Effect of hydrophilicity of carbon nanotube arrays on the release rate and activity of

recombinant human bone morphogenetic protein-2

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2011 Nanotechnology 22 295712

(http://iopscience.iop.org/0957-4484/22/29/295712)

Download details:

IP Address: 155.69.4.4

The article was downloaded on 22/08/2011 at 08:51

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 22 (2011) 295712 (8pp) doi:10.1088/0957-4484/22/29/295712

Effect of hydrophilicity of carbonnanotube arrays on the release rate andactivity of recombinant human bonemorphogenetic protein-2Zhao Jun Han1,2,5, Kostya (Ken) Ostrikov1,3, Cher Ming Tan2,Beng Kang Tay2 and Sean A F Peel4

1 Plasma Nanoscience Centre Australia (PNCA), CSIRO Materials Science and Engineering,Lindfield, New South Wales 2070, Australia2 School of Electrical and Electronic Engineering, Nanyang Technological University, 639798,Singapore3 School of Physics, University of Sydney, Sydney, New South Wales 2006, Australia4 Department of Dentistry, University of Toronto, Toronto, ON, M5G 1G6, Canada

E-mail: zhaojun.han@csiro.au

Received 2 May 2011, in final form 3 June 2011Published 21 June 2011Online at stacks.iop.org/Nano/22/295712

AbstractNovel nanostructures such as vertically aligned carbon nanotube (CNT) arrays have receivedincreasing interest as drug delivery carriers. In the present study, two CNT arrays with extremesurface wettabilities are fabricated and their effects on the release of recombinant human bonemorphogenetic protein-2 (rhBMP-2) are investigated. It is found that the superhydrophilicarrays retained a larger amount of rhBMP-2 than the superhydrophobic ones. Further use of apoloxamer diffusion layer delayed the initial burst and resulted in a greater total amount ofrhBMP-2 released from both surfaces. In addition, rhBMP-2 bound to the superhydrophilicCNT arrays remained bioactive while they denatured on the superhydrophobic surfaces. Theseresults are related to the combined effects of rhBMP-2 molecules interacting with poloxamerand the surface, which could be essential in the development of advanced carriers with tailoredsurface functionalities.

S Online supplementary data available from stacks.iop.org/Nano/22/295712/mmedia

(Some figures in this article are in colour only in the electronic version)

1. Introduction

Carbon nanotube (CNT) arrays have recently attractedincreasing interests in the development of novel and effectivecarriers for drug delivery systems and tissue engineering [1–3].Pristine CNTs possess exceptional mechanical, optical andelectrical properties, and a relatively good nanotube–cell andnanotube–tissue biocompatibility [4–6]. CNT arrays, whichconsist of vertically aligned individual or bundles of CNTs,show a high surface-to-mass ratio in addition to these excellentproperties, and therefore are highly suitable for carriers with

5 Author to whom any correspondence should be addressed.

an increased loading efficiency. Moreover, the surface of CNTarrays can be easily functionalized [1]. These advantageousfeatures have placed CNT arrays among the best candidatesfor economic, biocompatible and miniaturized drug deliverysystems [7].

Despite significant progress in device fabrication, theunderstanding of the mechanism of interactions betweenarrayed CNTs and proteins is far from complete. Suchunderstanding is important as protein immobilization is akey process in the early stage of implanted CNT-basedmaterials and determines many later cellular activities suchas cell attachment and proliferation [8, 9]. Several surface

0957-4484/11/295712+08$33.00 © 2011 IOP Publishing Ltd Printed in the UK & the USA1

Nanotechnology 22 (2011) 295712 Z J Han et al

physical and chemical properties, e.g. wettability, roughnessand topography, charge and functional groups, have beenextensively studied [10–14]. Among them, surface wettabilityhas been identified as the prime factor that governs not onlyprotein adsorption, but also protein-release kinetics over a longperiod of time [15–18]. Unfortunately, surface wettabilityaffects proteins in a complex way with a strong dependenceon the protein type. For instance, bovine serum albumin(BSA), a protein frequently used as a blocking agent, adsorbsstrongly on hydrophobic surfaces but poorly on hydrophilicsurfaces [10, 11]. Ferritin and fibrinogen also showed similaradsorption behaviors as that of BSA, while fibronectin andsome sera proteins adsorbed more strongly on hydrophilicsurfaces than on hydrophobic surfaces [19–22].

Recently, surfaces utilizing a gradient wettability wereengineered yet the wettability often varied in a narrowrange [23, 24]. In this study, we fabricate two CNTarrays with similar microstructures but vastly differentwettability, and extend the study of protein release onnanostructural surfaces into two extreme conditions, namelysuperhydrophobicity and superhydrophilicity. Recombinanthuman bone morphogenetic protein-2 (rhBMP-2) is usedas a model protein owing to its potential applications inbone regeneration and oral reconstruction [25, 26]. Ourmeasurements of rhBMP-2 release show that a strong proteinretention and a more sustainable release profile were achievedon the superhydrophilic surface, which can be further improvedby using a poloxamer diffusion layer. Bioactivity analysisindicated that rhBMP-2 bound on the superhydrophilic CNTarrays retained their bioactivity but denaturized on thesuperhydrophobic CNTs. These findings are importantfor utilizing CNT arrays in clinical applications and couldalso find medical applications in many biomedical devicesfunctionalized with superhydrophilic or superhydrophobiccoatings [27].

2. Experimental details

2.1. Fabrication of superhydrophilic and superhydrophobicCNT arrays

The pristine CNT arrays which rendered superhydrophilicitywere synthesized by using plasma-enhanced chemical vapordeposition (PECVD) [28–31]. Briefly, a thin layer of Nicatalyst (10 nm) was firstly deposited on smooth Si surface(n-type) by using the electron beam evaporator (AUTO 306,Edwards). The catalyst was then loaded into the PECVDreactor and dewetted at 800 ◦C for 2 min in the presence of240 sccm ammonia (pressure ∼10 mbar). Subsequently, a flowof 60 sccm acetylene was introduced and plasma was generatedat a DC power of 80 W. CNT arrays were grown for 10 minwith the working pressure kept at 12 mbar.

To prepare the superhydrophobic CNT arrays, theaforementioned pristine CNT arrays were placed on a metalelectrode in the presence of argon gas (pressure ∼ 0.1 mbar).Ar+ ions were then generated by applying a pulsed bias to themetal electrode at a voltage of −10 kV, a repetition rate of500 Hz and a pulse width of 20 μs. After 5–10 min, pseudo-spherical amorphous carbon (a-C) nanoparticles were found on

top of each nanotube [32]. This material was termed as ion-processed CNT (ipCNT) arrays and the schematic illustrationof their formation is shown in figures 1(a)–(c).

Surface wettability of both pristine and ipCNT arrayswas determined by water contact angle measurementsusing the sessile drop method (OCA-20, Dataphysics).Scanning electron microscopy (SEM; JSM5910LV, JEOL) andtransmission electron microscopy (TEM; JEM2010, JEOL)with an electron beam energy of 200 keV were used to examinethe microstructure. Prior to TEM observations, the sampleswere sonicated in ethanol for 5 min in a bath sonicator. Thesuspension containing either nanotube was then dropped ontoa holey carbon-coated copper grid and dried in air.

2.2. rhBMP-2 retention and release

Pristine, ipCNT and flat Si wafers (used as control samples)were placed in triplicate in a cell culture plate. The plate wasthen sterilized under UV light for 30 min in a biological safetycabinet (BSC). To ensure a fixed amount of protein loadingand to mimic the protein delivery in real clinical procedures,200 ng of rhBMP-2 (RnD Systems, Burlington) in distilledwater was dispersed onto the surface of each substrate anddried overnight.

To measure the amount of released rhBMP-2, 1 ml ofphosphate buffer saline with 0.1 wt% bovine serum albumin(PBS-BSA) was added to each well and maintained at 37 ◦C inan incubator. The PBS-BSA was replaced with fresh solutionat 10 min, 1, 4 and 8 h, and after 1, 2, 4 and 7 days.The collected solutions containing released rhBMP-2 werestored at −20 ◦C and subsequently measured using an enzyme-linked immunosorbent assay (ELISA; PeproTech, 900-K255)following the manufacturer’s instructions [33].

For the experiments with diffusion layers, sterile 33 wt%poloxamer 407 (Sigma Aldrich) in water was applied ontop of the substrates after rhBMP-2 loading and left to dryovernight. The same procedures to measure the amount ofreleased rhBMP-2 were then followed.

2.3. Bioactivity analysis

The bioactivity of rhBMP-2 bound on each substrate afterrelease was evaluated using the mouse myoblast cell (C2C12)-based assay [34]. Briefly, C2C12 cells (ATCC, Burlington)with two passages were seeded onto the substrates at 0.5 ×105 cells ml−1 alpha minimum essential medium (α-MEM)with 15% fetal bovine serum. Additional C2C12 cellsincubated in the medium with or without rhBMP-2 (40 ng)were used as positive and negative controls, respectively.The cell cultures were then terminated after 2 days andthe cell layer was lysed in CelLytic (Sigma-Aldrich). Thealkaline phosphatase (ALP) level of the lysates was determinedusing p-nitrophenol phosphate (pNP; Sigma-Aldrich) andnormalized to the protein content determined using theCoomassie plus assay (Fisher Scientific).

2.4. Statistical analysis

All statistical analyses were performed using SigmaStat (v3.5,Systat) and all data was tested for normality. Normal data

2

Nanotechnology 22 (2011) 295712 Z J Han et al

Figure 1. The formation and structure of the pristine and ipCNT arrays. (a)–(c) Schematic illustration of the formation of ipCNTs byenergetic ion bombardment. (d)–(e) SEM and TEM images of the pristine CNT arrays. Scale bar is 1 μm in (d). (f)–(g) SEM and TEMimages of the ipCNT arrays. a-C and CNT are clearly visible in (g). Scale bar is 1 μm in (f). (h) High-resolution TEM image and(i) selected-area electron diffraction pattern of a-C nanoparticle as shown in (g).

was compared by ANOVA with the SNK post hoc test, whilenon-normal data was compared using ANOVA on Ranks withDunnett’s post hoc test. The ALP levels measured fromdifferent substrates were compared with that of the substrateincubated without rhBMP-2. A p value <0.05 was consideredsignificant.

3. Results and discussion

The SEM and TEM images of pristine CNT arrays areshown in figures 1(d) and (e), respectively. As one cansee, an individual multi-walled nanotube has a diameter

of 50–200 nm, a height of 1–2 μm and an inter-tubedistance of ∼200 nm. These nanotubes are grown verticallyowing to the plasma-sheath-oriented growth in the PECVDprocess [35–39]. On the other hand, the ipCNT arraysshowed a slightly different microstructure. While the verticalalignment was retained, figures 1(f)–(i) indicated that pseudo-spherical nanoparticles with a diameter slightly larger than thenanotube were formed on the top surface, encapsulating eachnanotube. The high-resolution TEM image and the selected-area electron diffraction (SAED) pattern of the nanoparticles(figures 1(g)-(i)) revealed that they were amorphous carbon (a-C). We have previously shown that such formation was a result

3

Nanotechnology 22 (2011) 295712 Z J Han et al

Figure 2. Surface wettability of the pristine and ipCNT arrays. Water contact angle measurements show that (a) the pristine CNT arrays aresuperhydrophilic and (b) the ipCNT arrays are superhydrophobic. (c)–(d) Schematic illustration of water droplet on the pristine and ipCNTarrays, respectively.

of energetic ion bombardment on CNT arrays which smashedthe crystalline structure of CNTs and transformed them intoan amorphous structure in the presence of a strong electricfield [28, 35].

Figure 2(a) shows that the pristine CNT arrays weresuperhydrophilic and a nearly 0◦ contact angle was achievedwhen a water droplet was dispersed on the surface. In contrast,the ipCNT arrays featured superhydrophobic properties, asshown by a water contact angle larger than 150◦ shown infigure 2(b). This remarkable difference in surface wettabilitywas attributed to different microstructure of the two materials.For the pristine CNT arrays, the top surface is comprised of Nicatalysts which have a contact angle of ∼70◦. When a waterdroplet was in contact with the surface, it could easily seepinto the gaps between neighboring nanotubes through capillaryactions, resulting in a complete wetting state (figure 2(c)) [40].However, a slightly larger diameter of a-C caps gave a unique‘overhang’ structure in the ipCNT arrays [41]. Although thehydrophobicity of a-C nanoparticles was similar to that of theCNTs, its pseudo-spherical shape required a large pressure forthe water droplet to penetrate the surface and fill into the airgaps between neighboring nanotubes, a situation that was notfacilitated in the present case. Contact angle θ of the ipCNTarrays arising from the combined effects of surface roughnessand the unique ‘overhang’ structure is therefore given by the‘Cassie’ equation [28]:

cos θ = ϕs(cos θs + 1) − 1

where ϕs is the fraction of the solid surface in contact with thewater droplet and θs is the contact angle of a-C. This equationimplies that, even though θs is slightly hydrophilic, θ can bevery large if ϕs is small enough. Using the average diameterof a-C nanoparticles of 100 nm and the inter-tube distance of

300 nm (figure 1(g)), and θs = 80◦ for a-C [42], the contactangle θ was calculated to be 153.8◦, in good agreement withthe measurement.

With the prepared two CNT arrays having a similarmicrostructure but vastly different surface wettability, we theninvestigated their interactions with rhBMP-2, a protein thatcan stimulate the differentiation of human mesenchymal stemcells (hMSC) and promote bone formation in vivo [25, 26].During rhBMP-2 loading, we noted that the protein solutionrapidly filled the pristine CNT arrays and the morphologyof arrays transformed into a ‘pyramid’ structure after drying(available in supplementary information at stacks.iop.org/Nano/22/295712/mmedia). However, it retained a sphericalshape and a large contact angle on the ipCNT arrays throughoutthe drying process (available in supplementary informationat stacks.iop.org/Nano/22/295712/mmedia). These loadingcharacteristics agreed well with their corresponding surfacewettability and consequently affected their release profile, asdescribed below.

Figure 3(a) shows the release of rhBMP-2 measured atdifferent time intervals. For the flat Si substrates and thesuperhydrophobic ipCNT arrays, there were large amounts ofrhBMP-2 released over the initial 4 h, followed by a steadydecline at subsequent stages. In contrast, relatively constantamounts of release, albeit fluctuating, were observed on thesuperhydrophilic pristine CNT arrays at all time intervals. Thecumulative release of all three substrates is also shown infigure 3(b). From these curves, one can see that the totalamount of released rhBMP-2 was significantly reduced on boththe superhydrophilic and the superhydrophobic CNT arraysas compared to the Si wafer, i.e. an enhanced retentionwas achieved. Notably, the cumulative release from thesuperhydrophilic CNT arrays at 24 h (1.87 ± 0.35%) was fivetimes less than that of the Si wafer (10.04 ± 1.45%).

4

Nanotechnology 22 (2011) 295712 Z J Han et al

Figure 3. The release of rhBMP-2 on the superhydrophilic andsuperhydrophobic CNT arrays, and the Si control. (a) The release ateach time interval and (b) the cumulative profile. The releasepercentage was calculated with respect to the initial loading (200 ng).

Using nanostructured materials is known to increaseprotein adsorption through stronger interactions such as vander Waals and electrostatic forces [43], as shown by our resultshere. In addition, the difference between the superhydrophilicand superhydrophobic surfaces also suggested that surfacewettability affected the retention of rhBMP-2 and consequentlythe release profile both at the initial stage and in the longterm. In the case of pristine CNT arrays, as the entire surfacewas wetted by rhBMP-2 solution, the contact area betweenthe proteins and the nanotubes became much larger, resultingin stronger interactions and hence a maximum retention ofrhBMP-2. In contrast, the superhydrophobicity of ipCNTarrays to a large extent prevented rhBMP-2 molecules frominteracting with the nanotubes (available at stacks.iop.org/Nano/22/295712/mmedia), thereby weakening the effect of alarge surface-to-mass ratio which is considered beneficial foran enhanced rhBMP-2 adsorption.

The plots in figure 3(b) also indicated that the releaseof rhBMP-2 was significant at the initial 4 h, accounting for∼50% in the case of the superhydrophilic arrays and ∼80% in

the cases of the superhydrophobic arrays and the Si control.This initial burst release has been one of the obstacles forimplementing conventional carriers (e.g. collagen sponges,calcium phosphates, etc) in rhBMP-2 delivery [44, 45]. It couldgenerate a high local rhBMP-2 concentration which is thenlost prior to the arrival of the responsive hMSC at the repairsite. Such a situation not only wastes a significant amount ofthe prematurely released rhBMP-2, but also potentially resultsin side effects in host body systems, including excess boneformation, immunogenicity, formation of bone voids and evencancer [46].

An additional diffusion layer on top of the loaded rhBMP-2 was used to mitigate this problem. Poloxamer 407 waschosen as the diffusion layer because of its proven non-toxicity, demonstrated bone biocompatibility and effectivenessin delaying drug release [47, 48]. Another advantage of usingthis poloxamer is its reverse thermal property where it canchange between a liquid and a gel at its transition temperature.For 33 wt% poloxamer 407, it is a liquid at 4 ◦C but acquiresgel-like properties above 15 ◦C. We therefore were able toapply a cooled poloxamer to the surfaces as a liquid, let itbecome a gel and act as a diffusion barrier to the loadedrhBMP-2 upon warming to room temperature.

The same procedures to measure the release of rhBMP-2 were conducted for the CNT arrays coated with poloxamer.Figures 4(a) and (b) show the release at each time intervaland the cumulative release, respectively. Compared withthe previous results, the most significant feature was thatno rhBMP-2 was detected on the superhydrophilic arraysduring the initial 4 h and the maximum release rates forboth superhydrophilic and superhydrophobic arrays were onlyobserved after 25 h (figure 4(a)). This carrier structure witha diffusion barrier therefore showed a significant improvementin delaying the initial release, as compared to the simple dry-loading method. In addition, we noted that in the long termthe total amount of rhBMP-2 released from all substrates withthis poloxamer structure was higher as compared to that fromsubstrates without the poloxamer. The cumulative release hasincreased by 56%, 100% and 240% for Si, pristine and ipCNTarrays, respectively (figures 3(b) and 4(b)).

We attributed this observation to two possible comple-mentary mechanisms. The first mechanism was related to themolecular structure of the poloxamer. Poloxamer 407 is a non-ionic triblock copolymer which is composed of a hydrophobicbackbone of propylene oxide (PPO) and two hydrophilicchains of poly(ethylene oxide) (PEO) [48]. The hydrophilicPEO chains may interact with loaded rhBMP-2 molecules,weakening their interactions with the nanotubes and draggingthem into the aqueous solution, resulting in more rhBMP-2detaching from the surfaces. Another possible mechanism wasrelated to a process called the ‘Vroman effect’ [49, 50]. The‘Vroman effect’ occurs when more than one protein species isadsorbed on the surface. An initially adsorbed protein could beslowly displaced by a second protein and released into solutionif the second protein has a higher affinity to the surface. Forexample, when blood came in contact with a hydrophobicsurface, fibrinogen adsorbed rapidly, but was subsequentlydisplaced by apolipoproteins that were present at a much lower

5

Nanotechnology 22 (2011) 295712 Z J Han et al

Figure 4. The release of rhBMP-2 on the superhydrophilic andsuperhydrophobic CNT arrays and the Si control coated with thepoloxamer diffusion layer. (a) The release at each time interval and(b) the cumulative profile. The release percentage was calculatedwith respect to the initial loading (200 ng).

concentration but had a higher affinity to the surface [50]. In asimilar fashion we postulated that the hydrophobic propyleneoxide core of the poloxamer molecules had a higher affinity tothe nanotubes than the initially adsorbed rhBMP-2. When thepoloxamer was applied to the substrate it gradually displacedthe weakly bound rhBMP-2 which was merely physisorbed onthe surface. The displaced rhBMP-2 molecules could then bereleased into the buffer solution.

The preservation of native conformations of proteins isanother important issue affecting the efficacy of carriers. Herethe bioactivity of rhBMP-2 bound on the substrates wasevaluated by using the C2C12-based assay, which was basedon the principle that the level of ALP expressed by C2C12 cellsis proportional to the amount of bioactive rhBMP-2 presentin the incubation [34]. Both C2C12 cells cultured in thepresence and absence of rhBMP-2 were also used as positiveand negative controls. As shown in figure 5, the ALP level wasthe highest for C2C12 cells incubated on the superhydrophilicCNT arrays; while a significant value (∗ p < 0.01) wasobserved for cells incubated with 40 ng rhBMP-2 and on the Sicontrol sample. This result implicated that the rhBMP-2 bound

Figure 5. The ALP levels of C2C12 cells incubated with rhBMP-2bound on different substrates. Controls include cells incubated inmedium with and without 40 ng ml−1 hBMP-2. p values arecalculated between substrates as linked by the dashed lines. ∗ denotesthe significance p < 0.01.

to the superhydrophilic CNT arrays and the flat Si substrateremained active. The highest ALP level in the superhydrophilicarrays was consistent with the previous measurements thatmore rhBMP-2 was retained on the surface.

Figure 5 also showed that cells cultured on thesuperhydrophobic arrays did not display any significantincrease in ALP activity as compared to cells cultured withoutrhBMP-2. This was a surprising result as it was shown infigure 3(b) that only ∼4%-loaded rhBMP-2 were releasedfrom the superhydrophobic samples (note that this estimateis accurate if protein binding to the walls of the cell cultureplate is neglected). Although it was possible that surfaceroughness and chemistry could affect the ALP expressionlevel [51, 52], we attributed the lack of ALP to the denaturationof these proteins on the superhydrophobic surface. It hasbeen demonstrated previously that many proteins could breaktheir intramolecular bonds to establish new bonds betweenthe polypeptide chain and the substrate upon contact with ahydrophobic surface [11, 21, 53]. The water structure in thevicinity of a hydrophobic surface may also be changed by thesurface, forcing proteins to change their conformations evenbefore they arrived at the surface and resulting in partial orcomplete loss of activity [53].

4. Conclusions

In view of the potential applications of superhydrophobic andsuperhydrophilic surfaces, we have investigated the release ofrhBMP-2 on both pristine and ipCNT arrays. It has been foundthat the superhydrophilic arrays retained more rhBMP-2 thanthe superhydrophobic ones. The use of poloxamer in rhBMP-2 loading also delayed protein release by creating a diffusionbarrier, hence inhibiting the large initial burst commonlyobserved in conventional carriers. In the long term, thecombined effects of rhBMP-2 interacting with the poloxamerand the surface resulted in a larger amount of protein released.

6

Nanotechnology 22 (2011) 295712 Z J Han et al

Additionally, it was shown that rhBMP-2 bound to thesuperhydrophilic CNT arrays remained bioactive while theydenatured on the superhydrophobic arrays. These results arehighly relevant to the development of next-generation drug,gene and protein delivery systems.

Acknowledgments

We thank Dr H Y Yang, A J J Zhou and Z N Zhufor technical assistance and helpful discussion. ZJHis grateful to Professor C M L Clokie for a VisitingScholarship. This work was partially supported by theNanyang Technological University Seed Fund (Singapore), theOral Surgery Foundation of Canada (Canada), the AustralianResearch Council and CSIRO’s OCE Science LeadershipProgram (Australia).

References

[1] Wu J, Paudel K S, Strasinger C, Hammell D, Stinchcomb A Land Hinds B J 2010 Programmable transdermal drugdelivery of nicotine using carbon nanotube membranes Proc.Natl Acad. Sci. USA 107 11698–702

[2] Riggio C, Ciofani G, Raffa V, Cuschieri A and Micera S 2009Combination of polymer technology and carbon nanotubearray for the development of an effective drug deliverysystem at cellular level Nanoscale Res. Lett. 4 668–73

[3] Harrison B S and Atala A 2007 Carbon nanotube applicationsfor tissue engineering Biomaterials 28 344–53

[4] Jorio A, Dresselhaus G and Dresselhaus M S 2008 CarbonNanotubes: Advanced Topics in the Synthesis, Structure,Properties and Applications vol 111 (Berlin: Springer)

[5] Usui Y et al 2008 Carbon nanotubes with high bone-tissuecompatibility and bone-formation acceleration effects Small4 240–6

[6] Lahiri D, Rouzaud F, Namin S, Keshri A K, Valdes J J, Kos L,Tsoukias N and Agarwal A 2009 Polylactide-caprolactonecopolymer: mechanical strengthening and interaction withhuman osteoblasts in vitro ACS Appl. Mater. Interfaces1 2470–6

[7] Hilt J Z and Peppas N A 2005 Microfabricated drug deliverydevices Int. J. Pharm. 306 15–23

[8] Cai N, Wong C C, Gong Y X, Tan S C W, Chan V and Liao K2010 Modulating cell adhesion dynamics on carbonnanotube monolayer engineered with extracellular matrixproteins ACS Appl. Mater. Interfaces 2 1038

[9] Tutak W, Chhowalla M and Sesti F 2010 The chemical andphysical characteristics of single-walled carbon nanotubefilm impact on osteoblastic cell response Nanotechnology21 315102

[10] Arima Y and Iwata H 2007 Effect of wettability and surfacefunctional groups on protein adsorption and cell adhesionusing well-defined mixed self-assembled monolayersBiomaterials 28 3074–82

[11] Aronov D, Rosen R, Ron E Z and Rosenman G 2006 Tunablehydroxyapatite wettability: effect on adhesion of biologicalmolecules Process Biochem. 41 2367–72

[12] Bagno A and Di Bello C 2004 Surface treatments androughness properties of Ti-based biomaterials J. Mater. Sci:Mater. Med. 15 935–49

[13] Xie H-G, Li X-X, Lv G-J, Xie W-Y, Zhu J, Luxbacher T,Ma R and Ma X-J 2010 Effect of surface wettability andcharge on protein adsorption onto implantablealginate-chitosan-alginate microcapsule surfaces J. Biomed.Mater. Res. A 92A 1357–65

[14] Mu Q, Du G, Chen T, Zhang B and Yan B 2009 Suppression ofhuman bone morphogenetic protein signaling bycarboxylated single-walled carbon nanotubes ACS Nano3 1139–44

[15] Hao L and Lawrence J 2007 Wettability modification and thesubsequent manipulation of protein adsorption on ATi6Al4V alloy by means of CO2 laser surface treatmentJ. Mater. Sci: Mater. Med. 18 807–17

[16] Tan J L, Tien J and Chen C S 2002 Microcontact printing ofproteins on mixed self-assembled monolayers Langmuir18 519–23

[17] Koennings S, Berie A, Tessmar J, Blunk T and Goepferich A2007 Influence of wettability and surface activity on releasebehavior of hydrophilic substances from lipid matricesJ. Control. Release 119 173–81

[18] Maretschek S, Greiner A and Kissel T 2008 Electrospunbiodegradable nanofiber nonwovens for controlled release ofproteins J. Control. Release 127 180–7

[19] Nygren H, Alaeddin S, Lundstrijm I and Magnusson K-E 1994Effect of surface wettability on protein adsorption and lateraldiffusion. Analysis of data and a statistical model Biophys.Chem. 49 263–72

[20] Sigal G B, Mrksich M and Whitesides G M 1998 Effect ofsurface wettability on the adsorption of proteins anddetergents J. Am. Chem. Soc. 120 3464–73

[21] Xu L-C and Siedlecki C A 2007 Effects of surface wettabilityand contact time on protein adhesion to biomaterial surfacesBiomaterials 28 3273–83

[22] Hedlund J, Lundgren A, Lundgren B and Elwing H 2009A new compact electrochemical method for analyzingcomplex protein films adsorbed on the surface of modifiedinterdigitated gold electrodes Sensors Actuators B142 494–501

[23] Ivanov A E, Ekeroth J, Nilsson L, Mattiasson B, Bergenstahl Band Galaev I Y 2006 Variations of wettability and proteinadsorption on solid siliceous carriers grafted withpoly(N-isopropylacrylamide) J. Colloid Inteface Sci.296 538–44

[24] Vasilev K, Mierczynska A, Hook A L, Chan J, Voelcker N Hand Short R D 2010 Creating gradients of two proteins bydifferential passive adsorption onto a PEG-density gradientBiomaterials 31 392–7

[25] Kawasaki K, Aihara M, Honmo J, Sakurai S, Fujimaki Y,Sakamoto K, Fujimaki E, Wozney J M and Yamaguchi A2003 Effects of recombinant human bone morphogeneticprotein-2 on differentiation of cells isolated from humanbone, muscle, and skin Bone 23 223–31

[26] Wang E A et al 1990 Recombinant human bone morphogeneticprotein induces bone formation Proc. Natl Acad. Sci. USA87 2220–4

[27] Ye S-H, Johnson C A, Woolley J R, Snyder T A, Gamble L Jand Wagner W R 2009 Covalent surface modification of atitanium alloy with a phosphorylcholine-containingcopolymer for reduced thrombogenicity in cardiovasculardevices J. Biomed. Mater. Res. A 91A 18–28

[28] Han Z J, Tay B K, Shakerzadeh M and Ostrikov K 2009Superhydrophobic amorphous carbon/carbon nanotubenanocomposites Appl. Phys. Lett. 94 223106

[29] Cvelbar U, Chen Z, Sunkara M K and Mozetic M 2008Spontaneous growth of superstructure α-Fe2O3 nanowire andnanobelt arrays in reactive oxygen plasma Small 4 1610–4

[30] Ostrikov K, Levchenko I, Cvelbar U, Sunkara M andMozetic M 2010 From nucleation to nanowires: asingle-step process in reactive plasmas Nanoscale 2 2012–27

[31] Shieh J, Hou F J, Chen Y C, Chen H M, Yang S P, Cheng C Cand Chen H L 2010 Robust airlike superhydrophobicsurfaces Adv. Mater. 22 597–601

[32] Han Z, Tay B, Tan C, Shakerzadeh M and Ostrikov K K 2009Electrowetting control of cassie-to-wenzel transitions insuperhydrophobic carbon nanotube-based nanocompositesACS Nano 3 3031–6

7

Nanotechnology 22 (2011) 295712 Z J Han et al

[33] http://www.peprotech.com[34] Peel S A F, Hu Z M and Clokie C M L 2003 In search of the

ideal bone morphogenetic protein delivery system: in vitrostudies on demineralized bone matrix, purified, andrecombinant bone morphogenetic protein J. Craniofac. Sur.14 284–91

[35] Levchenko I and Ostrikov K 2007 Nanostructures of variousdimensionalities from plasma and neutral fluxes J. Phys. D:Appl. Phys. 40 2308–19

[36] Mariotti D, Svrcek V and Kim D-G 2007 Self-organizednanostructures on atmospheric microplasma exposedsurfaces Appl. Phys. Lett. 91 183111

[37] Keidar M, Levchenko I, Arbel T, Alexander M, Waas A M andOstrikov K 2008 Increasing the length of single-wall carbonnanotubes in a magnetically enhanced arc discharge Appl.Phys. Lett. 92 043129

[38] Chiang W-H and Sankaran R M 2009 Linking catalystcomposition to chirality distributions of as-grownsingle-walled carbon nanotubes by tuning Nix Fe1−x

nanoparticles Nat. Mater. 8 882–6[39] Ostrikov K 2005 Reactive plasmas as a versatile

nanofabrication tool Rev. Mod. Phys. 77 489–511[40] Wirth C T, Hofmann S and Robertson J 2008 Surface

properties of vertically aligned carbon nanotube arraysDiamond Relat. Mater. 17 1518–24

[41] Ahuja A, Taylor J A, Lifton V, Sidorenko A A, Salamon T R,Lobaton E J, Kolodner P and Krupenkin T N 2008Nanonails: a simple geometrical approach to electricallytunable superlyophobic surfaces Langmuir 24 9–14

[42] Yang P, Huang N, Leng Y X, Chen J Y, Fu R K Y,Kwok S C H, Leng Y and Chu P K 2003 Activation ofplatelets adhered on amorphous hydrogenated carbon(a-C:H) films synthesized by plasma immersion ionimplantation-deposition (PIII-D) Biomaterials 24 2821–9

[43] Gonzalez-Garcıa C, Sousa S R, Moratal D, Rico P andSalmeron-Sanchez M 2010 Effect of nanoscale topographyon fibronectin adsorption, focal adhesion size and matrixorganisation Colloids Surf. B 77 181–90

[44] Geiger M, Li R H and Friess W 2003 Collagen sponges forbone regeneration with rhBMP-2 Adv. Drug Deliv. Rev.55 1613–29

[45] Xu S, Long J, Sim L, Diong C H and Ostrikov K 2005 RFplasma sputtering deposition of hydroxyapatite bioceramics:synthesis, performance, and biocompatibility PlasmaProcess. Polym. 3 373–90

[46] Winn S R, Uludag H and Hollinger J O 1998 Sustained releaseemphasizing recombinant human bone morphogeneticprotein-2 Adv. Drug Deliv. Rev. 31 303–18

[47] Clokie C M L and Urist M R 2000 Bone morphogenetic proteinexcipients: comparative observations on poloxamer Plast.Reconstruct. Surg. 105 628–37

[48] Dumortier G, Grossiord J L, Agnely F and Chaumeil J C 2006A review of poloxamer 407 pharmaceutical andpharmacological characteristics Pharm. Res. 23 2709–28

[49] Vroman L, Adams A L, Fischer G C and Munoz P C 1980Interaction of high molecular weight kininogen, factor XII,and fibrinogen in plasma at interfaces Blood 55 156–9

[50] Goppert T M and Muller R H 2005 Adsorption kinetics ofplasma proteins on solid lipid nanoparticles for drugtargeting Int. J. Pharm. 302 172–86

[51] Zhao L, Hu L, Huo K, Zhang Y, Wu Z and Chu P K 2010Mechanism of cell repellence on quasi-aligned nanowirearrays on Ti alloy Biomaterials 31 8341

[52] Wu Y, Zitelli J P, TenHuisen K S, Yu X and Libera M R 2011Differential response of Staphylococci and osteoblasts tovarying titanium surface roughness Biomaterials 32 951

[53] Aggarwal N, Lawson K, Kershaw M, Horvath R and Ramsden J2009 Protein adsorption on heterogeneous surfaces Appl.Phys. Lett. 94 083110

8