As

MCa

Hb

a

ARRAA

KPFPSC

1

cimghpcattid

h0

Colloids and Surfaces B: Biointerfaces 122 (2014) 134–142

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces

jo ur nal ho me p ag e: www.elsev ier .com/ locate /co lsur fb

correlation study of protein adsorption and cell behaviors onubstrates with different densities of PEG chains

ingcong Suna, Jun Denga, Zengchao Tangb, Jindan Wua, Dan Lib, Hong Chenb,hangyou Gaoa,∗

MOE of Key Laboratory of Macromolecular Synthesis and Functionalization, Department of Polymer Science and Engineering, Zhejiang University,angzhou 310027, ChinaMacromolecules and Biointerface Lab, College of Chemistry, Chemical Engineering and Materials Science, Soochow University, Suzhou 215123, China

r t i c l e i n f o

rticle history:eceived 27 March 2014eceived in revised form 3 June 2014ccepted 19 June 2014vailable online 1 July 2014

eywords:rotein adsorptionibronectinoly(ethylene glycol)mooth muscle cellsellular behaviors

a b s t r a c t

The adsorption of proteins, in particular fibronectin (Fn), was studied on poly(ethylene glycol) (PEG,5 kDa)-grafted surfaces, and was correlated with the adhesion behaviors of smooth muscle cells (SMCs).The PEG molecules were covalently grafted on aldehyde-activated substrates with different densitiesof amino groups. The thickness of PEG layer increased nearly 10 fold in a hydrated state, reaching to27 nm on the surface of highest PEG chain density with a brush configuration. On the lower PEG-graftedsurfaces, however, the PEG molecules adopted a mushroom configuration. The adsorption of Fn withoutand with the competition of bovine serum albumin (BSA) and serum was studied by using ellipsometry,fluorescence microscopy and radio-labeling techniques. The adsorption amount of Fn in serum decreasedinitially with increased PEG chain density until 0.12 chains/nm2 PEG, and then slightly increased on the0.29 chains/nm2 PEG. A series of protein preadsorption experiments were carried out under differentconditions before SMCs culture in vitro. Compared with those substrates without Fn preadsorption, the

cell adhesion and spreading were significantly enhanced on all the PEG surfaces preadsorbed with Fn andserum, although they overall decreased along with the increase of PEG grafting density. The adhesion forceof Fn decreased monotonously with the increase of PEG grafting density, which was in accordance withthe cell adhesion force. The correlation between the PEG-grafted surfaces, Fn adsorption, and cellularbehaviors is finally suggested.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

The surface chemistry or physical structure, wettability, surfaceharge, topography, and viscoelasticity of biomaterials have a bigmpact on many cellular events such as cell adhesion, spreading,

igration and differentiation [1–11]. Particularly, poly(ethylenelycol) (PEG) has high hydrophilicity and the so-called steric-indrance effect. The surface-anchored PEG molecules can resistrotein adsorption (the anti-fouling property), and as a result,ell adhesion [12–16]. However, the interplay between the PEG-nchored surface and biological response is highly dependent onhe PEG density. For example, Hsu et al. found that the cell migra-

ion rate initially increases along with the increase of PEG contentn poly(�-caprolactione) (PCL)-b-PEG block copolymer, and thenecreases when the PEG content is higher than 77% [17]. They

∗ Corresponding author. Tel.: +86 571 87951108; fax: +86 571 87951108.E-mail addresses: cygao@mail.hz.zj.cn, cygao@zju.edu.cn (C. Gao).

ttp://dx.doi.org/10.1016/j.colsurfb.2014.06.041927-7765/© 2014 Elsevier B.V. All rights reserved.

attributed this phenomenon to the presence of nanoislands formedby phase separation, which greatly improve the surface roughness.Our group found that vascular smooth muscle cells (VSMC) migratefastest on the PEG brushes with an appropriate grafting density[18].

However, the underlying molecular mechanisms are essentiallytied with the protein adsorption behaviors. It is known that when abiomaterial contacts with physiological mediums such as serum, itssurface shall be instantaneously covered by water and salt ions, andthen by a wide variety of plasma proteins including albumin andcell growth factors [19]. In many cases, the cells can recognize theadsorbed cell growth factors or adhesive proteins through recep-tors, which in turn mediate the cell behaviors such as adhesion andmigration [20–23]. So far numerous experimental and theoreticalefforts have been focused on the investigation and understanding

of certain adhesive proteins and their effects on cell attachmentand other responses [24–30]. The type [21,31], adsorption amount[32], conformation/orientation [33,34], and bioactivity of the pro-tein layer are known to be modulated by the surface properties

es B: B

obia

t[oisatocsavduacacstbc

2

2

53m(otU

2

ldfoa[

2

wAm(t

aJuawf

WIZARD2 automatic �-Counters (Perkin-Elmer 1480, U.S.A.).

M. Sun et al. / Colloids and Surfac

f biomaterials. For example, Jia et al. modified titanium dioxidey electrostatic adsorption of poly(l-lysine)-g-PEG, and clarified

ts influence on the protein (fibrinogen and albumin) conformationnd orientation, and thereby the adhesion of fibroblasts [35].

Although the cell adhesion and spreading, and protein adsorp-ion on biomaterials surfaces have been independently studied26,36,37], the correlation between the physiochemical propertiesf biomaterials, protein adsorption behaviors, and cellular behav-ors remains far from understanding due to the complexity of theystem. In this work, ˛-methoxy poly(ethylene glycol-ω-amino)re grafted onto substrates with variable densities. Different fromhe many previous studies that focus on the antifouling propertiesf PEG, here the grafted PEG with variable densities allows thehange of adsorption of plasma proteins, and thereby cell adhe-ion and migration to different degrees. It is demonstrated thatmong the various types of adhesive proteins such as laminin [38],itronectin and fibronectin (Fn), the Fn secreted by SMCs highlyepends on the PEG densities [18], and thereby shall be partic-larly concerned. Therefore, in this study we shall focus on thedsorption of Fn on the PEG-grafted surfaces with and without theompetitive protein, albumin, which is most abundant in plasma,nd the Fn adhesion force which contributes significantly to theell adhesion strength. SMCs shall be cultured on the PEG-graftedurfaces at variable conditions. By this way, it is able to figure outhe correlation between the PEG surface, Fn adsorption, and cellehaviors, substantiating the decisive role of Fn, and interplay ofells and proteins.

. Experimental

.1. Materials

�-Methoxy-poly(ethylene glycol-�-amino) (mPEG-NH2, Mw

kDa) was purchased from Jiangsu PEG Co., Ltd (China).-Triethoxysilylpropylamin (APS) and 3-(trimethoxysilyl)propylethacrylate (TMSPMA) were purchased from J&K Company

China). All the chemicals were of analytical grade and used with-ut further purification if not otherwise stated. The water used inhe experiments was purified by a Milli-Q water system (Millipore,.S.A.).

.2. PEG grafting

Different ratios of TMSPMA and APS were covalently immobi-ized onto glass or polished silicon wafer slides by using a methodescribed previously [18]. The mPEG-NH2 was grafted onto the sur-aces with different densities via glutaraldehyde (GA) coupling. Inrder to avoid any possible binding with proteins or cells, the unre-cted aldehyde groups, if any, were blocked by lysine molecules39].

.3. Surface analysis

The chemical compositions of the surfaces after silanizationere detected by X-ray photoelectron spectrometry (XPS) using anxis Ultra spectrometer (Kratos Analytical, UK) with a monochro-ated Al K� source at different pass energies for survey spectra

160 eV) and core level spectra (80 eV). Data were analyzed withhe Kratos Vision Processing Software.

The dry thickness of PEG layer was measured by variable-ngle spectroscopic ellipsometry (VASE, model M2000D,.A.WoollamInc., U.S.A.). The samples in air were measured

nder ambient conditions at three angles of incidence (65◦, 70◦

nd 75◦) in the spectral range of 300–1700 nm. The data were fittedith the WVASE32 analysis software using a multilayer model

or a 2 nm silicon dioxide layer on 1 mm silicon layer, and the

iointerfaces 122 (2014) 134–142 135

PEG adlayer thickness was calculated by a Cauchy model (fixingAn = 1.45, Bn = 0.01, and Cn = 0). The measurement in water andphosphate buffered saline (PBS) was carried out in a liquid cell atan incidence angle of 75◦ in the spectral range of 350–1000 nm. Allthe measurements were carried out at least 3 times.

According to Zdyrko et al. [40], the brush thickness (h) in a drystate and the grafting density (�) have the following relationship:

� =(

�hNA

MW

)(1)

where, � is the density of PEG (1.09 g/cm3), NA is the Avogadro’snumber, and Mw is the molecular weight of PEG (5 kDa).

Moreover, the PEG density can be further converted into thedistance between two adjacent PEG chains (L) according to Eq. (2)[35]:

L =(

2√3�

)0.5(2)

It is known that the conformation of grafted PEG molecules isdetermined by L/2Rg, where Rg is the radius of gyration of PEGchains. It is calculated to be 2.36 nm for PEG 5 kDa according toan empirical equation (3) based on static light-scattering measure-ments [41,42].

Rg = s(

Xn

6

)0.5(3)

where Xn is the degree of polymerization and s is the segment lengthof PEG, which is about 0.35 nm [41].

2.4. Protein adsorption

2.4.1. Protein adsorption amount: Ellipsometry and fluorescencemicroscopy (FM)

The thickness of the protein adlayer was determined by ellip-sometry after subtracting the thickness of corresponding PEG layerfrom the total thickness. The optical constants of both the poly-mer and protein layers were fixed using the Cauchy model. FM wasapplied to measure the fluorescence intensity of adsorbed BSA or Fnon the PEG layer at a fixed exposure time of 1.2 s. At least 10 imageswere analyzed by Image J software to obtain the grey values. For sin-gle protein adsorption, the substrates were incubated with 3 mg/mlrhodamine B isothiocyanate (RITC)-labeled BSA (RITC-BSA, Sigma-Aldrich, U.S.A.) or 30 �g/ml RITC-labeled Fn (RITC-Fn, R&D Systems,U.S.A.) at 37 ◦C for 2 h, and followed by washing with PBS andwater. All experiments were carried out in triplicate. The compet-itive adsorption solution was consisted of RITC-BSA and unlabeledFn, or unlabeled BSA and RITC-Fn at the same concentration ratiounder the same experimental conditions.

2.4.2. Serum protein adsorption: Radio labelingFn was labeled with 125I using the iodine monochloride (ICl)

method as reported previously [43,44]. In order to study the Fnadsorption from buffer, the labeled proteins were mixed with unla-beled ones to give a total concentration of 30 �g/ml. To study theFn adsorption from serum, the labeled Fn molecules were added to10% serum at the same concentration. After the substrate surfaceswere equilibrated in PBS for 12 h prior to the adsorption experi-ment, they were incubated in the protein solution at 37 ◦C for 2 h,rinsed 3 times in PBS (10 min each), sucked with filter papers, andthen transferred to clean tubes for radioactivity measurement by

2.4.3. Protein adhesion force: Atomic force microscopy (AFM)Adhesion force of Fn on the PEG-grafted surface was mea-

sured by SPA400 AFM (Seiko Instruments Inc., Japan) using

136 M. Sun et al. / Colloids and Surfaces B: Biointerfaces 122 (2014) 134–142

Table 1Physiochemical properties of surface-grafted PEG brushes with different densities.

Reaction Time with TMSPMA (min) N/C molar ratio (silanized substrate)* Density � (chains/nm2)** Distance L (nm) L/2Rg Configuration

0 0.057 0.29 ± 0.030 2.0 ± 0.08 0.43 Brush5 0.045 0.12 ± 0.0079 3.1 ± 0.09 0.66 Brush20 0.025 0.04 ± 0.005 5.4 ± 0.21 1.14 Mushroom

* The absolute N/C ratios are all smaller than those of the previous values measured on the corresponding substrates prepared at the same conditions [18], due to thedifference in contamination of environmental carbon such as the pretreatment conditions and exposure time to the atmosphere before detection [61,62]. Nevertheless, therelative N/C ratios (e.g. normalized to the blank control) are rather close with a variation less than 20%. Therefore, the relative values can be reasonably used to compareb ard d

ing det

Fneitifeci33w

Ptii

2

wciUfitwt

d1ptcTswTalwc

2

wAs

etween different samples in one batch measurement. Values are mean ± SD (stand** Here, the PEG grafting mass expressed in literature [18] is replaced by the graft

he protein adsorption behaviors. Values are mean ± SD; n = 9.

n-functionalized cantilever. The commercially available siliconitride triangular cantilever (200 �m long) coated with Au andquipped with a pyramid-like-shaped tip, whose spring constants 0.08 N/m, was chosen for the measurement. The cantilever wasreated with oxygen plasma for 5 min, and was then incubatedn a 1.0 mmol/l 11-mercaptoundecanoic acid/ethanol solutionor 1 h to introduce free carboxyl groups. After washed withthanol and water, the cantilever was immersed in a solutionontaining 0.1 mol/l 1-ethyl-3-(3-dimethylaminopropyl) carbodi-mide hydrochloride and 0.05 mol/l N-hydroxysuccinimide for0 min. After rinsed with water the cantilever was incubated in0 �g/ml Fn solution (pH 7.4) at 37 ◦C for 1.5 h. Finally, it wasashed with PBS and stored at 4 ◦C before use.

The adhesion force between the immobilized proteins andEG layers was defined by the maximum pull-off force inhe force–distance curve during the retraction of the protein-mmobilized tip. All the measurements were repeated three timesn PBS.

.5. Protein pre-adsorption experiments and cell culture

The protein preadsorption experiments on the PEG surfacesith different grafting densities were carried out at the following

onditions. (a) Non preadsorption: VSMCs were cultured directlyn serum-free Dulbecco’s modified eagle medium (DMEM, Gibco,.S.A.). (b) Protein preadsorption samples: the substrates wererstly incubated with 30 �g/ml Fn solution for 2 h, and then washedhree times with PBS. (c) Serum preadsorption samples: all samplesere first incubated in a 10% serum solution for 2 h, and then rinsed

hree times in PBS.Before protein preadsorption experiments, the substrates with

ifferent PEG grafting densities were sterilized in 75% ethanol for h, followed with thoroughly washing in PBS (pH 7.4), and thenlaced in the wells of a 24-well culture plate (Corning, U.S.A.). Allhe SMCs were cultured on the substrates with a density of 2 × 104

ells/well in serum-free DMEM for 8 h at standard conditions [18].he images of cells were recorded under a fluorescence micro-cope (IX81, Olympus, Japan) after being incubated and stainedith 5 �g/ml fluorescein diacetate (FDA, Sigma-Aldrich) for 5 min.

he cell number and spreading area were analyzed, and all thebove experiments were repeated for at least one time. Cell out-ines were detected using Image J analysis software, and cell area

as calculated based on the number of pixels covered by theell.

.6. Statistical analysis

The significant difference between groups is analyzed using one-ay analysis of variance (ANOVA) (for two groups) and two-wayNOVA (for more than two groups) in the Origin software, and thetatistical significance was set as p < 0.05.

eviation); n (number of experiments) = 9.nsity, which is beneficial for the correlation of molecular conformation of PEG and

3. Results

3.1. PEG-grafted surfaces

In order to fabricate the PEG-grafted surfaces with different den-sities, the substrates were reacted with TMSPMA for different time,and then backfilled with APS, resulting in different densities ofamino groups. Changes in the chemical compositions of surfacesafter TMSPMA/APS reaction were analyzed by XPS (Table 1). Thenitrogen element could only be detected on the APS-treated sur-faces, enabling the use of N/C ratio to represent the relative contentof NH2 groups. The data show that along with the prolongation ofTMSPMA reaction time the content of NH2 groups decreased, andthereby the grafting density of mPEG-NH2 will be lower.

Ellipsometry characterization results of PEG grafted surfaceswith different reaction time of TMSPMA are shown in Fig. 1a andb. The dry thickness of PEG layer decreased rapidly within the first5 min, and then slowly along with the reaction time of TMSPMA,revealing the difference in grafting density. Fig. 1b shows that thehydrated thickness of PEG chains was similarly reduced after longerTMSPMA reaction time. However, the thickness of hydrated PEGchains dramatically increased, which is about 10 times thicker thanthat of the corresponding dry PEG layer. The maximum hydratedPEG thickness was measured to be 27.2 ± 2.5 nm on the surfaceof highest PEG chain density, which decreased to 18.9 ± 3.3 nm inPBS. Indeed, all the thicknesses of PEG layers in PBS were smallerthan their corresponding counterparts in water. The salts in PBSmay weaken the hydrogen bonding between PEG segments andwater molecules, reducing the steric-hindrance effect and therebyfavoring the collapse of PEG molecules to some extent. As a resultof variation of the PEG densities, the water contact angles variedcorrespondingly with the minimum and maximum values of 31◦

and 44◦ on the highest and lowest levels of PEG-grafted surfaces,respectively.

Table 1 summarizes the physiochemical properties of thesurface-grafted PEG brushes with different densities. According tothe above calculation results and analysis, the molecular confor-mation of PEG is illustrated in Fig. 1c. At the lowest grafting density(� < 0.12), the L/2Rg > 1, implying that the adjacent PEG molecules inthe hydrated state do not overlap and can rotate randomly withoutdisturbance, forming a mushroom-like regime. At the medium andhigh grafting densities (� ≥ 0.12), the L/2Rg = 0.43 or 0.66, revea-ling that the PEG molecules are in a crowded state, and therebytheir rotation shall be interfered with each other, enabling the PEGchains to stretch out and forming a so-called “brush” regime. Here,no state of L/2Rg = 1 appears at the present grafting density of PEG,where the PEG molecules start to overlap and show influence witheach other.

3.2. Protein adsorption

Quantitative adsorption of Fn and its most abundant competi-tor in plasma, albumin, was measured by ellipsometry, obtaining

M. Sun et al. / Colloids and Surfaces B: Biointerfaces 122 (2014) 134–142 137

Fig. 1. (a) Thickness of dry PEG layer prepared on substrates of different TMSPMA reaction time. (b) Thickness of hydrated PEG layer in water and PBS as a function ofTMSPMA reaction time. The refractive indices were fixed as 1.35 and 1.36 in water and PBS for the ellipsometry measurement, respectively. (c) Schematic of side view of PEGchain arrangement in the brush regime (L/2Rg < 1, upper and middle panels) and in the mushroom regime (L/2Rg > 1, lower panel). Data are mean ± SD; n = 9.

Fig. 2. (a) Ellipsometry measurement of protein adsorption on PEG-grafted surfaces after incubated in 3 mg/ml BSA, and 30 �g/ml Fn solution, respectively. FM measurementof relative adsorption of (b) 30 �g/ml Fn, and (c) 3 mg/ml BSA on PEG surfaces with and without the competitive counterpart, respectively. All the data are normalized tothose of the PEG 0 surface. (d) Radiolebeling measurement of adsorption of 30 �g/ml Fn in PBS or 10% serum on PEG-grafted surfaces with different densities. * Significantdifference between groups indicated by the lines at p < 0.05. Error bars are ± SD; n ≥ 12.

1 es B: B

tbmo0oaehtw

FwsctwrodmtlcPspb(ngstosOFctbfrid

stFssFdPsetdasTihpot

38 M. Sun et al. / Colloids and Surfac

he protein layer thickness which can be transferred into amountased on the Stenberg and Nygren formula [45]. The adsorptionass of both BSA and Fn decreased rapidly with the increase

f PEG grafting density when the PEG chain density was below.12 chains/nm2, and then slowly with a final value of ∼10 ng/cm2

n the 0.29 chains/nm2 PEG surface (Fig. 2a). Comparatively, thebsolute amount of Fn was larger than that of BSA on each surfacexcept for the highest PEG-grafted one, possibly due to the prettyigh Mw of Fn (440 kDa) over that of BSA (67 kDa). This result revealshe gradually enhanced repelling effect of the PEG-grafted surfacesith the increase of PEG chain density.

Next, the competitive adsorption was carried out between then and BSA with a fixed concentration ratio of Fn/BSA (1/100/w), and the Fn concentration was fixed at 30 �g/ml, which is

ame as that in the physiological condition [46]. Here fluores-ence microscopy (FM) was used to semi-quantitatively quantifyhe adsorption amounts of each type of proteins without andith the competitive counterpart. Fig. S2 shows that the FM

esults of single protein adsorption match very well with thosef the ellipsometry (Fig. 2a) in terms of the alteration ten-ency and the relative values, demonstrating the reliability of thisethod. Fig. 2b shows that in the presence of BSA, the adsorp-

ion amount of Fn was reduced sharply on the surfaces withower PEG-grafting densities (For example, 17% and 36% of theorresponding controls on the surfaces 0 and 0.12 chains/nm2

EG), and remained almost invariant on the 0.29 chains/nm2 PEGurface. Consequently, the absolute amount of Fn on the com-etitive condition (with BSA only) had no significant differenceetween all the surfaces regardless of the PEG grafting densityp > 0.05). By contrast, Fig. 2c shows that the presence of Fn didot significantly suppress the adsorption of BSA on all the PEG-rafted surfaces, except on the bare control slide (70% of theingle BSA). These results demonstrate that compared with Fn,he BSA molecules more easily adsorb onto the substrates with-ut PEG or with a smaller density of PEG, possibly due to theirmaller size and stronger affinity to the less anti-fouling surfaces.n the highest PEG-grafted surface, however, the adsorption ofn is not influenced by the repelling effect of the PEG chains,onveying the relatively stronger affinity of Fn than that of BSAo this anti-fouling surface. This conclusion is partially supportedy the slight reduction of BSA on the 0.29 chains/nm2 PEG sur-ace in the presence of Fn too. Therefore, the presence of BSAeduces the adsorption amount of Fn to different degrees depend-ng on the antifouling ability of the surfaces, i.e. PEG graftingensity.

Furthermore, the competitive adsorption of Fn in serum wastudied by radio labeling (Fig. 2d), which practically correlates withhe real application of biomaterials. Again, the adsorption of singlen decreased along with the increase of PEG density on the sub-trate (p < 0.05), although the decreasing tendency was relativelylower compared with those obtained by ellipsometry (Fig. 2a) andM (Fig. 2b and c). In serum, the adsorption amount of Fn wasramatically decreased with a fold of 20 (on the 0.29 chains/nm2

EG surface) to 40 (on the 0 chains/nm2 PEG surface, i.e. TMSPMAurface) (p < 0.05), whose degree is far larger than that in the pres-nce of BSA only (6 fold the maximum). Moreover, along withhe increase of PEG density the adsorption amount of Fn in FBSecreased from 9.7 ± 2.8 ng/cm2 to 3.3 ± 2.6 ng/cm2 on the 0.04nd 0.12 chains/nm2 PEG surfaces, respectively, but then increasedlightly to 6.2 ± 1.7 ng/cm2 on the 0.29 chains/nm2 PEG surface.his abnormal increase will be discussed later. Taking all the resultsnto consideration, one can conclude that adsorption of Fn in serum

as smaller variation on all the PEG-grafted surfaces, and otherlasma proteins contribute heavily to the competitive adsorptionf Fn besides albumin (which contributes about 18% of the compe-ition effect).

iointerfaces 122 (2014) 134–142

3.3. Adhesion forces of Fn and cells

It is reasonably assumed that the cell adhesion force is closelyassociated with the adsorbed proteins, especially the most abun-dant adhesive protein, Fn, in plasma. To disclose the relationship,the adsorption force of Fn on the PEG-grafted surfaces was deter-mined by AFM according to the method reported previously(Fig. 3a) [47]. Fig. S3 shows the representative histograms of Fnadhesion forces on different surfaces, whose values were averagedand shown in Fig. 3b. The average adhesion force of Fn decreasedfrom ∼90 pN to 10 pN on the TMSPMA and 0.29 chains/nm2 PEG-grafted surfaces, respectively. Fig. 3b further compares the Fnadsorption force with the cell adhesion force, revealing that theyhave the similar alteration tendency on the PEG-grafted surfacesalthough the cell adhesion force is much larger (300 pN on theTMSPMA surface). This result suggests that the adhesion strengthof a cell is greatly contributed by the adhesion strength of the previ-ously adsorbed Fn, which is known to interact with cell surface viathe integrin protein. It is likely that the cells may be removed fromthe substrate together with the tightly conjugated Fn (Of courseother adhesive proteins cannot be excluded) by the centrifugationforce, and thereby the cell adhesion force is greatly influenced bythe interaction between Fn and the substrate. It has to mention thatthe absolute values of the Fn adhesion and cell adhesion forces maynot be comparable directly since they are measured by differentmethods. Nevertheless, the general similar tendency suggests theintrinsic relationship between the Fn adsorption and cell adhesion.

3.4. Cell adhesion and spreading

It is known that cell adhesion, spreading and migration behav-iors are highly dependent on the protein adsorption. To elucidatethis influence, a series of protein preadsorption experiments werecarried out under different conditions before the SMCs were cul-tured in vitro. Figure S4 shows the morphology of SMCs beingcultured on the PEG-grafted substrates in serum-free mediumwithout and with Fn or serum preadsorption, respectively. On thePEG-grafted surfaces the total cell number (Fig. 4a) and average cellspreading area (Fig. 4b) show a generally decreasing tendency alongwith the increase of PEG chain density regardless of the preadsorp-tion conditions, suggesting that the mPEG chains of larger densitiesare more efficient in resisting cell adhesion and spreading. Detailanalysis finds that the cell adhesion number shows insignificantdifference on all the PEG-grafted substrates without preadsorp-tion of any proteins (p > 0.05). When the cells were cultured onthe substrates preadsorbed with Fn and FBS, both the cell numbersand spreading areas were significantly improved (p < 0.05) exceptfor the cell number on the 0.29 chains/nm2 PEG surface. Compar-atively, the substrates preadsorbed with FBS had a stronger effecton promoting the cell adhesion and spreading. These results con-firm the decisive role of the adsorbed proteins, in particular the Fn,on the cell behaviors. And this result is consistent with the pre-vious study that the polyelectrolyte multilayers terminated withFn can promote SMCs attachment by increasing the cell spreadingarea and number of pseudopodia [48]. Basically, the cell behaviorsare in good agreement with the adsorption behavior of Fn on thePEG-grafted surfaces, and other types of plasma proteins shouldalso take a role to some extent in promoting SMCs attachment.

4. Discussion

Since the adsorption behavior of Fn from serum matches wellwith the real situation of the PEG-grafted substrates for cell culture,the data in Fig. 2d are used for discussion of the correlation betweenbiomaterials, protein adsorption, and cell behaviors. Previous study

M. Sun et al. / Colloids and Surfaces B: Biointerfaces 122 (2014) 134–142 139

Fig. 3. (a) Representative force-versus-distance curves between Fn-immobilized cantilever and substrates with (1) TMSPMA, and PEG chains of different densities: (2) 0.04,( d cellt aged fr . Error

fadg[Srs

dwrsssmt(sd

e

Fea

3) 0.12 and (4) 0.29 chains/nm2. (b) Relationship between adhesion force of Fn anhereof too) and PEG chain density. The Fn adhesion force of one sample was averemains same for all the measurement because of the same treatment of these tips

ound that the gene expression of Fn by SMCs had a similar alter-tion tendency as that of the Fn adsorption (Fig. 2d), i.e. a loweregree of Fn gene was expressed when being cultured on the PEG-rafted surface with medium densities, such as 0.12 chains/nm2

18] (Fig. S5). On this type of surface, the SMCs migrated fastest (Fig.6). This positive correlation demonstrates that the cell migrationate is intrinsically governed by the adsorbed Fn on the PEG-graftedubstrates.

It is known that many cellular behaviors such as migration areominantly governed by the initial cell adhesion and spreading,hich are influenced by the protein adsorption behaviors too. Fig. 4

eveals that the difference in Fn adsorption on the PEG-graftedurfaces is well correlated with the different cell adhesion andpreading properties, confirming the significant role of this adhe-ive protein. However, the cell adhesion and spreading decreasedonotonously on all the Fn or FBS-pretreated surfaces along with

he increase of PEG density, regardless of the culture conditionswith or without serum-containing medium) [18], which are incon-

istent with the smaller adsorption amount of Fn on the mediumensity of PEG surface.

Protein can adsorb onto polymer brushes by mainly three differ-nt modes depending on the distance between neighboring chains,

Cell

are

a (

µµm2)

a

0

20

40

60

80

Cell

den

sit

y (

cell

s/m

m2)

PEG cha in dens ity (cha ins/nm2)

No pre-adsorpti on

Fn

FBS

0 0.04 0.12 0.29

*

** *

****

ig. 4. Cell density (a) and spreading area (b) of SMCs after being cultured on PEG-graftedither not treated or pre-treated in 30 �g/ml Fn and 10% serum for 2 h, respectively, folre ± SD with n (number of cell images) ≥ 20 and n ≥ 100 for cell density and spreading ar

adhesion force (Data from literature [15]. The experimental details can be foundrom at least 100 measurements. The total number of Fn modified on the AFM tips

bars are ± SD; n ≥ 100 for protein adhesion force measurement.

molecular weight, and the protein–surface attraction. The primaryadsorption (2Rg < <L) involves an attractive contact with the sur-face, while the secondary adsorption (2Rg > >L) occurs at the outeredge of the brushes [49,50]. At the intermediate chain densities,the proteins generally penetrate into the brushes but not directlyto the substrate through the weak protein-PEG attraction (ternaryadsorption) [28,51].

In this study the polymerization degree of PEG is notlarge enough to exclude some additional factors such asthe protein dimension and conformation. BSA is commonlymodeled as a “cigar-like” prolate ellipsoid with a dimension of14 nm × 4 nm × 4 nm [50]. Therefore, BSA can directly contact withthe substrate (primary adsorption) considering that the minor axisis shorter than the distance between PEG chains at the low PEG-grafted surface, and would mainly adopt the ternary adsorptionmode at medium and high PEG densities (L < 4 nm).

Although the size of Fn is much larger than that of BSA, it couldadopt different conformations depending strongly on the environ-

mental conditions and different characteristics of biomaterials sur-face [52] due to its flexible and special structure, which includes twoalmost identical subunits of about 220 kDa each, and is crosslinkedby disulfide bonds in the C-terminal with a maximum length of

0

1000

2000

3000

4000

PEG cha in dens ity (cha ins/nm2)

No pre-adsorpti on

Fn

FBS

0 0.04 0.12 0.29

b

*

*

** *

*

*

*

*

surfaces with different densities for 8 h in serum-free DMEM. The substrates werelowed with thorough rinsing in PBS. * Significant difference at p < 0.05. Error barsea studies, respectively.

140 M. Sun et al. / Colloids and Surfaces B: Biointerfaces 122 (2014) 134–142

on PEG

1[stadmtsittavw[oa

baoaswsassmwSanPrcwsttfistm

Scheme 1. Schematic of Fn adsorption and SMCs attachment

40 nm and average diameter of the strand 2 nm for each subunit53,54]. Each subunit of Fn contains several cell binding regionsuch as RGD domain [55–57], to which SMCs can attach directlyhrough integrins �5�1 or �v�3 [58,59]. In this study, the celldhesion and spreading decreased with the increase of PEG chainensity, whereas the adsorbed Fn amount was the lowest on theedium density of PEG-grafted surface. The inconsistency between

he adsorbed Fn amount and the SMCs behaviors on the PEG-graftedurface gives a hint that the conformation of Fn should also take anmportant role on the cellular behaviors. Previous studies foundhat fibroblasts and endothelial cells behave similarly and havehe fasted migration rates at a moderate PEG content [17,23]. Tzi-mpazis et al. reported that the bioactivity of adsorbed Fn remainedery high on a copolymer film with a lower level of PEG (0–6%), andas then reduced sharply when the PEG content increased to 8%

23]. They attributed this phenomenon to the conformation changef Fn from a tight configuration to a fully extended one, and finally

close-packed one with a large degree of RGD buried.Taking all the results into consideration, the interaction

etween the PEG-grafted surfaces, adsorption behavior of Fn,nd cellular behaviors is suggested (Scheme 1). On the TMSPMAr lower density of PEG-grafted surface, larger amount of Fn isdsorbed. It adopts a tightly tethered configuration bound to theubstrate with multiple sites, exposing sufficient RGD sequenceshich are favorable for SMCs adhesion with a larger number and

preading area. However, owning to the high adhesion force of Fnnd thereby the stronger adhesion of cells, the cells stick to theurface and their tails are not easily contracted to release from theurface. Therefore, the cells have a relatively low mobility. On theedium density PEG-grafted surface, less amount of Fn is adsorbedith a looser configuration due to the limited adhesive space for Fn.

ome numbers of the RGD sequences can be exposed and provideppropriate adhesion force for cells, leading to the reduced cellumber and spreading area but fastest mobility. On the highestEG density surface, however, the PEG chains with the brushegime did not further reduce the adsorption amount of Fn. In thisase, the Fn most possibly adopts an extended configuration boundith the substrate weakly at a very few sites. The cell adhesion

ites such as the RGD peptide sequences, and the 4F1.5F1 segmentshat are located in the heparin I and fibrin binding domains nearhe N terminal of the Fn molecule [60], shall be buried by theully extended PEG chains (∼30 nm). Consequently, the weak

nteractions between the Fn, substrate, and cells result in themallest cell adhesion number. The cells turn nearly round withhe weakest cell adhesion force, which cannot drive the cell

ovement. Therefore, the adsorption amount, adhesion force and

-grafted surfaces with different densities. For detail, see text.

conformation variations of Fn definitely influence the cellularbehaviors. It is quite possible to control the protein adsorp-tion behaviors, including the adsorption amount, orientationand conformation, by controlling the PEG chain density in anappropriate range and thereby the cell adhesion and migrationbehaviors.

5. Conclusions

Covalent grafting of PEG molecules was implemented on amino-functionalized substrate, forming a mushroom-like regime at thelowest grafting density (� < 0.12) and brush regimes at interme-diate and high grafting densities (� ≥ 0.12). The adsorption massof both Fn and BSA decreased with the increase of PEG graftingdensity. The presence of BSA reduced the adsorption amount ofFn to some degree. With a smaller density of PEG the Fn adsorp-tion was reduced significantly in the presence of BSA, whereas notinfluenced with the highest PEG density. In serum the adsorptionamount of Fn was dramatically decreased, conveying that otherplasma proteins contribute heavily to the competitive adsorptionof Fn as well.

The mPEG chains of a larger density are more efficient in resist-ing cell adhesion and spreading. Pre-adsorption of Fn and serumon the substrates could significantly improve both the cell numbersand spreading areas. The cell behaviors are in good agreement withthe adsorption behavior of Fn, and other types of plasma proteinsshould also take a positive role to some extent.

Taking all the results into consideration, the interaction betweenthe PEG-grafted surfaces, adsorption behavior of Fn, and cellularbehaviors is suggested (Scheme 1). The difference in PEG graft-ing density influences the adsorption amount and configuration ofFn, which in turn influences the Fn adhesion force and cell adhe-sion force, and eventually governs the adhesion, spreading andmigration behaviors of SMCs. The fundamental understanding ofthe complex relationship between biomaterials surface, proteinadsorption, and cellular behaviors has considerable implications forbiomaterials used in tissue engineering and regenerative medicine,especially for the surface design and functionalization.

Acknowledgement

This study is financially supported by the National BasicResearch Program of China (2011CB606203), and the National Nat-ural Science Foundation of China (21374097 and 51120135001).

es B: B

A

i2

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M. Sun et al. / Colloids and Surfac

ppendix A. Supplementary data

Supplementary material related to this article can be found,n the online version, at http://dx.doi.org/10.1016/j.colsurfb.014.06.041.

eferences

[1] T. Satomi, Y. Nagasaki, H. Kobayashi, H. Otsuka, K. Kataoka, Density control ofpoly(ethylene glycol) layer to regulate cellular attachment, Langmuir 23 (2007)6698–6703.

[2] N.P. Reynolds, K.E. Styan, C.D. Easton, Y. Li, L. Waddington, C. Lara, J.S. Forsythe,R. Mezzenga, P.G. Hartley, B.W. Muir, Nanotopographic surfaces with definedsurface chemistries from amyloid fibril networks can control cell attachment,Biomacromolecules 14 (2013) 2305–2316.

[3] T. Sun, G. Qing, B. Su, L. Jiang, Functional biointerface materials inspired fromnature, Chem. Soc. Rev. 40 (2011) 2909–2921.

[4] X. Wang, H. Gan, M. Zhang, T. Sun, Modulating cell behaviors on chiral poly-mer brush films with different hydrophobic side groups, Langmuir 28 (2012)2791–2798.

[5] G. Qing, T. Sun, Chirality-driven wettability switching and mass transfer,Angew. Chem. Int. Ed. 53 (2014) 930–932.

[6] H.A. Pan, J.Y. Liang, Y.C. Hung, C.H. Lee, J.C. Chiou, G.S. Huang, The spa-tial and temporal control of cell migration by nanoporous surfaces throughthe regulation of ERK and integrins in fibroblasts, Biomaterials 34 (2013)841–853.

[7] Z.A. Cheng, O.F. Zouani, K. Glinel, A.M. Jonas, M.C. Durrieu, Bioactive chemicalnanopatterns impact human mesenchymal stem cell fate, Nano Lett. 13 (2013)3923–3929.

[8] P.Y. Wang, L.R. Clements, H. Thissen, A. Jane, W.B. Tsai, N.H. Voelcker, Screeningmesenchymal stem cell attachment and differentiation on porous silicon gra-dients, Adv. Funct. Mater. 22 (2012) 3414–3423.

[9] W. Wei, L. Bo, G. Changyou, Modulating the differentiation of BMSCs by surfaceproperties of biomaterials, Prog. Chem. 23 (2011) 2160–2168.

10] X. Yao, Y. Hu, B. Cao, R. Peng, J. Ding, Effects of surface molecular chiral-ity on adhesion and differentiation of stem cells, Biomaterials 34 (2013)9001–9009.

11] X. Dongming, M. Lie, G. Changyou, Preparation and application of orderedand gradient scaffolds in regenerative medicine, Prog. Chem. 23 (2011)2550–2559.

12] H. Chen, Z. Zhang, Y. Chen, M.A. Brook, H. Sheardown, Protein repellant siliconesurfaces by covalent immobilization of poly(ethylene oxide), Biomaterials 26(2005) 2391–2399.

13] W.R. Gombotz, W. Guanghui, A.S. Hoffman, Immobilization of poly(ethyleneoxide) on poly(ethylene terephthalate) using a plasma polymerization process,J. Appl. Polym. Sci. 37 (1989) 91–107.

14] E. Ostuni, R.G. Chapman, R.E. Holmlin, S. Takayama, G.M. Whitesides, A surveyof structure−property relationships of surfaces that resist the adsorption ofprotein, Langmuir 17 (2001) 5605–5620.

15] J.G. Archambault, J.L. Brash, Protein resistant polyurethane surfaces by chemicalgrafting of PEO: amino-terminated PEO as grafting reagent, Colloids Surf., B:Biointerfaces 39 (2004) 9–16.

16] P. Kingshott, S. McArthur, H. Thissen, D.G. Castner, H.J. Griesser, Ultrasensi-tive probing of the protein resistance of PEG surfaces by secondary ion massspectrometry, Biomaterials 23 (2002) 4775–4785.

17] S.H. Hsu, C.M. Tang, J.J. Chiu, T.C. Liao, C.C. Lin, H. Iwata, Cell migrationrate on poly(epsilon-caprolactone)/poly(ethylene glycol) diblock copolymersand correlation with the material sliding angle, Macromol. Biosci. 7 (2007)82–90.

18] J. Wu, Z. Mao, C. Gao, Controlling the migration behaviors of vascular smoothmuscle cells by methoxy poly(ethylene glycol) brushes of different molecularweight and density, Biomaterials 33 (2012) 810–820.

19] H. Chen, L. Yuan, W. Song, Z. Wu, D. Li, Biocompatible polymer materials: roleof protein–surface interactions, Prog. Polym. Sci. 33 (2008) 1059–1087.

20] R.O. Hynes, Integrins: bidirectional, allosteric signaling machines, Cell 110(2002) 673–687.

21] D. Yang, X. Lu, Y. Hong, T. Xi, D. Zhang, The molecular mechanism of media-tion of adsorbed serum proteins to endothelial cells adhesion and growth onbiomaterials, Biomaterials 34 (2013) 5747–5758.

22] E. Ruoslahti, M. Pierschbacher, New perspectives in cell adhesion: RGD andintegrins, Science 238 (1987) 491–497.

23] E. Tziampazis, J. Kohn, V.P. Moghe, PEG-variant biomaterials as selectivelyadhesive protein templates: model surfaces for controlled cell adhesion andmigration, Biomaterials 21 (2000) 511–520.

24] M.R. Kreke, A.S. Badami, J.B. Brady, R.M. Akers, A.S. Goldstein, Modulation ofprotein adsorption and cell adhesion by poly(allylamine hydrochloride) hep-arin films, Biomaterials 26 (2005) 2975–2981.

25] Y. Arima, H. Iwata, Effect of wettability and surface functional groups on pro-tein adsorption and cell adhesion using well-defined mixed self-assembled

monolayers, Biomaterials 28 (2007) 3074–3082.

26] C.C. Barrias, M.C. Martins, G. Almeida-Porada, M.A. Barbosa, P.L. Granja, Thecorrelation between the adsorption of adhesive proteins and cell behaviour onhydroxyl-methyl mixed self-assembled monolayers, Biomaterials 30 (2009)307–316.

[

[

iointerfaces 122 (2014) 134–142 141

27] S. Becerra-Bayona, V. Guiza-Arguello, X. Qu, D.J. Munoz-Pinto, M.S. Hahn,Influence of select extracellular matrix proteins on mesenchymal stem cellosteogenic commitment in three-dimensional contexts, Acta Biomater. 8(2012) 4397–4404.

28] C. Xue, B.C. Choi, S. Choi, P.V. Braun, D.E. Leckband, Protein adsorption modesdetermine reversible cell attachment on poly(N-isopropyl acrylamide) brushes,Adv. Funct. Mater. 22 (2012) 2394–2401.

29] V.J. Mukhatyar, M. Salmerón-Sánchez, S. Rudra, S. Mukhopadaya, T.H. Barker,A.J. García, R.V. Bellamkonda, Role of fibronectin in topographical guid-ance of neurite extension on electrospun fibers, Biomaterials 32 (2011)3958–3968.

30] C.R. Jenney, J.M. Anderson, Adsorbed serum proteins responsible for surfacedependent human macrophage behavior, J. Biomed. Mater. Res. 49 (2000)435–447.

31] C.D. Walkey, J.B. Olsen, H. Guo, A. Emili, W.C. Chan, Nanoparticle size and surfacechemistry determine serum protein adsorption and macrophage uptake, J. Am.Chem. Soc. 134 (2012) 2139–2147.

32] S. Choi, B.C. Choi, C. Xue, D. Leckband, Protein adsorption mechanisms deter-mine the efficiency of thermally controlled cell adhesion on poly(N-isopropylacrylamide) brushes, Biomacromolecules 14 (2013) 92–100.

33] N. Giamblanco, E. Martines, G. Marletta, Laminin adsorption on nanostructures:switching the molecular orientation by local curvature changes, Langmuir 29(2013) 8335–8342.

34] J. Deng, M. Sun, J. Zhu, C. Gao, Molecular interactions of different sizeAuNP–COOH nanoparticles with human fibrinogen, Nanoscale 5 (2013)8130–8137.

35] J. Pei, H. Hall, N.D. Spencer, The role of plasma proteins in cell adhesion toPEG surface–density-gradient-modified titanium oxide, Biomaterials 32 (2011)8968–8978.

36] J.H. Seo, K. Sakai, N. Yui, Adsorption state of fibronectin onpoly(dimethylsiloxane) surfaces with varied stiffness can dominate adhesiondensity of fibroblasts, Acta Biomater. 9 (2013) 5493–5501.

37] L. Marquardt, R.K. Willits, Student award winner in the undergraduate’s degreecategory for the society for biomaterials 35th annual meeting, Orlando, FloridaApril 13–16, 2011, J. Biomed. Mater. Res. A 98A (2011) 1–6.

38] H. Yamashita, M. Tripathi, M.P. Harris, S. Liu, B. Weidow, R. Zent, R. Quaranta,The role of a recombinant fragment of laminin-332 in integrin �3�1-dependent cell binding, spreading and migration, Biomaterials 31 (2010)5110–5121.

39] H.W. Zhao, C.Z. Huang, Y.F. Li, A novel optical immunosensing system basedon measuring surface enhanced light scattering signals of solid supports, Anal.Chim. Acta 564 (2006) 166–172.

40] B. Zdyrko, S.K. Varshney, I. Luzinov, Effect of molecular weight on synthesisand surface morphology of high-density poly(ethylene glycol) grafted layers,Langmuir 20 (2004) 6727–6735.

41] T. Drobek, N.D. Spencer, M. Heuberger, Compressing PEG brushes, Macro-molecules 38 (2005) 5254–5259.

42] I. Luzinov, D. Julthongpiput, H. Malz, J. Pionteck, V.V. Tsukruk, Polystyrenelayers grafted to epoxy-modified silicon surfaces, Macromolecules 33 (2000)1043–1048.

43] D. Li, H. Chen, W. Glenn McClung, J.L. Brash, Lysine-PEG-modified polyurethaneas a fibrinolytic surface: effect of PEG chain length on protein inter-actions, platelet interactions and clot lysis, Acta Biomater. 5 (2009)1864–1871.

44] H. Chen, Y. Zhang, D. Li, X. Hu, L. Wang, W.G. McClung, J.L. Brash, Surfaces havingdual fibrinolytic and protein resistant properties by immobilization of lysineon polyurethane through a PEG spacer, J. Biomed. Mater. Res. A 90A (2009)940–946.

45] M. Stenberg, H. Nygren, The use of the isoscope ellipsometer in the study ofadsobed proteins and biospecific binding reactions, J. Phys. Colloq. 44 (1983),C10-83-C10-86.

46] K.M. Yamada, K. Olden, Fibronectins—adhesive glycoproteins of cell surface andblood, Nature 275 (1978) 179–184.

47] Y. Inoue, T. Nakanishi, K. Ishihara, Adhesion force of proteins againsthydrophilic polymer brush surfaces, React. Funct. Polym. 71 (2011) 350–355.

48] M. Li, D.K. Mills, T. Cui, M.J. McShane, Cellular response to gelatin-and fibronectin-coated multilayer polyelectrolyte nanofilms, IEEE Trans.NanoBiosci. 4 (2005) 170–179.

49] A. Halperin, G. Fragneto, A. Schollier, M. Sferrazza, Primary versus ternaryadsorption of proteins onto PEG brushes, Langmuir 23 (2007) 10603–10617.

50] A. Halperin, Polymer brushes that resist adsorption of model proteins: designparameters, Langmuir 15 (1999) 2525–2533.

51] A. Halperin, M. Kroger, Ternary protein adsorption onto brushes: strong versusweak, Langmuir 25 (2009) 11621–11634.

52] D.J. Iuliano, S.S. Saavedra, G.A. Truskey, Effect of the conformation and orienta-tion of adsorbed fibronectin on endothelial cell spreading and the strength ofadhesion, J. Biomed. Mater. Res. 27 (1993) 1103–1113.

53] H.P. Erickson, N. Carrell, Fibronectin in extended and compact conformations.Electron microscopy and sedimentation analysis, J. Biol. Chem. 2581 (1983)14539–14544.

54] H.P. Erickson, J.L. Inglesias, A six-armed oligomer isolated from cell surface

fibronectin preparations, Nature (1984).

55] R.O. Hynes, The dynamic dialogue between cells and matrices: implications offibronectin’s elasticity, Proc. Nat. Acad. Sci. 96 (1999) 2588–2590.

56] M. Bergkvist, J. Carlsson, S. Oscarsson, Surface-dependent conformations ofhuman plasma fibronectin adsorbed to silica, mica, and hydrophobic surfaces,

1 es B: B

[

[

[

[

[

42 M. Sun et al. / Colloids and Surfac

studied with use of atomic force microscopy, J. Biomed. Mater. Res. A 64A (2003)349–356.

57] H.P. Erickson, N. Carrell, J. McDonagh, Fibronectin molecule visualized in elec-tron microscopy: a long, thin, flexible strand, J. Cell Biol. 91 (1981) 673–678.

58] S. Miyamoto, B.-Z. Kathz, R.M. Lafrenie, K.M. Yamada, Fibronectin and integrins

in cell adhesion, signaling, and morphogenesis, Ann. N.Y. Acad. Sci. 857 (1998)119–129.

59] M. Obara, M.S. Kang, K.M. Yamada, Site-directed mutagenesis of the cell-binding domain of human fibronectin: separable, synergistic sites mediateadhesive function, Cell 53 (1988) 649–657.

[

iointerfaces 122 (2014) 134–142

60] N. Giamblanco, M. Yaseen, G. Zhavnerko, J.R. Lu, G. Marletta, Fibronectin confor-mation switch induced by coadsorption with human serum albumin, Langmuir27 (2010) 312–319.

61] K.M. Kallury, P.M. Macdonald, M. Thompson, Effect of surface water and basecatalysis on the silanization of silica by (aminopropyl) alkoxysilanes studied

by X-ray photoelectron spectroscopy and 13C cross-polarization/magic anglespinning nuclear magnetic resonance, Langmuir 10 (1994) 492–499.

62] P. Merel, M. Tabbal, M. Chaker, S. Moisa, J. Margot, Direct evaluation of thesp3 content in diamond-like-carbon films by XPS, Appl. Surf. Sci. 136 (1998)105–110.