Titanium nanotubes induce osteogenic differentiation through the FAK/RhoA/YAP cascade

Abstract

Clarifying mechanisms underlying various nanotopography-mediated changes in adherent cell function is requisite to understanding and improving the tissue–alloplastic interface for clinical applications. In this study, pure titanium foils and implants were modified with titanium nanotubes (TNTs) to verify their osteogenic abilities both in vitro and in vivo. The intracellular molecular dynamics of the FAK/RhoA/YAP cascade were assessed as a function of TNTs nanotopography. MC3T3-E1 cells adherent to large diameter TNTs presented limited cell spreading, reduced focal adhesions (FAs), and increased osteogenic differentiation, compared with those adherent to smooth Ti surfaces. Significant attenuation of FAK recruitment and RhoA activity was observed in TNTs adherent cells, and YAP localization moved from the nucleus to the cytoplasm. Moreover, YAP overexpression in MC3T3-E1 cells prevented TNTs-induced osteogenic differentiation. The demonstrated relationship between TNTs-mediated spreading behaviors, altered signaling through FAK and RhoA and YAP translocation suggests that nanotopographic cues influence osteoblastic differentiation in a YAP-dependent manner.

---

Introduction

Attached cells are inherently sensitive to extracellular matrix (ECM) substrate topography in various cultured conditions.^1–3 The surface topography features of biomaterials for bone tissue engineering play a pivotal role in regulating osseointegration by affecting cell adhesion, morphology, proliferation and differentiation of osteogenitor cells.^4,5 Titanium (Ti) can be modified with nanotubes fabricated through electrochemical oxidation methods. Due to the nanoscale dimension and physicochemical properties, Ti nanotubes (TNTs) mimic the hierarchical structures of natural osteogenic tissues. TNTs promote mesenchymal stem cell differentiation into osteoblasts.^6–9 However, the detailed biomolecular signaling through which TNTs affect cellular behavior and tissue responses is still unclear. In the present study, the mechanism behind cell fate regulated by nanotopographic TNTs cues was investigated at the level of signaling via YAP.

The mechanical stimulus sensed by cells is “translated” into biochemical signals during their interaction with the ECM. Focal adhesions (FAs) serve as an initiation point for the transition from mechanical force to biological signals. These large, dynamic cytoplasmic complexes form at the ECM-bound integrin receptor as a central point for signaling.¹⁰ Thus Focal adhesion kinase (FAK) is one mediator of downstream signaling reportedly involved in mechanotransduction and acts synergistically with GTPase proteins to regulate osteogenic differentiation in stem cells.^11–13 Ras homolog gene family member A (RhoA) is a small GTPase protein able to sense and respond to mechanical cues.¹⁴ RhoA activation regulates the synthesis of actin bundles, stress fibers, and tensile actomyosin structures.^13,15–18 RhoA activity can also be influenced by FAK in response to mechanical stimulation, again controlling the polymerization of the cytoskeleton to transmit signals to the nucleus and regulate gene transcription.^15,19,20 While each of these pathways are implicated in the signaling from alloplastic nanoscale features,²¹ further details of nanoscale cueing of mesenchymal stem cell differentiation remain unexplored.

More recently, mechanosensing has been studied at the level of Yes-associated protein (YAP) and its control of gene regulation. Dupont et al. described a role for YAP and TAZ (transcriptional coactivator with PDZ-binding motif) in relaying mechanical signals to the nucleus.²² YAP activation is associated with its nuclear localization and has been shown to induced osteogenic gene expression.²³ This process of nuclear localization is, partly, regulated by cell–ECM interactions and particularly is activated by cell spreading.²⁴ Because RhoA is often central to the process of cytoskeletal remodeling during spreading, the inhibition of RhoA activity was found to maintain the cytoplasmic location of YAP, whereas activated RhoA led to the nuclear translocation of YAP and the induction of gene expression changes.¹⁵ Because nanotopography is able to modulate cell spreading, the cytoskeleton, and RhoA activity. However, the influence of nanotopography on YES function and osteoblastic differentiation in surface adherent cells has not been considered.

In this study, we hypothesized that the biological effects of nanotopography are exerted via a mechanism involving FAK, RhoA activity, and YAP distribution. For confirmation, we fabricated Ti nanotubes of different diameters on pure Ti foils and implants, and evaluated the osteoconduction or osteoinduction abilities of TNTs both in vitro and in vivo. Afterward, we analyzed the cell functions, FAK recruitment in focal adhesions, RhoA activity, and YAP subcellular localization in MC3T3-E1 cells on different substrates. Our experimental results shed light on the signaling crosstalk within MC3T3-E1 cells growing on TNTs of different diameters, and the role of YAP subcellular localization in nanotopography induced cell responses.

Results and discussion

Topography feature of TNTs and adhered MC3T3-E1 cells

TNTs can emulate the ECM structures and direct stem cell differentiation to a specific lineage, similar to what natural ECM does in vivo. The nanotubes with various diameters apply different effects on the behavior and differentiation of bone precursor cells.²⁵ To investigate how TNTs affect cell differentiation, we fabricated TNTs with two distinct diameters: STNTs (10v) and LTNTs (30v). Smooth Ti was used as the control group. After fabrication, SEM images showed that the outside diameter (OD), inside diameter (ID), and the wall thickness of the nanotubes developed with 10v were measured to be 50 nm, 30 nm, and 10 nm, respectively, and 150 nm, 90 nm, and 30 nm for 30v (Fig. 1A). The cell spreading patterns were different on the distinct surfaces, i.e., the LTNTs limited the cell spreading more than STNTs, and both of them reduced the cell spreading dimension compared with the smooth surface. The leading edge of MC3T3-E1 cells attached on smooth Ti had broad lamellipodia, while typical filopodia were found on the cells cultured on LTNTs (Fig. 1B and S1†). Because of the limited surface area for cell attachment on the nanotubes (Fig. 2A and B), the space for cell spreading will be restricted. The ratio of the available area for cell-substrate adherence (AACA) to the total area on the LTNTs was 31.8% for smooth Ti, while for STNTs, this ratio was 58.8% (Fig. 2C). Fig. S2† shows the schematic diagram of the anodizing process TNT fabrication.

Fig. 1 (A) SEM image of smooth Ti, STNTs, and LTNTs (scale bars = 200 nm); (B) SEM image of adhered MC3T3-E1 cells on different substrates (scale bars = 500 nm).

Fig. 2 (A) The available area for cell-substrate adhesion (AACA) on smooth Ti and TNTs (in red, scale bars = 50 nm). (B) Diagram of cell attachment area on different substrates (in red). (C) The quantification of AACA/total area for each substrate (**p < 0.01 and ***p < 0.001 vs. smooth Ti; ^##p < 0.01 vs. STNTs, n = 6).

Osteogenic differentiation assay

Prior studies indicated that TNTs could induce the osteogenic differentiation of stem cells.²⁶ In the initial cell response to surfaces, the nanotube dimensions play an important role.²⁵ To verify the effect of the Ti surface topography on cell fate, MC3T3-E1 cells adhered on differently fabricated TNTs were cultured for 7 days. Real-time PCR results showed that the TNTs significantly promoted the gene expression of the bone differentiation markers, Runt-related transcription factor 2 (Runx2) and osterix (Osx), compared with flat Ti (Fig. 3). Two- and three-fold increases (p < 0.01) in both Runx2 and Osx expression respectively were observed for cells adherent to STNTs and LTNTs.

Fig. 3 Osteogenic markers Runx2 and Osx expression in MC3T3-E1 cells on different substrates (**p < 0.01 and ***p < 0.001 vs. smooth Ti; ^##p < 0.01 vs. STNTs. n = 4).

To confirm in vivo the osteogenic effects of TNTs, the pure Ti implants modified with large TNTs (TNTI) were engrafted into rat tibiae, and the non-anodized smooth implants (SI) were applied as a control. The SEM images of SI and TNTI surfaces are shown in Fig. 4A. μ-CT analysis at 2 weeks following implant placement indicated that the TNTI implants accrued grated interfacial bone volume and higher trabecular number/mm than SI (Fig. 4B and C). Qualitatively, H&E staining showed a continuous line of new bone formation along the interface between the bone and TNTI, whereas SI groups lacked the new bone formation in the same area (Fig. 4D). Taken together, nanoscale modification of titanium surfaces with titanium nanotubes increased osteoblast differentiation and osseointegration. Larger diameter nanotubes induces a stronger osteogenic response.

Fig. 4 (A) SEM images of implant surfaces before and after anodization: (A1–A4), smooth surface of implant (SI); (A5–A8), LTNTs fabricated on the surface of implant (TNTI). (B) 3D reconstruction performed for SI and TNTI. (C) The average bone volume per total volume (BV/TV) and the trabecular number (Tb. number) analysis (*p < 0.05 vs. smooth Ti. n = 5). (D) Histomorphological analysis by H&E staining (scale bar = 200 μm and 50 μm).

TNTs topography decreased FAK recruitment to FAs in MC3T3-E1

FAK is concentrated near focal adhesions and plays an important role in signaling events involving growth factors, ECM molecules, and stress signals.²⁷ FAK activation has been shown to occur with both outside–in and inside–out mechanical signaling. Phosphorylation at tyrosine residue 397 of FAK (pY397-FAK) is critical for regulation of cell signaling in response to mechanical stimulation.¹⁴ To investigate FAK involvement in TNT regulation of osteogenic differentiation, we seeded MC3T3-E1 cells on the different surfaces. The western blot result (Fig. 5A) showed that both total FAK and pY397-FAK were significantly decreased on LTNTs compared with smooth Ti, while their levels demonstrated relatively less reduction on STNTs. Vinculin, a stably FA-bound structural protein, was immunoprecipitated from cell lysates, and western blotting was used to detect the FAK pulled down in the precipitate. The association of FAK with vinculin on LTNTs also declined to a greater degree than on STNTs (Fig. 5B). Immunofluorescent staining with an anti-vinculin antibody showed that the FAs appeared as obvious green punctate spots at the periphery of the spreading cells on smooth Ti, and few bright FAs were observed in the unspread cells on LTNTs (Fig. 5C, as indicated by yellow arrows). Simultaneously, both the cell size and actual adhesive area were much smaller on LTNTs than on STNTs and smooth Ti (Fig. 5D and S3†). As a result, we concluded that the smaller AACA of TNTs compared with smooth Ti limited the cell spreading and FA formation, and therefore, reduced FAK recruitment to FAs and FAK activity.

Fig. 5 TNTs topography induce total FAK and pY397-FAK decrease. (A) Western blot results of total FAK and FAK Y397 phosphorylation. (B) Densitometric analysis of recruitment of FAK to focal adhesions. Vinculin was immunoprecipitated using an anti-vinculin antibody and blotted for FAK (*p < 0.05 and **p < 0.01 vs. smooth Ti; ^#p < 0.05 vs. STNTs. n = 3). (C) Immunofluorescent staining of focal adhesions (green) with anti-vinculin antibodies (as indicated by yellow arrows, scale bars = 100 μm). (D) Actual adhesive area of the cell grown on three substrates (***p < 0.001 vs. smooth Ti; ^##p < 0.01 vs. STNTs. n = 3).

TNTs triggered RhoA inactivation by impairing FAK function in MC3T3-E1 cells and further played a role in cell osteodifferentiation

RhoA and FAK interact with each other to sense mechanical stimuli and regulate cell differentiation.¹³ To ascertain whether RhoA played a role in TNT-induced osteogenic differentiation, the RhoA activity on different substrates was first detected. RhoA activity was significantly reduced in the cells grown on TNTs compared with on smooth Ti (Fig. 6A). RhoA activity of the cells on the LTNTs showed the greatest decrease. Additionally, to clarify whether the effects of RhoA activity on cell osteodifferentiation is topography-dependent, C3 was used to block RhoA activity in the cells cultured with 6-well plate and to mimic the effects of TNTs to down regulate RhoA activation. The result showed higher bone marker gene expression when no mechanical stimulation was offered in C3+ group (Fig. 6B). This observation was similar to the effect of TNT surface on osteogenesis (Fig. 3). The results of both Fig. 6A and B suggested that the osteogenic effects of TNTs were RhoA related. Because RhoA is an important signaling node that regulates many of the cytoskeletal changes induced by force,²⁸ we predicted that the alteration of RhoA activity by TNTs would also influence the cell skeleton. As expected, the diminished cell cytoskeleton (F-actin staining, in red) was observed in the confocal laser scanning microscopy images on TNTs where very few mature FA spots (in green) (Fig. 6C) were visualized. Oppositely, on smooth Ti, the robust cytoskeleton connected to FAs and was distributed all over the cell. Meanwhile, it's been known that RhoA also serves as downstream target of FAK in the mechanotransduction cascades,¹⁵ we asked whether the impeded RhoA activity was caused directly by TNTs topography or indirectly by the inhibited FAK function on TNTs. To answer the question, we applied pY397-FAK pharmacological inhibitor PF573228 to the cells cultured with 6-well plate, and the immunoprecipitation of GTP-RhoA from FAK inhibitor treated cells in Fig. 6D reflected the RhoA activity decrease caused by pY397-FAK inhibitor. This phenomenon were consistent with Burridge's conclusion²⁸ and proved that on TNTs the weakened FAK function itself could down-regulate RhoA activation even without topographical cues. All these results indicated that the mechanical stimulus generated by the interaction between the TNT surface and cells was converted into the biochemical signaling following the FAK/RhoA cascade, which played a role in the subsequent cell osteodifferentiation. The degree of the effect on this signaling varied in the cells on different sized nanotubes.

Fig. 6 TNTs triggered RhoA inactivation associated with osteogenic differentiation in MC3T3-E1 cells. (A) Densitometric analysis of RhoA activity on smooth Ti and two TNTs substrates. GTP-RhoA were immunoprecipitated and blotted for RhoA (*p < 0.05 and **p < 0.01 vs. smooth Ti; ^##p < 0.01 vs. STNTs. n = 3). (B) Osteogenic gene expression in MC3T3-E1 cells cultured on 6-well plate (polystyrene, flat bottom) with and without C3 (*p < 0.05 vs. control group without C3 treatment, n = 4). (C) TRITC-phalloidin for F-actin staining (red), anti-vinculin FITC for focal adhesion staining (green) and DAPI for nucleus staining (blue) on three substrates (scale bars = 20 μm). (D) Immunoprecipitation western blot analysis of RhoA activity in 6-well plate (polystyrene, flat bottom) with and without PF 573228 (*p < 0.05 vs. control group without PF 573228 treatment, n = 3).

YAP cytoplasmic localization, in a RhoA dependent manner, was associated with TNT-induced osteogenic differentiation

A previous study showed that RhoA and the actin cytoskeleton were required to maintain nuclear YAP in MSCs, and the inhibition of RhoA activity and formation of the actin cytoskeleton changed YAP transcriptional activity.²² In addition, Zaidi et al. found that YAP associated with the native Runx2 protein, a master regulator on osteogenic differentiation, and suppressed Runx2 transcriptional activity. Inactivation of YAP by phosphorylation dissociated endogenous Runx2/YAP complexes and expelled YAP from the nucleus, and in this situation the unleashed Runx2 would induce the skeletal gene expression.²⁹ On the other hand, the nuclear localization of YAP enhances cell proliferation.^30,31 In the light of these findings, we hypothesized that TNTs induced osteogenic differentiation by attenuating FAK/RhoA activity, which resulted in YAP exportation from the nucleus, and activation of Runx2. Our previous investigation³² on the proliferation assay (BrdU assay) showed that LTNTs had promotive effects on cell proliferation at early culture stage, however, the overall trend of cell growth was significantly reduced over time compared with that of the cells on smooth Ti (Fig. 7A and S4†). The results of the nuclear/cytoplasmic fractionation showed that the nuclear YAP gradually decreased in the order of smooth Ti > STNTs > LTNTs (Fig. 7B). Consistent with this finding, the immunofluorescence staining showed that YAP remained in the nucleus of 87% cells on smooth Ti, and was expelled from the nucleus in 79% cells on LTNTs (Fig. 7C). The nuclear YAP was retained in approximately 35% of cells on STNTs (Fig. 7C). We further confirmed this finding with a YAP-responsive luciferase reporter (8× GTIIC-luciferase),³² as direct read-out of YAP translocation (Fig. S5†). These results indicated that the surface topography influenced the YAP allocation in the cells. The nanotubes caused YAP to be expelled from the nucleus, and the larger nanotubes had a stronger effect on the cells than the smaller ones.

Fig. 7 (A) The proliferation level of MC3T3-E1 cells cultured on three substrates quantified by immunofluorescent labeling of BrdU (*p < 0.05 and **p < 0.01 vs. smooth Ti; ^#p < 0.05 and ^##p < 0.01 vs. STNTs. n = 6). (B) Densitometric analysis of YAP levels in the nuclear and cytoplasm based on nuclear/cytoplasmic fractionation (*p < 0.05 and ***p < 0.001 vs. smooth Ti; ^#p < 0.05 vs. STNTs. n = 3). (C) Immunofluorescence staining of YAP (red) with anti-YAP antibody, cell nucleus (blue) with DAPI (scale bars = 20 μm) and the quantification of cells that displaying different YAP subcellular location on different substrates (n = 3).

RhoA has been acknowledged to be correlated with YAP nuclear location,²² and in our study, the cell appeared impeded RhoA activity as well as YAP cytoplasm distribution on LTNTs, which was concordant with Piccolo's findings. As mentioned above, FAK/RhoA cascade was proved to be engaged in the process of cell osteodifferentiation, to further ascertain if the expulsion of YAP from the nucleus acts the downstream of RhoA to regulate the TNT-induced osteogenic differentiation, we overexpressed YAP in the cells cultured on LTNTs (Fig. S6†) and activated RhoA with CN01. As shown in Fig. 8A, around 85% of YAP overexpressed cells failed to expel YAP from their nucleus and opposed the osteogenic induction of TNTs (Fig. 8B). Therefore, TNTs impeded the level of YAP in the nucleus and induced cell osteogenesis, conversely, the elevated nuclear YAP on TNTs overruled the topographical effects on cell osteodifferentiation.

Fig. 8 (A) The immunofluorescence staining and quantification of different YAP subcellular location in the cells cultured on LTNTs with/without YAP overexpression and CN01 treatment (scale bars = 20 μm, n = 3). (B) Osteogenic marker gene expression in MC3T3-E1 cells cultured on LTNTs with and without YAP overexpression (*p < 0.05 vs. control group, n = 4). (C) Osteogenic gene expression in MC3T3-E1 cells cultured on LTNTs with and without CN01 treatment (*p < 0.05 vs. control group, n = 4).

After RhoA was activated by CN01 in the cells grown on LTNTs, 72% of cells presented with totally or partially expelled YAP from the nucleus (Fig. 8A). Furthermore, the TNT-induced osteogenic differentiation was also inhibited by activating RhoA (Fig. 8C). Hence, we concluded that a decrease of nuclear YAP, in a RhoA-dependent manner, was required for TNT-induced osteogenic differentiation.

As discussed earlier, different dimensions of nanotubes triggered different cell responses to the surfaces. The results from other researchers' studies concluded that stem cells responded to the change in size of Ti nanotubes in a range of 15 nm to 100 nm, and small nanotubes with diameters less than 30 nm increase cell adhesion, proliferation, and integrin clustering/focal adhesion formation. These reactions tend to decline significantly with increasing pore size.^4,33,34 Our results agreed with these observations and suggested that the larger nanotubes decreased the rate of cell proliferation and reduced the focal adhesion formation more significantly compared with the smaller ones. Intriguingly, these differences were reflected as the stronger osteogenic differentiation of the cells on the larger nanotubes. Since YAP has been demonstrated to interact with Runx2 and inhibit Runx2 activity in the regulation of its downstream genes in the nucleus, we supposed that osteogenesis occurring on TNTs was associated with YAP translocation into cytoplasm, and therefore reduced the YAP/Runx2 binding probability. Such YAP translocation-mediated cell commitment is not independent of the tension of the cytoskeleton,²² and of note, the impeded RhoA activity on TNTs influenced the shuttling of YAP and altered the pattern of osteogenic gene expression accordingly. Given that a bridge to transmit the mechanical force between the ECM topography and the cellular cytoskeleton is indispensable, our evidence showed that the cell–TNT interaction influenced FAK/RhoA signaling and furthermore initiated the conversion between physical stimuli and the biochemical cues. Taken together, TNTs aroused a series of intracellular molecular dynamics which involved FAK/RhoA/YAP, and the chain reaction of this mechanism was not only reflected in the changes of cell spreading and proliferation but also the differentiation ability. A challenge of future studies will be to clarify the role of the FAK/RhoA/YAP pathway in the osteogenesis induced by nanotubular topography in vivo, and the bone regeneration investigation based on this mechanism will also contribute to a better understanding of the cellular responses and body reaction to biofunctional nanomaterials.

Experimental

TNTs fabrication, surface characterization and observation of adhered cells

Commercial pure Ti foils (Sigma-Aldrich, 99.7%, 0.25 mm thick) were cut into 1 × 1 cm² or 2 × 2 cm² squares. The natural oxide layer of the foils was removed using 600 and 1200 grit SiC abrasive paper (Electron Microscopy Sciences). Prior to anodization, Ti foils were sonicated in 100% acetone, 70% ethanol, and distilled water sequentially for 15 min each. The cleaned samples were then air dried and used as the anode while a platinum foil (Alfa Aesar) was used as the cathode. Both were connected to an electrochemical reaction flask that had a two-electrode configurations, and a DC power supply was used (Thermo Electron). In order to create nanotubes with different diameters on Ti surfaces, 10v or 30v was applied and a glycerol-based electrolyte was prepared with 0.25 wt% ammonium fluoride (Alfa Aesar, 96%) and 2 wt% deionized water; small Ti nanotubes (STNTs) were fabricated at 10v anodizing voltage and large Ti nanotubes (LTNTs) were fabricated at 30v anodizing voltage. To crystallize the fabricated amorphous structured Ti oxide (TiO₂) nanotubes into an anatase after anodizing, the samples were washed with deionized water, dried at 80 °C, and annealed with a muffle furnace (Thermolyne 6000) at 500 °C for 3 h. The non-anodized pure Ti pieces were used as a control (smooth Ti) after being polished, soaked in the same electrolyte for 3 hours, washed, and finally air dried. All the smooth Ti and TNT samples prepared for cell culture experiments were sterilized with 70% ethanol for 10–12 h followed by UV light overnight prior to use. The pure Ti implants for in vivo study were also prepared with same methods. Surface topography and cell adhesion were observed with a scanning electron microscope (SEM, Hitachi S-4700) and the available adhesive area of each substrate was analyzed with Image J (n = 6).

Cell culture and reagents

Newborn mouse calvaria-derived MC3T3-E1 preosteoblast cells (ATCC) were maintained in 1× MEM Alpha (Gibco) containing 10% FBS (Gibco) and 1% penicillin–streptomycin (Gibco) in an incubator with a humidified atmosphere of 5% CO₂ at 37 °C. The same culture conditions were applied to all the cells grown on experimental substrates. 200 ng mL⁻¹ C3 (Cytoskeleton)³⁵ and 500 nM PF573228 (Sigma-Aldrich)^36,37 were used to inhibit RhoA activity and FAK Y397 phosphorylation, respectively, for cells seeded in polystyrene, flat bottom, 6-well plates (Corning). To estimate the effects of RhoA inhibition on the osteogenic ability of MC3T3-E1 cells grown on 6-well plates, 10 mM β-glycerophosphate sodium (Sigma-Aldrich) and 0.2 mM ascorbic acid (Fisher Science) were used to induce osteogenic differentiation for both the control group (no C3 treatment) and C3-treated cells. MC3T3-E1 cells from passages 7–10 were seeded on substrates at 500 cells per cm² for SEM characterization and immunofluorescence staining; 5 × 10³ cells per cm² were used for all the other experiments unless otherwise specified. The Xpress (XPR)-tagged full-length YAP (1–472) plasmid was a kind gift from Dr Gary S. Stein at the University of Vermont. The plasmid was transfected into cells using the Attractene transfection reagent (Qiagen).

Immunofluorescence staining

After 24 h culture, immunofluorescence staining of focal adhesions, the cytoskeleton, and YAP was performed on the cells seeded on smooth Ti, STNTs, and LTNTs. The samples were first fixed with 4% formaldehyde (Tousimis® formaldehyde) for 15–20 min at room temperature, and then 0.1% Triton X-100 (Sigma-Aldrich) was used for permeabilization, followed by 1% bovine serum albumin (Gibco) blocking for 30 min. For vinculin and cytoskeleton staining, the primary and secondary antibody incubations were carried out using the Actin Cytoskeleton and Focal Adhesion Staining Kit (Millipore) according to the manufacturer's instructions. Vinculin, F-actin, and nuclei were visualized using a fluorescence microscope (Nikon) with TRITC, FITC, and DAPI filters. An anti-YAP antibody (Santa Cruz) was used as the primary antibody for the immunofluorescence staining of YAP. The BrdU assay was performed at 3 days, 5 days, and 7 days after seeding on the three substrates. For each time point, cells were incubated with BrdU (BD Biosciences) for 2 hours prior to fixing. A monoclonal mouse anti-BrdU antibody was used for primary antibody incubation at 4 °C overnight, followed by secondary antibody incubation with biotinylated goat anti-mouse IgG (H + L) (Invitrogen) for 60 min, and DAPI (Millipore) for 15 min. The proliferation level was indicated by the ratio of BrdU-labeled cells to all the adhered cells on each sample (an average of 200 cells counted, n = 6). For YAP and FA staining assays, 200 and 150 cells were counted in each sample (6 samples were involved, the experiment was repeated 3 times, n = 3). The images of FA and cytoskeleton staining were obtained with a Nikon Eclipse Ti-U digital camera (Japan) and the fluorescent images were merged by Nikon NIS Elements software. The translocation of YAP was visualized and photographed by a Zeiss LSM 710 Spectral Confocal Laser Scanning Microscope (Germany). The area of the single adherent cells was measured on the three substrates and the formula for calculating the actual adhesive area was as follows: the actual adhesive area = cell area × AACA/total surface area.

Quantitative real-time PCR

The expression levels of osteogenic markers were analyzed by quantitative real-time PCR using 18S as a control (n = 4). The RNA of MC3T3-E1 cells grown on both metal substrates and the 6-well plates was extracted using the TRIzol Reagent (Invitrogen) at day 7 after seeding. Reverse transcription was performed with a iScript™ cDNA synthesis kit (Bio-Rad) and PCR was performed with Taq DNA Polymerase (Invitrogen) using the Applied Biosystems 7500 system. The osteogenic-specific primers used (Runt-related transcription factor 2 [Runx2] and osterix [Osx]) are shown in Table 1.

Table 1 Primers used for real-time PCR

Gene	Forward primer sequence (5′-3′)	Reverse primer sequence (5′-3′)
Runx2	GAATGGCAGCACGCTATTAAATCC	GCCGCTAGAATTCAAAACAGTTGG
Osx	CCTCTCGACCCGACTGCAGATC	AGCTGCAAGCTCTCTGTAACCATGAC
18S	TGCATGGCCGTTCTTAGTTG	AGTTAGCATGCCAGAGTCTCGTT

---

Immunoprecipitation and western blot

Whole cell lysates were prepared for the immunoprecipitation of vinculin. 1.0 μg IgG was used to preclear the lysate and 10 μL anti-vinculin antibody (Santa Cruz) was added followed by end-to-end rotation at 4 °C overnight. The immunoprecipitates (IP) were collected and resuspended in 20 μL PBS before being boiled with loading buffer for SDS-PAGE (10%). To pull down active RhoA, 30 μg of rhotekin-RBD beads (Cytoskeleton) were applied to immunoprecipitate GTP-RhoA as described in a previous study.³⁸ The beads were collected and washed 3 times before resolving on 15% SDS-PAGE. For western blots, all proteins were electrotransferred to PVDF film and were probed with anti-vinculin (Santa Cruz), anti-pFAK, anti-FAK, and anti-RhoA antibodies (Cell Signaling). After reacting with the secondary antibody (HRP-conjugated IgG, GE Healthcare Life Sciences), immunoreactive bands were visualized using ECL (Thermo Scientific) and the densitometry analysis was performed with Image J and carried out from three independent experiments. Blots were stripped and reprobed for the loading control, GAPDH (Cell Signaling).

MTS assay

The effects of different RhoA activator, CN01 (Cytoskeleton), concentrations on MC3T3-E1 cell viability were assessed using an MTS assay (Promega). MC3T3-E1 cells cultured in a 96-well plate were treated with CN01 at different concentrations for 72 hours: 0 U mL⁻¹, 0.1 U mL⁻¹, 0.2 U mL⁻¹, 0.3 U mL⁻¹, 0.4 U mL⁻¹, 0.5 U mL⁻¹, 0.6 U mL⁻¹, 0.7 U mL⁻¹, 0.8 U mL⁻¹, 0.9 U mL⁻¹, and 1 U mL⁻¹. The absorbance of each group was measured at 490 nm using a plate reader (Biorad, Hercules, CA, USA).

Luciferase assay

The MC3T3-E1 cells were grown to 60–70% confluence before being trypsinized and plated on three substrates at a density of 0.06 × 10⁶ cells per mL with Opti-MEM (Gibco) containing 3% FBS (Gibco) and 1× MEM non-essential amino acids solution (Gibco). Meanwhile, 50 ng YAP/TAZ reporter, 8× GTIIC-luciferase DNA (a gift from Stefano Piccolo, Addgene plasmid # 34615),²² and 50 ng pGL3b empty vector were transfected into cells with Attractene reagent (Qiagen). The pRL Renilla luciferase control reporter vector (Promega) was included as a control for transfection efficiency. The cells were lysed after 24 h and luciferase activity was assessed with a luciferase assay kit (Promega) in the same cell lysate.

Animal and surgical procedures

For in vivo experiments, 10 six-week-old female Sprague Dawley rats (Charles River) were divided into 2 groups. The pure Ti implants with smooth surface (SI) were used as the control and the TNT implants (TNTI) after anodization at 30v voltage were used for the experimental group. To minimize the individual differences in the response to the different implants, each animal received two implants: right tibia for SI and left tibia for TNTI. The study protocol was approved by the Ethics Committee of Chongqing Medical University. All implants were implanted into the tibia metaphysis of the rats with average weights of 150–200 g.³⁹ Before operation, all rats were anesthetized with an intraperitoneal injection of ketamine (90 mg kg⁻¹) and xylazine (10 mg kg⁻¹). The hair of the surgical area was shaved and disinfected with chlorhexidine. After the incision, the tibia metaphysis was exposed and a hole (diameter: 1.7 mm; depth: 2 mm) was drilled with an electrical drill and a sterile pilot bur (1.7 mm). Sterile saline (0.9%, w/v) was used to wash the hole before the implantation of SIs and TNTIs. The muscle tissues were closed in different layers after implantation and the skin was closed with absorbable sutures. All rats were examined every 2 days for any sign of discomfort or infection. All the samples were harvested after 14 days.

Micro-computed tomography (μ-CT) analysis

The samples were harvested 14 days after surgery and fixed in 4% neutral formalin solution for 3 days followed by scanning with μ-CT (VivaCT 40, Scanco Medical, Bassersdorf, Switzerland) at 70 kV_p, 1000 mA with a 10 μm voxel size. The bone around the implants (0.3–0.5 mm) was selected as the regions of interest (ROI). Three-dimensional (3D) models were analyzed by Image Processing Language (IPL) to evaluate the formation of new bone. The bone volume per total volume (BV/TV) and the trabecular number (Tb. number) were used to indicate the degree of osteogenesis that occurred around the implant surface. All images were prepared at the same threshold values.

Histology staining

For histological observations, after fixation, the sample were decalcified with EDTA (10%, pH = 7.3) for 20 days and the implants were gently removed before the tissue was dehydrated, cleared, and embedded with paraffin. The sections were cut at 5 μm and processed with hematoxylin–eosin (H&E) staining. All images were captured with a Nikon Eclipse Ti-U digital camera (Japan).

Statistical analysis

Statistical significance was assessed by either a one-way analysis of variance (ANOVA) or a Student's t-test. All experiments were replicated at least three times to assure reproducibility, and the values are shown in the form of mean ± SD.

Conclusions

Our results indicated that nanotubes limited cell spreading, thereby restraining the formation of FAs. Consistent with this, FAK/RhoA signaling was impeded, which was followed by YAP export from the nucleus and Runx2 activation. As a result, the cells on the nanotubes initiated osteogenic differentiation. Although ECM/FAK-mediated changes in cell function have been suggested before, we first confirmed that the FAK/RhoA/YAP cascade triggered by TNT topography was to the mechanism behind the induction of osteogenesis. Further detailed research of other components relevant to the mechanical inputs that dictate cell responses can aid our approach to achieve optimal osteointegration by manipulating the surface topography of Ti implants.

Acknowledgements

We thank Ching-Chang Ko for assistance with the establishment of the anodization system and the gift of the Xpress (XPR)-tagged full-length YAP (1–472) plasmid, Gary S. Stein for the gift of the YAP-responsive luciferase reporter, and Nagasawa Masako for the suggestion and instruction of in vivo experiments. This work is financially supported by the National Natural Science Foundation of China (81500894), the Specialized Research Fund for the Doctoral Program of China (SRFDP20125503120009) and the Program for Young Scholars of Chongqing Medical University (CYYQ201506) to Dr S. Yang.

References

1. M. M. Stevens and J. H. George, Science, 2005, 310, 1135–1138.
2. M. Stevens, Mater. Today, 2008, 11, 18–25.
3. E. Kingham and R. O. C. Oreffo, ACS Nano, 2013, 7, 1867–1881.
4. J. K. Park, Y. J. Kim, J. Yeom, J. H. Jeon, G. C. Yi, J. H. Je and S. K. Hahn, Adv. Mater., 2010, 22, 4857–4861.
5. B. K. Teo, S. T. Wong, C. K. Lim, T. Y. Kung, C. H. Yap, Y. Ramagopal, L. H. Romer and E. K. Yim, ACS Nano, 2013, 7, 4785–4798.
6. N. Wang, H. Li, W. Lu, J. Li, J. Wang, Z. Zhang and Y. Liu, Biomaterials, 2011, 32, 6900–6911.
7. L. Zhao, S. Mei, P. K. Chu, Y. Zhang and Z. Wu, Biomaterials, 2010, 31, 5072–5082.
8. L. Zhao, H. Wang, K. Huo, X. Zhang, W. Wang, Y. Zhang, Z. Wu and P. K. Chu, Biomaterials, 2013, 34, 19–29.
9. S. Yang, M. Wang, H. Zhang, K.-y. Cai, X.-k. Shen, F. Deng, Y. Zhang and L. Wang, RSC Adv., 2014, 4, 65163–65172.
10. C. S. Chen, J. Tan and J. Tien, Annu. Rev. Biomed. Eng., 2004, 6, 275–302.
11. R. M. Salasznyk, R. F. Klees, A. Boskey and G. E. Plopper, J. Cell. Biochem., 2007, 100, 499–514.
12. R. M. Salasznyk, R. F. Klees, W. A. Williams, A. Boskey and G. E. Plopper, Exp. Cell Res., 2007, 313, 22–37.
13. B. Xu, G. Song, Y. Ju, X. Li, Y. Song and S. Watanabe, J. Cell. Physiol., 2012, 227, 2722–2729.
14. P. P. Provenzano and P. J. Keely, J. Cell Sci., 2011, 124, 1195–1205.
15. E. C. Lessey, C. Guilluy and K. Burridge, Biochemistry, 2012, 51, 7420–7432.
16. R. McBeath, D. M. Pirone, C. M. Nelson, K. Bhadriraju and C. S. Chen, Dev. Cell, 2004, 6, 483–495.
17. E. H. Ahn, Y. Kim, Kshitiz, S. S. An, J. Afzal, S. Lee, M. Kwak, K.-Y. Suh, D.-H. Kim and A. Levchenko, Biomaterials, 2014, 35, 2401–2410.
18. H. Sunami, I. Yokota and Y. Igarashi, Biomater. Sci., 2014, 2, 399.
19. C. H. Seo, H. Jeong, Y. Feng, K. Montagne, T. Ushida, Y. Suzuki and K. S. Furukawa, Biomaterials, 2014, 35, 2245–2252.
20. H. Chikumi, S. Fukuhara and J. S. Gutkind, J. Biol. Chem., 2002, 277, 12463–12473.
21. Y. Ogino, R. Liang, D. B. S. Mendonça, G. Mendonça, M. Nagasawa, K. Koyano and L. F. Cooper, J. Cell. Physiol., 2016, 231, 568–575.
22. S. Dupont, L. Morsut, M. Aragona, E. Enzo, S. Giulitti, M. Cordenonsi, F. Zanconato, J. Le Digabel, M. Forcato, S. Bicciato, N. Elvassore and S. Piccolo, Nature, 2011, 474, 179–183.
23. K. H. Wrighton, Nat. Rev. Mol. Cell Biol., 2011, 12, 404.
24. F. X. Yu, B. Zhao, N. Panupinthu, J. L. Jewell, I. Lian, L. H. Wang, J. Zhao, H. Yuan, K. Tumaneng, H. Li, X. D. Fu, G. B. Mills and K. L. Guan, Cell, 2012, 150, 780–791.
25. C. J. Frandsen, K. S. Brammer and S. Jin, Int. J. Biomater., 2013, 2013, 513680.
26. A. Pozio, A. Palmieri, A. Girardi, F. Cura and F. Carinci, Dent. Res. J., 2012, 9, S164–S168.
27. D. D. Schlaepfer, C. R. Hauck and D. J. Sieg, Prog. Biophys. Mol. Biol., 1999, 71, 435–478.
28. R. J. Marjoram, E. C. Lessey and K. Burridge, Curr. Mol. Med., 2014, 14, 199–208.
29. S. K. Zaidi, A. J. Sullivan, R. Medina, Y. Ito, A. J. van Wijnen, J. L. Stein, J. B. Lian and G. S. Stein, EMBO J., 2004, 23, 790–799.
30. M. Ota and H. Sasaki, Development, 2008, 135, 4059–4069.
31. N. McCarthy, Nat. Rev. Mol. Cell Biol., 2007, 8, 854.
32. H. Zhang, S. Yang, N. Masako, D. J. Lee, L. F. Cooper and C. C. Ko, RSC Adv., 2015, 5, 38117–38124.
33. J. Park, S. Bauer, K. von der Mark and P. Schmuki, Nano Lett., 2007, 7, 1686–1691.
34. S. Bauer, J. Park, J. Faltenbacher, S. Berger, K. von der Mark and P. Schmuki, Integr. Biol., 2009, 1, 525–532.
35. J. Heusinger-Ribeiro, M. Eberlein, N. A. Wahab and M. Goppelt-Struebe, J. Am. Soc. Nephrol., 2001, 12, 1853–1861.
36. A. Rape, W. H. Guo and Y. L. Wang, J. Cell Sci., 2011, 124, 4233–4240.
37. A. J. Keung, E. M. de Juan-Pardo, D. V. Schaffer and S. Kumar, Stem cells, 2011, 29, 1886–1897.
38. J. C. Krepinsky, J. Am. Soc. Nephrol., 2003, 14, 2790–2800.
39. M. Nagasawa, L. F. Cooper, Y. Ogino, D. Mendonca, R. Liang, S. Yang, G. Mendonca and K. Uoshima, J. Dent. Res., 2016, 95, 319–326.

---

Footnote

† Electronic supplementary information (ESI) available: The experimental strategy for synthesis of TNTs, SEM images of adherent cells on all substrates, vinculin staining, BrdU staining, luciferase activity of the YAP reporter, western blot result of YAP overexpression, and MTS assay for CN01 are shown in Fig. S1–S7. See DOI: 10.1039/c6ra04002k

---