Biomaterials 58 (2015) 93e102

Contents lists available at ScienceDirect

Biomaterials journal homepage: www.elsevier.com/locate/biomaterials

In situ hybridization of carbon nanotubes with bacterial cellulose for three-dimensional hybrid bioscaffolds Subeom Park a, 1, Jooyeon Park b, 1, Insu Jo a, Sung-Pyo Cho a, Dongchul Sung c, Seungmi Ryu d, Minsung Park e, Kyung-Ah Min c, Jangho Kim f, Suklyun Hong c, Byung Hee Hong a, **, 2, Byung-Soo Kim b, d, g, *, 2 a

Department of Chemistry, Seoul National University, Seoul 151-744, South Korea School of Chemical and Biological Engineering, Seoul National University, Seoul 151-744, South Korea Graphene Research Institute & Department of Physics, Sejong University, Seoul 143-747, South Korea d Interdisciplinary Program of Bioengineering, Seoul National University, Seoul 151e744, South Korea e Department of Biosystems & Biomaterials Science and Engineering, Seoul National University, Seoul 151-742, South Korea f Department of Rural and Biosystems Engineering, Chonnam National University, Gwangju 500-757, South Korea g Bio-MAX Institute, Institute of Chemical Processes, Engineering Research Institute, Seoul National University, Seoul 151-744, South Korea b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 April 2015 Accepted 10 April 2015 Available online 11 May 2015

Carbon nanotubes (CNTs) have shown great potential in biomedical fields. However, in vivo applications of CNTs for regenerative medicine have been hampered by difficulties associated with the fabrication of three-dimensional (3D) scaffolds of CNTs due to CNTs' nano-scale nature. In this study, we devised a new method for biosynthesis of CNT-based 3D scaffold by in situ hybridizing CNTs with bacterial cellulose (BC), which has a structure ideal for tissue-engineering scaffolds. This was achieved simply by culturing Gluconacetobacter xylinus, BC-synthesizing bacteria, in medium containing CNTs. However, pristine CNTs aggregated in medium, which hampers homogeneous hybridization of CNTs with BC scaffolds, and the binding energy between hydrophobic pristine CNTs and hydrophilic BC was too small for the hybridization to occur. To overcome these problems, an amphiphilic comb-like polymer (APCLP) was adsorbed on CNTs. Unlike CNT-coated BC scaffolds (CNT-BC-Imm) formed by immersing 3D BC scaffolds in CNT solution, the APCLP-adsorbed CNT-BC hybrid scaffold (CNT-BC-Syn) showed homogeneously distributed CNTs throughout the 3D microporous structure of BC. Importantly, in contrast to CNT-BC-Imm scaffolds, CNT-BC-Syn scaffolds showed excellent osteoconductivity and osteoinductivity that led to high bone regeneration efficacy. This strategy may open a new avenue for development of 3D biofunctional scaffolds for regenerative medicine. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Bacterial cellulose Bone regeneration Carbon nanotubes Hybrid materials Self-assembly

1. Introduction Carbon nanotubes (CNTs) have been of great interest to researchers in biomedical arena [1,2] due to their ability to enhance

* Corresponding author. School of Chemical and Biological Engineering, Seoul National University, Seoul 151-744, South Korea. Tel.: þ82 2 880 1509; fax: þ82 2 888 1604. ** Corresponding author. Tel.: þ82 2 880 6569; fax: þ82 2 889 1568. E-mail addresses: [email protected] (B.H. Hong), [email protected] (B.-S. Kim). 1 S. Park and J. Park contributed equally. 2 Co-corresponding authors. http://dx.doi.org/10.1016/j.biomaterials.2015.04.027 0142-9612/© 2015 Elsevier Ltd. All rights reserved.

cell functionality [3] and direct differentiation [4]. In particular, CNTs have emerged as a promising functional nanomaterial for bone regeneration as they can promote osteogenesis of mesenchymal stem cells [5], as well as osteoblast functioning [6], and bone calcification [7]. Previous studies reported that extracellular matrix proteins adsorbed on CNTs mediate cell adhesion and subsequently remodel cell cytoskeleton [8,9]. The cytoskeletal remodelling of the cells activates focal adhesion kinase and triggers cell signalling that promotes osteogenic differentiation [8,9]. However, despite all the findings discovered through in vitro cell culture on CNTs, in vivo applications of CNTs are still very limited due to the difficulties to fabricate three-dimensional (3D) microporous structures with nano-scale CNTs [9e12]. 3D microporous

94

S. Park et al. / Biomaterials 58 (2015) 93e102

structure is required for tissue engineering scaffolds to provide sufficient surface and space for cell adhesion, migration, growth, and tissue formation [13]. In addition, nano-scale fibers [14] and suitable mechanical properties [15] are preferable for effective bone grafts. Bacterial cellulose (BC), synthesized by Gluconacetobacter xylinus (G. xylinus), has 3D microporous structure composed of nanofibrous networks in layer-by-layer arrangement [16]. BC has many structural aspects favourable for tissue engineering scaffold, including large pores and nano-scale fibers in 3D structure [17e19]. Therefore, we thought that the hybridization of CNTs with BC would provide an environment suitable for bone regeneration in vivo considering osteogenic effects of CNTs and the adequate tissue-engineering scaffold properties of BC. An ideal CNT-BC hybrid scaffold would have CNTs homogeneously distributed throughout 3D microporous BC scaffold and stably hybridized with BC fibers. We sought to achieve this by culturing G. xylinus in culture media containing CNTs. However, pristine CNTs tend to aggregate with each other due to strong van der Waals interactions [20]. In addition, the binding energy between hydrophobic pristine CNTs [21] and hydrophilic BC nanofibrils [22] is too small for integration to occur (Fig. 2b). Hence, CNTs needed be modified to enhance the colloidal stability of CNTs and the hybridization between CNTs and BC. We utilized an amphiphilic comb-like polymer (APCLP) to prevent CNT aggregation and promote CNT-hybridization with BC. APCLP consists of hydrophobic backbone and hydrophilic side chains. APCLP coating on CNTs was achieved by hydrophobic interactions between ACPLP and CNTs. The APCLP coating would promote colloidal stability of CNTs by the hydrophilic side chains of APCLP. In addition, the hydrophilic side chains of APCLP would enable strong interactions between CNTs and hydrophilic BC nanofibers. We hypothesized that the simple method of culturing G. xylinus in culture media containing APCLPcoated CNTs would stably hybridize CNTs with BC nanofibers and homogeneously distribute CNTs throughout the 3D microporous BC scaffolds. A range of methods were applied to analyse the chemical, mechanical, and electrical properties of the scaffolds. The in vivo bone regeneration efficacy of the CNT-BC hybrid scaffolds was evaluated by implanting the scaffolds into mouse calvaria. 2. Materials and methods 2.1. Preparation of APCLP APCLP was synthesized by free radical polymerization of MMA (Aldrich), poly(ethylene glycol) methacrylate (Aldrich, Mn 360 g/ mol, corresponding to n ¼ 6) and poly(ethylene glycol) methyl ether methacrylate (Aldrich, Mn 475 g/mol, corresponding to n ¼ 9), in tetrahydrofuran for 18 h. The CNTs (purity >95%, lljin Nanotech Co., Korea) produced by the chemical vapour deposition method were used without further purification or treatment. The CNTs were determined to have an outer diameter of about 10e20 nm and a length of 150e200 mm. A total of 1 mg of CNT was added to 10 mL of APCLP solution (30% ethanol, 0.001% APCLP), after which the samples were sonicated using a horn-type ultrasonic generator (Fisher Scientific Co., USA) with a frequency of 23 kHz and a power of 30 W for 20 min at room temperature. Colloidal stability was measured using an ultravioletevisible spectrometer (UV-3600, Shimadzu, Japan). 2.2. G. xylinus culture G. xylinus (KCCM 40216) was obtained from the Korean Culture Center of Microorganisms. The bacterium was cultured on medium composed of 2.5% (w/w) mannitol, 0.5% (w/w) yeast extract and 0.3% (w/w) bacto-peptone. The culture media were sterilized in an

autoclave at 120  C for 20 min and then poured into 500 mL flasks. The pre-inoculum for all experiments was prepared by transferring a single G. xylinus colony grown on agar culture medium into a 100 mL Erlenmeyer flask filled with mannitol culture medium. The optimal culture conditions were determined empirically. 2.3. Computational methods To explain binding mechanism among CNT, polymer and cellulose, we have performed the density functional theory (DFT) calculation within generalized gradient approximation (GGA) using the Vienna ab initio simulation package (VASP) [23e25]. The projector augmented wave (PAW) potentials, as implemented in the VASP, were employed to describe the potentials from atom centers. The energy cutoff for the plane-wave basis was set to 400 eV in GGA. Geometries were optimized until the HellmaneFeynman forces acting on the atoms became smaller than 0.03 eV/Å. For investigation of binding mechanism, amphiphilic comb-like polymers (MMA and HPOEM) is considered between cellulose and CNT. To include weak van der Waals (vdW) interactions among them, we adopt the Grimme's DFT-D2 vdW correction based on a semi-empirical GGA-type theory [26]. For the Brillouin-zone interaction we used a 3  1  1 grid in Monkhorst-Pack special k-point scheme. In order to explain the behaviours of CNT-BC hybrids for large-scale dynamic system, we also have performed molecular dynamics (MD) simulation for CNT hybrids structures at room temperature (300 K). We performed NVT-MD simulations using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) with a reactive force field (ReaxFF) potential [27] for 12.5 ps. 2.4. Preparation of CNT-BC The pre-cultured culture cell suspension was introduced into 1  103% (w/v) multi-walled CNT-dispersed culture medium (pH 6.0) at a ratio of 1:10 and incubated at 28  C for 2 weeks. The CNTincorporated BC (CNTeBC) membrane biosynthesized in the medium was simply harvested and purified by boiling in 1 wt% sodium hydroxide for 2 h at 90  C. Subsequently, the membrane was thoroughly washed with running distilled water, after which it was soaked in 1 wt% aqueous sodium hydroxide solution for 24 h at room temperature to eliminate the cell debris and components of the culture liquid. The pH was then reduced to 7.0 by repetitive washing with distilled water. Next, the membrane was bleached by immersion in 1 wt% aqueous sodium hypochlorite for 2 h, after which it was washed with running distilled water until pH 7 was attained. Finally, the membrane was vacuum-dried at 60  C for 12 h. The normal BC membrane was prepared by harvesting a single G. xylinus colony in mannitol culture medium without CNTs, purifying and bleaching as described above. The membrane stored in distilled water was immersed in 1  103% (w/v) CNT-dispersed culture medium (pH 6.0) at 28  C for 2 weeks. The BC membranes were washed with running distilled water and vacuumdried at 60  C for 12 h. 2.5. Analyses of the scaffolds The samples were characterized by transmission electron microscopy (TEM), thermogravimetric analysis (TGA), optical microscopy, scanning electron microscopy (SEM), Raman spectroscopy, and atomic force microscopy (AFM). TEM (JEOL 2100, Japan) analyses were operated at 200 kV. TGA was performed under a nitrogen flow using a TA instrument Q-5000 IR. For optical microscopy analyses of the scaffolds, BC, CNT-BC-Imm, and CNT-BC-Syn were embedded in OCT compound, and were frozen at 70  C. The

S. Park et al. / Biomaterials 58 (2015) 93e102

samples were cut into 10 mm sections, and visualized using optical microscopy (Zeiss, Germany). For SEM analysis, BC, CNT-BC-Imm and CNT-BC-Syn were frozen at 70  C, followed by freeze-drying for 1 day. The surface morphology of the samples was observed at an acceleration voltage of 5 kV using SEM (JSM-6330F, JEOL). Raman spectra were taken from top, middle and bottom layer. Raman maps are of the G peak (1560e1620 cm1) bands. Raman spectra were recorded using a Renishaw inVia micro-Raman spectrometer (l laser ¼ 514 nm, ~500 nm spot size, 100  objector). The AFM observations and measurements were carried out by means of a Park systems XE-100 scanning probe microscope at ambient conditions. Silicon cantilever NSC-36 C (Mikromasch Inc) having pyramidal tips with 10 nm nominal radii of curvature(Rc). Cantilever spring constant of 0.60 N m1 were used. Topological data was employed a compressive load of 10 nN during scanning. After getting topography images, z-scanner displacement versus force curves were recorded. Forward and backward rate was 0.3 mm/s and a maximum compressive load of 40 nN was applied to the surface during data acquisition. The Young's modulus (E) of the samples were calculated using the Hertzian model, equation. Poisson's ratio(n) is set to 0.5 and parabolic geometry of the indenter is set. 2.6. In vivo implantation Six-week-old, female ICR mice (Koatech, Pyeongtaek, Korea) were anesthetized with xylazine (10 mg kg1) and ketamine (100 mg kg1) intraperitoneally. After shaving the scalp hair, a longitudinal incision was made in the midline of the cranium from the nasal bone to the posterior nuchal line and the periosteum was elevated to expose the surface of the parietal bones. Using a surgical trephine bur (Ace Surgical Supply Co., Brockton, MA, USA) and a low speed micromotor, two transosseous defects (4 mm diameter, circular) were produced in the skull. The defect size corresponded to the critical defect size for the mouse calvarial defect model. The drilling site was irrigated with saline, and bleeding points were electrocauterized. The periosteum was removed and never restored. The defect was filled with nothing, BC, CNT-BC-Imm, CNTBC-Syn, or Col-BMP-2 (n ¼ 6 implants per group). The periosteum and skin were then closed in layers with resorbable 6-0 Vicryl® sutures (Ethicon, Edinburgh, UK). The mice were housed singly after surgery. The study was approved by the Institutional Animal Care and Use Committee of Seoul National University (SNU-1101213). The implants were retrieved for analysis at 8 weeks after the surgery. 2.7. Analyses of bone regeneration Eight weeks after implantation, the animals were euthanized, and the skulls were harvested for analysis. Bone formation was evaluated with micro-CT scans (n ¼ 4 per group). The micro-CT images were obtained using a micro-CT scanner (SkyScan-1172; Skyscan, Belgium). After micro-CT imaging, the specimens were immersed in 4% paraformaldehyde solution, dehydrated in alcohol solutions of increasing concentrations, clarified in xylene, and embedded in paraffin. The specimens were sectioned transversely at a thickness of 4 mm. Then, the sections were soaked in xylene, hydrated in alcohol solutions of decreasing concentration, and washed with PBS solutions. The tissue sections were histologically stained with Goldner's Trichrome staining and immunohistologically stained with antibodies against osteocalcin (Abcam, UK). The immunostaining signal was visualized with rhodamine isothiocyanate-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, USA). The slides were counterstained with DAPI (Vector Laboratories, USA) to stain the nuclei of cells.

95

2.8. Statistical analysis Quantitative data were expressed as mean ± standard deviation. Statistical analyses were performed using analysis of variance (ANOVA). A p value of less than 0.05 was considered statistically significant. 3. Results 3.1. The colloidal stability of APCLP-coated CNTs APCLP is composed of long a hydrophobic methyl methacrylate (MMA) backbone and short hydrophilic side chains that consist of hydroxyl polyoxyethylene methacrylate (HPOEM) and polyethylene glycol methacrylate (POEM) (Fig. 1a). Molecular dynamics (MD) simulation showed that the hydrophobic backbone was attached to CNTs via hydrophobicehydrophobic interactions, and the hydrophilic side chains wrapped around CNTs to form an amphiphilic surface (Fig. 1a). The coating of APCLP facilitated the dispersion of CNTs in the culture medium (Fig. 1). Images obtained by SEM and TEM indicated that CNTs were uniformly dispersed after being treated with APCLP (Fig. 1b). We further confirmed the APCLP coating on CNTs by means of fourier transform infrared spectroscopy (Fig. 1c), and the results showed that APCLP-coated CNTs were well dispersed in culture medium for more than 3 months after APCLP coating (Fig. 1d). In addition, the colloidal stability of APCLP-coated CNTs was not influenced by pH change (Fig. 1e). 3.2. The effects of APCLP on CNT-BC hybridization The coating of APCLP not only facilitates the dispersion of CNTs, but also induces the hybridization of CNTs and BC (Fig. 2a). To clearly understand the APCLP-mediated binding of CNTs to BC, we performed ab initio calculations and MD simulations. First, the binding energies of CNT-BC and APCLP-coated CNT-BC were compared (Fig. 2b). The binding energy, Eb, is defined as Eb ¼ [Etotal  Econstituent1  Econstituent2], where Etotal, Econstituent1 and Econstituent2 are the energies of the total system (i.e., CNT-BC), constituent 1 (BC), and constituent 2 (CNT), respectively. The binding energy of CNTs and BC nanofibrils was 0.05 eV (per given unit), while the binding energy between BC nanofibrils was 0.68 eV (Fig. 2b). Therefore, BC nanofibrils entangle around each other without having interactions with CNTs. On the other hand, the binding energy of APCLP-coated CNTs and BC nanofibrils was 0.71 eV (Fig. 2b), causing BC nanofibrils to bind to and wrap around APCLP-coated CNTs. In this regard, MD simulation was performed to investigate the mechanism of CNT-BC and APCLPcoated CNT-BC formation (Fig. 2c). In the case of CNT-BC, BC assembled with each other without being hybridized with CNTs. On the contrary, in the case of APCLP-coated CNT-BC, BC nanofibrils wrapped around APCLP-coated CNTs to form a hybrid, indicating the essential role of APCLP in the hybridization of CNTs with BC. All CNTs used in the following experiments were present in the form of APCLP-coated CNTs, as pristine CNTs agglomerate and would be unsuitable for use in this study. Therefore, the word “CNTs” in all instances in the following paper indicate APCLPcoated CNTs. Due to the impracticality of using CNTs (without APCLP-coating) for the experiments, we compared the characteristics of the CNT-BC hybrid (i.e., CNT-BC-Syn) with those of CNTcoated BC formed by immersion (i.e., CNT-BC-Imm). The fabrication processes of both are shown in Fig. 2d.

96

S. Park et al. / Biomaterials 58 (2015) 93e102

Fig. 1. APCLP-mediated dispersion of CNTs. a) MD simulation of APCLP-coated CNT. b) SEM and TEM images of CNTs before and after APCLP coating. c) Fourier transform infrared spectroscopy spectrum of CNT, APCLP and APCLP-coated CNT. d) Colloidal stability of CNT and APCLP-coated CNT after centrifugation and after 3 months. e) Colloidal stability of APCLP-coated CNT in various pH conditions of culture medium.

3.3. Analysis of the hybridization between CNTs and BC TEM was performed to characterize the hybridization between CNTs and BC. TEM analysis showed that CNT-BC-Syn exhibited a coreeshell structure, where CNTs were packed by BC nanofibril entanglements (Fig. 3a). In contrast, CNTs and BC nanofibrils in CNT-BC-Imm were separate from each other (Fig. 3a). It is noteworthy that the BC nanofibrils in CNT-BC-Syn did not cover the entire surface of CNTs, resulting in a partial exposure of the CNTs to

the surrounding environment (Fig. 3b). The average thickness of the BC nanofibril entanglements was 4.3 nm, and about 3.9% of the CNT surface was uncovered (Fig. 3b). The exposure of CNTs and thin BC coating would enable the interaction of cells with CNTs. The electron energy loss spectroscopy (EELS) spectrum collected from BC showed a broad peaks at 291 eV assigned to the 1s to 2s* transition of carbon (Fig. 3c). On the other hand, EELS spectrum of CNT-BCSyn showed a sharp peak at 291 eV due to the 1s to p-orbital anti-bonding 2s* band transition (Fig. 3c), which was a

S. Park et al. / Biomaterials 58 (2015) 93e102

97

Fig. 2. APCLP-induced hybridization of CNTs and 3D scaffolds of BC. a) Schematic representation of APCLP-coated CNT-BC hybridization. b) Energies for CNT-BC binding (Eb) as measured by ab initio calculations. c) MD simulations for pristine CNT-BC hybridization and APCLP-coated CNT-BC hybridization (CNT-BC-Syn). d) Schematic illustration for the 3D scaffold fabrication processes of BC, CNT-BC-Imm (CNT-coated BC which was formed by immersing BC in APCLP-coated CNT solution) and CNT-BC-Syn.

98

S. Park et al. / Biomaterials 58 (2015) 93e102

Fig. 3. Analyses of the hybridization between CNTs and BC. a) TEM images of BC, CNT-BC-Imm and CNT-BC-Syn. b) Distribution of thickness of BC layer on CNTs, showing the partial exposure of bare CNTs. c) EELS spectra of BC and CNT-BC-Syn, showing the coreeshell hybridization structure of CNT-BC-Syn.

characteristic of CNTs. This difference in EELS spectrum indicated that CNTs were embedded in the thin layer of BC in CNT-BC-Syn. We examined the thermal decomposition of CNT-BC-Syn using TGA to determine the amount of CNTs in the scaffolds. The thermal degradation that occurred at 250350  C is attributed to the

depolymerization and cleavage of glycosidic linkages in BC (Fig. S1). The second decomposition at 350400  C resulted from the decomposition of the six-member cyclic structure of BC (Fig. S1). After decomposition of BC, 20 to 25 wt% of CNT-BC-Syn remained undecomposed, which is the amount of remaining CNTs (Fig. S1).

S. Park et al. / Biomaterials 58 (2015) 93e102

99

3.4. Analysis of CNT distribution in CNT-BC-Syn

3.5. Bone regeneration efficacy of CNT-BC-Syn

Intensive accumulations of black dots (i.e., CNTs) were observed on the edge of CNT-BC-Imm in the high resolution optical image (Fig. 4a). In contrast, CNTs in CNT-BC-Syn were homogeneously distributed throughout the scaffold. The SEM images showed that the accumulation of CNTs caused pore clogging at the CNT-BC-Imm surface (Fig. 4b); on the other hand, no CNT accumulation, thus no clogged pores were observed in CNT-BC-Syn (Fig. 4a and b). All scaffolds maintained the original layer-by-layer structure of BC (Fig. 4b). G-peak intensity mapping by means of Raman spectroscopy was performed to clearly depict the distribution of CNTs (Fig. 4c). Although the top and bottom surfaces of CNT-BC-Imm were fully covered with CNTs, CNTs were rarely observed inside the scaffolds. This result indicated that CNTs were predominantly present on the surface and absent from the interior of the scaffold in CNT-BC-Imm (Fig. 4c). On the other hand, CNTs were homogeneously distributed over the entire scaffold in CNT-BC-Syn (Fig. 4c). The high Young's modulus of the CNT-BC-Imm surface measured by AFM indicated that the surface of the scaffold was fully covered by CNT agglomerates (Fig. 4d). In addition, the homogenous distribution of CNTs in CNT-BC-Syn resulted in a reduced specific resistance and enhanced electrical conductivity, as compared with that of BC and CNT-BC-Imm (Table 1).

To evaluate the in vivo performance of CNT-BC-Syn, it was applied as a bone graft for bone regeneration application, and the results were compared with those obtained by using BC, CNT-BCImm, and collagen sponge loaded with bone morphogenetic protein-2 (Col-BMP-2), a clinically used bone graft [28]. The microscopic computed tomographic (micro-CT) analyses performed 8 weeks after the implantation of the scaffolds into the critical-sized defect in mouse calvaria showed a similar bone regeneration efficacy of CNT-BC-Syn, and Col-BMP-2, which was much higher than that of BC or CNT-BC-Imm (Fig. 5a). The analytical results of Goldner's trichrome staining of the histological sections revealed insufficient bone regeneration by BC and CNT-BC-Imm; in contrast, the use of CNT-BC-Syn resulted in extensive bone regeneration similar to that caused by Col-BMP-2 (Fig. 5b). It is noteworthy that although Col-BMP-2 is capable of forming new bone with the area similar to that formed by CNT-BC-Syn, the density of the regenerated bone was lower than that formed by CNT-BC-Syn (Fig. 5b). This indicates that CNT-BC-Syn is suitable for use as a bone graft material due to its capability in forming new bone with high density. In addition, 40 ,6-diamidino-2-phenylindole (DAPI, blue) and osteocalcin staining showed that the cells from the surrounding tissues have migrated into and have undergone

Fig. 4. Analyses showing the homogeneous distribution of CNTs in 3D hybrid scaffolds of CNT-BC-Syn and accumulation of CNTs on CNT-BC-Imm scaffold surface. a) Optical images of cross-sections of BC, CNT-BC-Imm and CNT-BC-Syn. Black or arrows indicate CNTs. Scale bars, 20 mm b) SEM images of BC, CNT-BC-Imm and CNT-BC-Syn. The surface of CNT-BCImm scaffolds were covered by CNTs. Scale bars, 5 mm c) Raman mapping of CNT-BC-Imm and CNT-BC-Syn. The red color indicates a high density of CNTs. d) Compressive force mapping of BC, CNT-BC-Imm and CNT-BC-Syn on the scaffold surface, as measured by contact-mode AFM. Scale bars, 2 mm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

100

S. Park et al. / Biomaterials 58 (2015) 93e102

Table 1 Properties of 3D scaffolds of BC, CNT-BC-Imm, and CNT-BC-Syn. Sample

Porosity (%)

Average pore diameter (mm)

Young's modulus at surfacea (kPa)

Specific resistance (kU/sq)

Electrical conductivity (S/cm)

BC CNT-BC-Imm CNT-BC-Syn

53.2 ± 6.7 56.8 ± 3.8 65.3 ± 5.3

41.1 ± 3.2* 51.7 ± 3.4* 85.4 ± 2.8*

30.4 ± 4.3* 474.0 ± 22.7* 56.6 ± 5.8*

500 ± 20* 123 ± 38* 5 ± 1*

0.13 ± 0.02 x 103* 2.98 ± 0.38  103* 0.08 ± 0.01*

*p < 0.05 compared to any other group. a The Young's modulus was the compressive modulus measured via AFM.

Fig. 5. Bone regeneration efficacy of the scaffolds, as evaluated by implanting scaffolds in mouse calvarial defects for 8 weeks. Collagen scaffolds loaded with BMP-2 (Col-BMP-2) served as a positive control. a) Bone regeneration evaluated by micro-CT analyses and quantification of the bone regeneration area in defects. b) Goldner's trichrome staining of mouse calvarial defect areas and quantification of bone formation area and new bone density in defects. Arrows indicate the bone defect margin. c) Immunohistochemistry of mouse calvarial defect areas. DAPI (blue) staining indicates osteoconductivity of the scaffolds. Osteocalcin (red) staining indicates osteoinductivity of the scaffolds. Scale bars, 2 mm *p < 0.05 compared to any other group. #p < 0.05 compared to either no treatment, BC or CNT-BC. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

S. Park et al. / Biomaterials 58 (2015) 93e102

osteogenic differentiation throughout CNT-BC-Syn in the defect area. However, cell migration and osteogenesis only occurred on the surface of CNT-BC-Imm. 4. Discussion We were able to successfully fabricate CNT-BC hybrid scaffold through a simple and facile method by utilizing APCLP. APCLP plays a critical role in the fabrication of CNT-BC-Syn as it facilitates the dispersion of CNTs and induces the hybridization of CNTs with BC. MD simulation suggested that APCLP modified the CNT surface through its unique comb-like molecular structure. The amphiphilic surface was able to facilitate the dispersion of APCLP-coated CNTs in the culture medium for at least 3 months (Fig. 1). In addition, the APCLP coating significantly enhanced the binding energy between CNTs and BC nanofibrils, which induced the hybridization of CNTs with BC (Fig. 2). In the absence of APCLP, the binding energy between CNTs and BC nanofibrils is too low for any interaction to occur between CNTs and BC, which would cause BC nanofibrils to assemble with each other rather than wrapping around CNTs. CNTs were stably hybridized with BC in CNT-BC-Syn (Fig. 3). TEM images demonstrated that BC nanofibrils wrapped around CNTs, and formed an ultra-thin layer of BC coating on each CNT (Fig. 3). The embedding of the CNTs occurred homogeneously throughout CNT-BC-Syn, resulting in homogeneous distribution of CNTs (Fig. 2). The 3D microporous structure of BC was unaffected by the hybridization of CNTs (Fig. 2b). The structure of CNT-BC-Syn is favourable for a bone graft. The pore size of CNT-BC-Syn, as measured by the prosimeter, was 85 mm (Table 1), which is suitable for osteogenic cell migration [29]. The pore sizes of the scaffolds in the SEM images (Fig. 4b) seem much smaller than the pore sizes measured by the porosimeter, which may be because of the scaffold shrinkage that occurred during SEM sample preparation. The nano-scale fibers constituting CNT-BC-Syn are favourable for cell attachment, proliferation, alkaline phosphatase synthesis and extracellular calcium deposition, which are all essential aspects of bone regeneration [30,31]. The resulting CNT-BC-Syn showed mechanical strength appropriate for osteogenic differentiation [32]. In addition, the hydrogel-like BC layers surrounding CNTs may act as a reservoir for accommodating various growth factors [33,34], which in turn can enhance the cell differentiation to facilitate bone regeneration. Previous studies have demonstrated that despite limited degradation of CNTs and BC in the body, both have shown a high level of biocompatibility [35,36]. In the present study, 20 to 25 wt % of CNT-BC-Syn was composed of CNTs (Fig. S1), and the weight of CNT-BC-Syn for implantation was 0.10 ± 0.01 mg. Therefore, the amount of CNTs incorporated in two CNT-BC-Syn scaffolds implanted to a mouse is 40e50 mg. A previous study has demonstrated that although CNTs were retained in vivo over 3 months, toxicity was minimal at a dosage less than 200 mg per mouse [37]. Therefore, the amount of CNTs in CNT-BC-Syn would not exhibit notable toxicity in vivo. In addition, CNTs showed minimized local inflammatory reactions and were well integrated into the newly formed bone after in vivo implantation for bone regeneration [35]. Similarly, BC was well integrated into the host tissue without inducing any inflammation when implanted in vivo [36]. BC is composed of only cellulose without any unwanted biogenic compounds, and it is less immune-stimulatory as compared with collagen, a commonly used scaffold material [38]. With above considerations, we have demonstrated the bone regeneration efficacy of CNT-BC-Syn in vivo. The implantation of

101

CNT-BC-Syn promoted new bone formation with high bone density (Fig. 5). The bone regeneration efficacy of CNT-BC-Syn was comparable to that of Col-BMP-2 scaffolds (Fig. 5). A combination of collagen sponges and BMP-2, which is clinically used, has been proven to be very effective in regenerating bone as collagen sponges can serve as osteoconductive matrix for cell migration and deliver BMP-2, the most potent osteoinductive factor [39]. Although BMP-2 has been proven to be effective in promoting bone regeneration [40], it has some drawbacks such as expensive cost, requirement of large doses, and short half-life in vivo [41]. Therefore, in our study, we utilized excellent osteoconductivity (migration of osteogenic cells from the surrounding environment to the defect area) [42] and osteoinductivity (promotion of osteogenic differentiation) [43] of CNT-BC-Syn to promote new bone formation without exogenous BMP-2. The 40 ,6-diamidino-2-phenylindole (DAPI, blue) staining of cell nuclei confirmed the cell migration into the scaffolds. The fluorescence of DAPI was visible throughout the defect area where CNT-BC-Syn was implanted, demonstrating that CNT-BC-Syn exhibited excellent osteoconductivity for promoting cell migration (Fig. 5c). The osteocalcin staining (red) shows that the cells have undergone pronounced osteogenic differentiation. Most of the cells in CNT-BC-Syn were successfully labelled with osteocalcin (Fig. 5c), suggesting that CNT-BC-Syn showed a high osteoinductivity. As regards to CNT-BC-Imm, the cells were found to only differentiate on the periphery of the defect area (Fig. 5c). The possible explanation for the enhanced osteoconductivity and osteoinductivity of CNT-BC-Syn, as compared with BC and CNT-BCImm, would be the homogeneous distribution of CNTs throughout 3D microporous BC scaffolds. In the present study, CNTs are absent inside the CNT-BC-Imm, therefore osteogenesis occurred only on the surface of the scaffold (Fig. 5c). 5. Conclusion In this study, an important method was introduced for the development 3D biofunctional scaffolds through hybridization with biofunctional nanomaterials for biomedical applications. BC was chosen as the 3D scaffold material for the hybridization of CNTs due to its ideal structure for tissue engineering scaffolds. APCLP was utilized to promote colloidal stability of CNTs and induce the hybridization between CNTs and BC, which enabled homogeneous and effective hybridization of CNTs with BC. Unlike conventional CNT-composite scaffolds fabricated through immersion of 3D BC scaffolds in CNT solution, CNT-BC-Syn scaffolds showed homogeneously distributed CNTs throughout the 3D microporous structure of BC. This resulted in excellent osteoinductivity and osteoconductivity of CNT-BC-Syn hybrid scaffolds, and in turn, high bone regeneration efficacy when implanted in vivo. This strategy for hybridizing 3D scaffolds with functional nanomaterials may present a new perspective for regenerative medicine. Acknowledgements This research was supported by the National Research Foundation of Korea (2014073757), Korea Health 21 R&D Project (HI14C1550 and HI14C3270) funded by the Ministry of Health and Welfare, Nano Material Technology Development Program (2012M3A7B4049888), and Nano Material Technology Development Program (2013069673), Republic of Korea. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.biomaterials.2015.04.027.

102

S. Park et al. / Biomaterials 58 (2015) 93e102

References [1] A. Bianco, K. Kostarelos, C.D. Partidos, M. Prato, Biomedical applications of functionalised carbon nanotubes, Chem Commun (2005) 571e577. [2] C. Cha, S.R. Shin, N. Annabi, M.R. Dokmeci, A. Khademhosseini, Carbon-based nanomaterials: multifunctional materials for biomedical engineering, ACS Nano 7 (2013) 2891e2897. [3] S.R. Shin, S.M. Jung, M. Zalabany, K. Kim, P. Zorlutuna, S.B. Kim, et al., Carbonnanotube-embedded hydrogel sheets for engineering cardiac constructs and bioactuators, ACS Nano 7 (2013) 2369e2380. [4] T.I. Chao, S. Xiang, C.S. Chen, W.C. Chin, A.J. Nelson, C. Wang, et al., Carbon nanotubes promote neuron differentiation from human embryonic stem cells, Biochem Biophys Res Commun 384 (2009) 426e430. [5] X.M. Li, H.F. Liu, X.F. Niu, B. Yu, Y.B. Fan, Q.L. Feng, et al., The use of carbon nanotubes to induce osteogenic differentiation of human adipose-derived mscs in vitro and ectopic bone formation in vivo, Biomaterials 33 (2012) 4818e4827. [6] S. Shao, S. Zhou, L. Li, J. Li, C. Luo, J. Wang, et al., Osteoblast function on electrically conductive electrospun pla/mwcnts nanofibers, Biomaterials 32 (2011) 2821e2833. [7] M. Shimizu, Y. Kobayashi, T. Mizoguchi, H. Nakamura, I. Kawahara, N. Narita, et al., Carbon nanotubes induce bone calcification by bidirectional interaction with osteoblasts, Adv Mater 24 (2012) 2176e2185. [8] S. Namgung, T. Kim, K.Y. Baik, M. Lee, J.M. Nam, S. Hong, Fibronectin-carbonnanotube hybrid nanostructures for controlled cell growth, Small 7 (2011) 56e61. [9] S. Ryu, C. Lee, J. Park, J.S. Lee, S. Kang, Y.D. Seo, et al., Three-dimensional scaffolds of carbonized polyacrylonitrile for bone tissue regeneration, Angew Chem Int Ed Engl 126 (2014) 9367e9371. [10] A. Abarrategi, M.C. Gutierrez, C. Moreno-Vicente, M.J. Hortiguela, V. Ramos, J.L. Lopez-Lacomba, et al., Multiwall carbon nanotube scaffolds for tissue engineering purposes, Biomaterials 29 (2008) 94e102. [11] M.C. Serrano, M.C. Gutierrez, F. del Monte, Role of polymers in the design of 3d carbon nanotube-based scaffolds for biomedical applications, Prog Polym Sci 39 (2014) 1448e1471. [12] S. Nardecchia, D. Carriazo, M.L. Ferrer, M.C. Gutierrez, F. del Monte, Three dimensional macroporous architectures and aerogels built of carbon nanotubes and/or graphene: synthesis and applications, Chem Soc Rev 42 (2013) 794e830. [13] M.A. Correa-Duarte, N. Wagner, J. Rojas-Chapana, C. Morsczeck, M. Thie, M. Giersig, Fabrication and biocompatibility of carbon nanotube-based 3d networks as scaffolds for cell seeding and growth, Nano Lett 4 (2004) 2233e2236. [14] K. Fujihara, M. Kotaki, S. Ramakrishna, Guided bone regeneration membrane made of polycaprolactone/calcium carbonate composite nano-fibers, Biomaterials 26 (2005) 4139e4147. [15] R.J. DeVolder, I.W. Kim, E.S. Kim, H. Kong, Modulating the rigidity and mineralization of collagen gels using poly(lactic-co-glycolic acid) microparticles, Tissue Eng Pt A 18 (2012) 1642e1651. [16] M. Henriksson, L.A. Berglund, Structure and properties of cellulose nanocomposite films containing melamine formaldehyde, J Appl Polym Sci 106 (2007) 2817e2824. [17] E.J. Vandamme, S. De Baets, A. Vanbaelen, K. Joris, P. De Wulf, Improved production of bacterial cellulose and its application potential, Polym Degrad Stab 59 (1998) 93e99. [18] M. Yeo, H. Lee, G. Kim, Three-dimensional hierarchical composite scaffolds consisting of polycaprolactone, beta-tricalcium phosphate, and collagen nanofibers: fabrication, physical properties, and in vitro cell activity for bone tissue regeneration, Biomacromolecules 12 (2011) 502e510. [19] J. Venugopal, S. Low, A.T. Choon, S. Ramakrishna, Interaction of cells and nanofiber scaffolds in tissue engineering, J Biomed Mater Res B Appl Biomater 84 (2008) 34e48.

[20] L.Q. Jiang, L. Gao, J. Sun, Production of aqueous colloidal dispersions of carbon nanotubes, J Colloid Interface Sci 260 (2003) 89e94. [21] C.G. Hu, C.H. Yang, S.S. Hu, Hydrophobic adsorption of surfactants on watersoluble carbon nanotubes: a simple approach to improve sensitivity and antifouling capacity of carbon nanotubes-based electrochemical sensors, Electrochem Commun 9 (2007) 128e134. [22] P.A. Charpentier, A. Maguire, W.K. Wan, Surface modification of polyester to produce a bacterial cellulose-based vascular prosthetic device, Appl Surf Sci 252 (2006) 6360e6367. [23] G. Kresse, J. Furthmuller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys Rev B 54 (1996) 11169e11186. [24] G. Kresse, J. Furthmuller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Comp Mater Sci 6 (1996) 15e50. [25] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple, Phys Rev Lett 77 (1996) 3865e3868. [26] S. Grimme, Semiempirical gga-type density functional constructed with a long-range dispersion correction, J Comput Chem 27 (2006) 1787e1799. [27] A. Strachan, A.C.T. van Duin, D. Chakraborty, S. Dasgupta, W.A. Goddard, Shock waves in high-energy materials: the initial chemical events in nitramine rdx, Phys Rev Lett (2003) 91. [28] H.S. Yang, W.G. La, Y.M. Cho, W. Shin, G.D. Yeo, B.S. Kim, Comparison between heparin-conjugated fibrin and collagen sponge as bone morphogenetic protein-2 carriers for bone regeneration, Exp Mol Med 44 (2012) 350e355. €la €, H.O. Yla €nen, C. Ekholm, K.H. Karlsson, H.T. Aro, Pore diameter of [29] A.I. Ita more than 100 mm is not requisite for bone ingrowth in rabbits, J Biomed Mater Res 58 (2001) 679e683. [30] P.X. Ma, Biomimetic materials for tissue engineering, Adv Drug Deliv Rev 60 (2008) 184e198. [31] K.L. Elias, R.L. Price, T.J. Webster, Enhanced functions of osteoblasts on nanometer diameter carbon fibers, Biomaterials 23 (2002) 3279e3287. [32] A.J. Engler, S. Sen, H.L. Sweeney, D.E. Discher, Matrix elasticity directs stem cell lineage specification, Cell 126 (2006) 677e689. [33] A.K. Silva, C. Richard, M. Bessodes, D. Scherman, O.W. Merten, Growth factor delivery approaches in hydrogels, Biomacromolecules 10 (2009) 9e18. [34] Q. Shi, Y. Li, J. Sun, H. Zhang, L. Chen, B. Chen, et al., The osteogenesis of bacterial cellulose scaffold loaded with bone morphogenetic protein-2, Biomaterials 33 (2012) 6644e6649. [35] Y. Usui, K. Aoki, N. Narita, N. Murakami, I. Nakamura, K. Nakamura, et al., Carbon nanotubes with high bone-tissue compatibility and bone-formation acceleration effects, Small 4 (2008) 240e246. [36] G. Helenius, H. Backdahl, A. Bodin, U. Nannmark, P. Gatenholm, B. Risberg, In vivo biocompatibility of bacterial cellulose, J Biomed Mater Res A 76 (2006) 431e438. [37] S.T. Yang, X. Wang, G. Jia, Y.Q. Gu, T.C. Wang, H.Y. Nie, et al., Long-term accumulation and low toxicity of single-walled carbon nanotubes in intravenously exposed mice, Toxicol Lett 181 (2008) 182e189. [38] N. Petersen, P. Gatenholm, Bacterial cellulose-based materials and medical devices: current state and perspectives, Appl Microbiol Biotechnol 91 (2011) 1277e1286. [39] M. Geiger, R.H. Li, W. Friess, Collagen sponges for bone regeneration with rhbmp-2, Adv Drug Deliv Rev 55 (2003) 1613e1629. [40] D.H.R. Kempen, L.C. Lu, A. Heijink, T.E. Hefferan, L.B. Creemers, A. Maran, et al., Effect of local sequential vegf and bmp-2 delivery on ectopic and orthotopic bone regeneration, Biomaterials 30 (2009) 2816e2825. [41] S.E. Kim, S.H. Song, Y.P. Yun, B.J. Choi, I.K. Kwon, M.S. Bae, et al., The effect of immobilization of heparin and bone morphogenic protein-2 (bmp-2) to titanium surfaces on inflammation and osteoblast function, Biomaterials 32 (2011) 366e373. [42] A.R. Vaccaro, The role of the osteoconductive scaffold in synthetic bone graft, Orthopedics 25 (2002) s571es578. [43] P. Habibovic, K. de Groot, Osteoinductive biomaterialseproperties and relevance in bone repair, J Tissue Eng Regen Med 1 (2007) 25e32.