 4c), and the values were plotted in Figure 4e. The results indicate that hMSCs on larger square patterns exhibited easier commitment to the osteogenic lineage. This is consistent with our hypothesis that larger cell spreading in swCNT patterns enhanced the osteogenesis.

As previously reported, the enhanced adhesion might be linked to Rho-family GTPase (guanine triphosphatase) signaling and nonmuscle myosin contraction within the cell. They showed that the overexpression of ras homolog gene family member A (RhoA) or Rho-associated coiled-coil-containing protein kinase 1 (Rock1) stimulated myosin contraction and promoted osteogenesis.^[11] Microarray analysis and pathway inhibition studies should be followed to elucidate a more detailed mechanism. Furthermore, the effect of plasma treatment on swCNT should be studied systemically. As the plasma treatment was known to enhance protein adsorption onto swCNT surfaces, it might have an influence on the cellular interaction with the nanostructured surface. Interestingly, surface chemical modulation with −OH and −COOH groups itself was reported to downregulate the osteogenesis.^[28] This implies that there is a synergistic effect of both surface roughness and chemistry on O-swCNT substrates.

In summary, we showed that the osteogenic differentiation of hMSC was promoted from swCNT monolayers without differentiation-inducing media. In addition, we found that simple oxygen plasma treatment amplifies the adhesion, proliferation, and even osteogenic differentiation of hMSCs by adding chemical effects on the main topographical effects. The stress on hMSCs due to the enhanced cell spreading on swCNT monolayers was proposed as a possible explanation. However, a more detailed mechanism, involving such details as the signal pathways in the stem cells, is not yet clear and requires further study. In any case, this work suggests that the swCNTs can be a powerful scaffold in mesenchymal stem cell engineering.