Acta Biomaterialia 6 (2010) 2711–2720

Contents lists available at ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /actabiomat

Tuning cell adhesion by controlling the roughness and wettability of 3Dmicro/nano silicon structures

A. Ranella a, M. Barberoglou a,b, S. Bakogianni a,b, C. Fotakis a,b, E. Stratakis a,c,*

a Institute of Electronic Structure and Laser, Foundation for Research and Technology—Hellas, (IESL-FORTH), P.O. Box 1527, Heraklion 711 10, Greeceb Physics Department, University of Crete, Heraklion 710 03, Greecec Materials Science and Technology Department, University of Crete, Heraklion 710 03, Greece

a r t i c l e i n f o a b s t r a c t

Article history:Received 21 May 2009Received in revised form 14 November 2009Accepted 11 January 2010Available online 18 January 2010

Keywords:Cell adhesionSiliconScaffoldSurface roughnessSurface energy

1742-7061/$ - see front matter � 2010 Acta Materialdoi:10.1016/j.actbio.2010.01.016

* Corresponding author. Address: Institute of ElecFoundation for Reasearch and Technology Hellas,Department of Materials Science and Technology, VBox 1527, GR-711 10, Greece. Tel.: +30 2810 391274;

E-mail address: stratak@iesl.forth.gr (E. Stratakis).

The aim of this study is to investigate fibroblast cell adhesion and viability on highly rough three-dimen-sional (3D) silicon (Si) surfaces with gradient roughness ratios and wettabilities. Culture surfaces wereproduced by femtosecond (fs) laser structuring of Si wafers and comprised forests of conical spikes exhib-iting controlled dual-scale roughness at both the micro- and the nano-scale. Variable roughness could beachieved by changing the laser pulse fluence and control over wettability and therefore surface energycould be obtained by covering the structures with various conformal coatings, which altered the surfacechemistry without, however, affecting morphology. The results showed that optimal cell adhesion wasobtained for small roughness ratios, independently of the surface wettability and chemistry, indicatinga non-monotonic dependence of fibroblast adhesion on surface energy. Additionally, it was shown that,for the same degree of roughness, a proper change in surface energy could switch the behaviour fromcell-phobic to cell-philic and vice versa, transition that was always correlated to surface wettability.These experimental findings are discussed on the basis of previous theoretical models describing the rela-tion of cell response to surface energy. The potential use of the patterned Si substrates as model scaffoldsfor the systematic exploration of the role of 3D micro/nano morphology and/or surface energy on celladhesion and growth is envisaged.

� 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

The eventual goal of artificial cell culture scaffold design is tomimic the natural extracellular environment features, in a waythat cells can function as if they in vivo. In this respect, designingsurfaces for controlling cell–material interactions is a considerablyinteresting subject, highly significant in the development andeventual success of implantable medical devices and engineeredtissues [1–3]. Using different approaches, various materials havebeen surface engineered in order to guide cell adhesion and mod-ulate cell–biomaterial interactions [4–6], indicating that cellgrowth, division and migration are highly dependent on theirimmediate culture substrate. Hence, surface chemistry [7–9], wet-tability [10,11] and roughness [2,12,13] are found to be three of themost important factors influencing biological reactions at biomate-rial surfaces [14]. In spite of such numerous investigations, anintrinsic and deep understanding of the factors governing attach-

ia Inc. Published by Elsevier Ltd. A

tronic Structure and Lasers,University Of Crete, Hellas,outes Heraklion Crete, P.O.fax: +30 2810 391305.

ment of cells to biomedical surfaces, is still limited [15,16]. Moststudies have been using substrates covering only a small range ofroughness, while the effects of nanoscale or microscale roughnesswere in most cases separately investigated [2,17–21]. There is,however, increasing evidence indicating that cell–surface interac-tions occur at multiple length scales [6]. Indeed, in the context ofa natural environment, the cells included in a tissue are sur-rounded by a three-dimensional (3D) dynamic extracellular matrix(ECM), which provides instructive cues at meso- micro- and nano-scales necessary to maintain cell phenotype and behaviour. There-fore, in order to reach the level of ECM complexity, biomedical sub-strates must interlace hierarchically organized multiple-scalestructures. Due to such demanding design requirements, cell re-sponse on sub-micrometer as well as micrometer scale rough sub-strates, has not yet been systematically studied. Thus, theproduction of surface engineering schemes enabling controlledand reproducible structuring of biomaterials at both micro- andnano-scales, is required [22].

Another limiting parameter of the so far available literature in-cludes the instability of chemical composition of the surfaces,which by changing deliberately or unintentionally during themicro/nano structuring process, damage the cells, since they areinherently sensitive to variations of surface chemistry [6].

ll rights reserved.

2712 A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720

Consequently, most of studies raise the question of whether cellbehaviour, related to viability, proliferation, motility, adhesion,morphology, cytoskeletal arrangement and gene expression, isinfluenced by the topographical features and/or the surface chem-istry of a scaffold, parameters that determine the surface tensionand wettability [23,24]. In order to overcome such controversies,it is crucial to find pathways to discriminate among the differenteffects. In this respect, the ability (a) to control the surface chem-istry, while keeping the same micro/nano morphology or vice versaand (b) to tune surface wettability from super-hydrophobicity tosuper-hydrophilicity are highly important tools allowing success-ful attachment of fully functional cells to biomedical surfaces.

The results presented here demonstrate the ability to tune celladhesion using biomimetic artificial substrates in a 3D-design,comprising hierarchical micro- and nanostructures produced byultrafast laser patterning of silicon (Si). Tailoring of the structures’morphological features can be advantageously achieved by varyingcertain fabrication parameters. This capability allows one to con-trol the wetting properties and thus, surface tension of the Si sub-strates. Further control over wettability can be achieved by alteringthe surface chemistry through coating the structures with variousconformal layers, while keeping the same morphology. As a conse-quence, it is possible to preferentially tune the wettability of theartificial substrates from super-hydrophobicity to super-hydrophi-licity through a proper combination of surface topography andchemistry. In this respect, it was demonstrated that the patternedSi substrates can be potentially used as model scaffolds for the sys-tematic exploration of the role of 3D micro/nano morphology and/or surface energy on cell adhesion and growth. The different struc-tures obtained by this method can be transferred to various typesof polymeric materials as well through replication molding tech-niques [25]. The simplicity of the structuring process and the flex-ibility of fast patterning by laser beam scanning, together with itspotential to tailor the wettability of different classes of materials,are certainly useful for creating patterned interfaces on biomateri-als devices.

2. Materials and methods

2.1. Micro and nanostructure fabrication

A simple but effective method to fabricate silicon micro/nanostructures over a large area with superior control of structuregeometry and pattern regularity was used based on ultrafast laserstructuring [26]. Briefly, single crystal n-type Si (1 0 0) wafers weresubjected to laser irradiation in a vacuum chamber evacuateddown to a residual pressure of 10�2 mbar. A constant SF6 pressureof 500 Torr was maintained during the process through a precisionmicro valve system. The irradiating laser source was constituted bya regenerative amplified Ti:Sapphire (k = 800 nm) delivering 150 fspulses at a repetition rate of 1 kHz. The sample was mounted on ahigh-precision X–Y translation stage normal to the incident laserbeam. A mechanical shutter was synchronized to the translationstages, exposing any given spot on the Si surface to an average of500 pulses. Each micro-structured surface was fabricated at con-stant fluence ranging from 0.34 to 1.69 J/cm2. After laser irradia-tion, the micro-structured surfaces were cleaned with ultrasonicbaths of trichloroethylene, followed by acetone and ethanol. Theneach surface was morphologically characterized by field emissionscanning electron microscopy (SEM). An image processing algo-rithm was implemented in order to obtain quantitative informa-tion concerning the topological characteristics of the formedstructures i.e. spikes’ density, profile, height, and surface area, fromtop, side-view and cross-sectional SEM pictures of the structuredregions. The roughness ratio, r, was calculated by dividing the ac-

tual, unfolded, surface area of spikes over the total irradiated area.In order to alter the surface chemistry, freshly prepared patternedsurfaces were covered with either a thermally grown, hydrophilicoxide layer or with a hydrophobic silane coating through vapouradsorption from solution. Both layers are known to form high qual-ity conformal coatings on Si surfaces. For oxidation, the sampleswere placed in a box furnace and heated at 1000 �C for 30 min inair. For silanization, the samples were placed in a flask containing0.5 ml of dichlorodimethylsilane (CH3)2SiCl2 (DMDCS) reagent,where hydrophobic DMDCS monolayers were deposited on theirsurface via adsorption reactions. The silanization process em-ployed is similar to that reported in the literature [27].

2.2. Wettability measurements

Static contact angle measurements were performed by an auto-mated tensionmeter, using the sessile drop method. A 2 ll dis-tilled, deionised Millipore water droplet was gently positioned onthe surface using a microsyringe and images were captured tomeasure the angle formed at the liquid–solid interface. The meanvalue was calculated from at least five individual measurements.Successive measurements were reproducible within ±1�.

2.3. Cell cultures

Prior to cell culture the structured surfaces were sterilized andthen transferred onto sterile 6 well plates (Sarstedt; Numbrecht,Germany).

The NIH/3T3 cells were suspended to a concentration of105 cells/mL in Dulbecco’s modified Eagle’s medium (DMEM) sup-plemented with 10% fetal bovine serum (FBS), and 1% antibioticsolution (GIBCO, Invitrogen, Kalsruhe, Germany) and 3 ml of cellsuspension was added in the 6 well plate and cultured at 37 �C,for 72 h in an atmosphere of 5% CO2. Before seeding the cells onthe different surfaces, cells were grown to confluency, detachedwith 0.05% trypsin/EDTA (GIBCO, Invitrogen, Kalsruhe, Germany)and diluted in complete medium at an appropriate density. Allexperiments were done in triplicates to ensure reproducibilityand obtain better statistics.

2.4. Cell viability assay

In order to assess cell viability, the Live-Dead Cell Staining Kit(BioVision) was used. The kit utilizes Live-Dye, a cell-permeablegreen fluorescent dye (Ex/Em = 488/518 nm), to stain live cells.Dead cells can be easily stained by propidium iodide (PI), a cellnon-permeable red fluorescent dye (Ex/Em = 488/615). At the endof the incubation time (72 h), structured surfaces with the cellswere covered with the staining solution and incubated for15 min at 37 �C. Cells were observed immediately under a fluores-cence microscope. Stained live and dead cells can be visualized byfluorescence microscopy using a band-pass filter (detects FITC andrhodamine). Healthy cells stain only with the cell-permeable Live-Dye, fluorescing green. Dead cells can stain with both the cell-per-meable Live-Dye and the cell non-permeable PI (red), the overlayof green and red appears to be yellow–red. The experiments weredone in triplicates and for each surface the mean number of livecells was calculated.

2.5. Laser scanning confocal microscopy

For the double staining of F-actin and Vinculin the focal adhe-sion staining kit (Chemicon International Inc., Temecula, CA, USA)was used. The cells were fixed with 4% paraformaldehyde for15 min and permeabilized with 0.1% Triton X-100 (Merck KGaA,Darmstadt,Germany) in PBS for 3–5 min on ice. The non-specific

A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720 2713

binding sites were blocked with 1% BSA in PBS for 30 min. Actinand focal adhesion complexes were stained by incubating cells indiluted primary antibody (anti-vinculin) in blocking solution(1:200) for 1 h and subsequently labelling them with diluted sec-ondary antibody (1:200) (mouse–anti-mouse FITC conjugate) (Sig-ma–Aldrich Chemie GmbH, Munich, Germany) for 45 min, withsimultaneous incubation with diluted tetramethyl rhodamine iso-thiocyanate-conjugated phalloidin. The samples were then washedwith PBS and stored in 10% glycerol in PBS in dark. Confocalmicroscopy was performed using a ‘Zeiss AxiosKop 2 plus’ laserscanning confocal microscope.

2.6. Scanning electron microscopy

The morphologies of NIH/3T3 fibroblasts seeded on patternedsurfaces or on flat silicon wafer were observed by SEM. After incu-bation, cells were washed with 0.1 M of sodium cacodylate buffer(SCB) and then incubated with the SCB for 15 min. This step wasrepeated twice and followed by the fixation of the cells using a fix-ative buffer (2% glutaraldehyde, 2% formaldehyde in 1% SCB) for 1 hat 4 �C. Subsequently, the surfaces were washed twice (from15 min each time) with 1% SCB at 4 �C. After that, cells were dehy-drated through a graded ethanol series (from 10% to 100%) andincubated for 15 min on dry 100% ethanol. Prior to electron micros-copy examination the samples were sputter coated with a 10 nmgold layer. SEM was performed on a JEOL 7000 field emission scan-ning electron microscope with an acceleration voltage of 10 kV.

2.7. Statistical analysis

Student’s t-test was used to compare the significance levels(p < 0.001) between control and test values.

Fig. 1. (a) Picture of a polished Si wafer (i) and side SEM views of the as-prepared Si spikcm2(A2), (iv) 0.90 J/cm2(A3), and (v) 1.69 J/cm2(A4); (b) high magnification SEM imagespatterned Si surfaces; and (d) confocal laser microscopy pictures of fibroblast cells cultu

3. Results

3.1. Surface topography and wettability

The manufactured substrates, which possess a 3D-design, wereproduced by simultaneous structuring at both micrometer andnanometer length scales. As shown in the SEM micrographs of pat-terned Si surfaces produced at different incident laser energies perunit area (fluences, Fig. 1a and b), the treated areas comprise for-ests of conical microstructures (spikes) exhibiting structures atthe micro- and nano-scale. Variation of laser fluence causedremarkable changes in the structures as to ratio, dimension anddensity. Besides directly affecting the micrometer-scale surfacetopology, increasing fluence is also crucial to induce a more pro-nounced sub-micrometer decoration on the spikes walls. In partic-ular, the spikes height varied from one to ten micrometers, whilethe size of nano-protrusions ranged from tens to a few hundredsof nanometers, providing a double length-scale pattern on the sil-icon surface (Fig. 1b). The micrometer-scale features, together withthe nanoscale protrusions resulted into a significant increase of theoverall roughness. Consequently, these etched substrates shouldallow more 3D free space perpendicular to the culture plane andshould also provide physical cues to facilitate cell adhesion andspreading.

In this study, three series of substrates were tested: Type A cor-responds to the, as-prepared, patterned Si surfaces comprising foursamples with gradient roughness ratios, denoted as A1–A4; Type Bcorresponds to the same substrates coated with a hydrophilic ther-mally grown oxide layer (B1-B4); (c) Type C corresponds to thesame substrates coated by a hydrophobic silane layer. Types Aand C samples had shown the same qualitative results and willbe considered as equivalent cases thereafter. In all series, the

es surfaces structured at four different laser fluencies (ii) 0.34 J/cm2(A1), (iii) 0.56 J/of the corresponding Si cones obtained; (c) photographs of water droplets on thered for three days on the respective surfaces.

Table 1General properties of the different samples used for this study. Series A denote as-prepared patterned Si surfaces and series B denote the patterned surfaces after thermaloxidation treatment.

Sample Flat Si Oxidized Flat Si A1 B1 A2 B2 A3 B3 A4 B4

Laser fluence (J/cm2) – – 0.34 0.34 0.56 0.56 0.90 0.90 1.69 1.69Roughness ratio 0 0 2.6 2.6 3.3 3.3 6.0 6.0 6.9 6.9Wetting angle (�) 67 25 105 20 121 20 152 0 154 0

2714 A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720

corresponding flat surfaces were tested as control samples. Theprincipal properties of the above substrate types are summarizedin Table 1. Images of water droplets lying on flat and as-preparedmicro-structured Si surfaces are shown in Fig. 1c. The correspond-ing dependence of the wetting angle (WA) on the laser fluence androughness ratio attained is plotted in Fig. 2a. It is evident that thelaser assisted texturing of Si induced a remarkable increase in itshydrophobicity which was more pronounced at high fluences[28]. This was a direct consequence of the hierarchical morphologyof the artificial structures, which mimicked that of the naturalsuper-hydrophobic surfaces [29]. Indeed, the WA of the patternedsurface structured at the highest fluence was 154�, which is closeto the average WA measured on the most popular natural ultrahy-drophobic surface – the Lotus (nelumbo nucifera) leaf [30].

3.2. Cell spreading and viability

In order to investigate whether these surfaces were able tomodulate cellular responses, cell adhesion experiments were per-formed using the fibroblast NIH/3T3 cell line. Optical and SEM pho-tomicrographs of NIH/3T3 fibroblasts, seeded on patternedsurfaces, showed that both the cell concentration and shape aredifferent depending on the culture substrate (Figs. 1d and 3,respectively). The number of attached cells per unit area decreased

. . . . .

Fig. 2. (a) Relation between the laser fluence and surface wettability for the series Apatterned Si substrates; the corresponding roughness ratio for each sample is alsoindicated; (b) cell density after 72 h incubation as a function of the surfacewettability for the flat Si and structured surfaces of series A. All experiments weredone on triplicates and the cell density values plotted are the calculated meanvalues. The lines serve as a guide for the eye.

as the roughness ratio and WA increased, denoting that cell attach-ment was favoured on more hydrophilic surfaces. Concurrently,cell morphology changed from a well spread polygonal phenotypeon smoother surfaces to a round-shaped phenotype on highlyrough super-hydrophobic substrates and a smaller cell size, indic-ative of poor adhesion. Additionally, the SEM micrographs of Fig. 3showed that in super-hydrophobic substrates, cells appeared toaccumulate and form multilayers, probably due to limited oppor-tunity for sufficient traction and stretch out in order to minimizeunfavourable contacts with the substrate. This layer of elongatedcells could serve as a basal structure for other cells to populate thatregion, resulting in the formation of cell multilayers (Fig. 3f). Theextent of cells spreading on a material provides a visual qualitativeindicator of the strength of the cell–surface interaction [23]. Polyg-onal cell spreading implies extensive interaction between the sur-face and the cells, whereas weak cell–surface interaction results incell clustering [23,31].

Cell viability of the attached cells on the different surfaces wasassessed using the ‘live-dead’ kit (Fig. 4). This staining protocolfacilitates counting the number of live (depicted as green) anddead cells (depicted as red/yellow) per unit area. According to thistest cell cultures on hydrophilic surfaces (A1) represented a verylow percentage of dead cells, equal to 2%. This percentage seemedto increase proportional to the roughness ratio, namely on surfacesA2 and A4 the particular percentages were 17% and 42%, respec-tively. The corresponding dependence of the live cell density onthe laser fluence and surface wettability is presented in Figs. 2band 5, respectively. It is obvious that the induction of roughnessfacilitated cell spreading up to an optimum roughness ratio. Onthe contrary, cell spreading was inhibited on highly rough andsuper-hydrophobic substrates. Furthermore, hydrophobization ofsubstrates using a hydrophobic silane coating, failed to reversethe inhibition of cell spreading. In order to further characterizethe principal surface property determining cell response, the wet-tability of the substrates was modified by coating the structureswith a hydrophilic oxide layer. As a result, all the structured sur-faces were converted to super-hydrophilic, without altering theirinitial morphology (Fig. 6a). In this case, a remarkable change incell behaviour was obtained, since they spread well in the highlyrough oxidized surfaces, B3 and B4 (Fig. 6b). Similar results wereobtained by culturing HeLa cells on the above mentioned surfaces(data not shown). Following the corresponding evolution of celldensities for the hydrophobic and hydrophilic surface chemistries(Figs. 5 and 7), it can be argued that the cell response is not solelydependent on surface morphology or surface chemistry, but ontheir combination, which represents surface energy. Thus, fibro-blasts preferentially adhered to high surface energy or hydrophilicsubstrates. The ability to tune the surface energy of the patternedSi substrates, by changing either the total roughness or chemistry,could allow advantageous manipulation of the cell behaviour andvice versa.

3.3. Cell adhesion

Once fibroblasts detect their target location using the filopodiaon their leading edge, the formation of lamellipodia facilitates the

Fig. 3. SEM micrographs of NIH/3T3 cells cultured on flat (a) and A1 (a and d), A2 (b and e), A4 (c and f) patterned Si substrates with low, mid and high roughness ratiorespectively. The corresponding roughness ratios are also shown. The dashed lines indicate the border between flat and patterned regions.

Fig. 4. Fluorescence microscopy images of live (green) and dead (yellow–red) cells cultured on A1 (a), A2 (b) and A4 (c) patterned Si substrates.

Fig. 5. Cell density of fibroblasts on flat and oxidized Si, the as-prepared (series A)and oxidized (series B) patterned Si surfaces after 72 h incubation. All experimentswere done on triplicates and the cell density values shown are the calculated meanvalues.

A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720 2715

cell movement to the desired site [32]. While a lot of studies havereported that cells are sensitive to micro/nanoscale topography[17,33,34], it is generally accepted that cells use filopodia for spa-

tial sensing in their movement and spread on structured surfaces.Thus, the ability of fibroblast cells to recognize and adhere to thedifferent surfaces described here was examined.

Confocal microscopy analysis, following actin distribution in thecells growing on the different structured surfaces, clearly showedmorphological changes in the various cases tested here (Fig. 8).Thus, fibroblast cells growing on more hydrophobic surfaces weremostly rounded (Fig. 8g and i), while a significantly larger numberof cells growing on that surface (Fig. 8c and e) as well as on flatcontrol (Fig. 8a) had produced lamellae. In all cases the actin fila-ments were located at the cell periphery. The high magnificationSEM images confirmed the differences in lamellipodium formation,also showing that cells on less hydrophobic structured areas werepolarized with areas of dense filopodia extension (Fig. 8d and f).Some of these cells were showing a kind of flattening, but reducedsize as compared to cells growing on the flat controls (Fig. 8b).These processes were spread in multiple planes perpendicular tothe patterned area, suggesting a 3D cell proliferation mode(Fig. 8d and f). On the contrary, cells growing on the super-hydro-phobic surfaces seemed to be smaller, with decreased number offilopodia extensions. This type of morphology could therefore beused to assess the quality of cell–surface interactions. Furthermore,the numeral processes developed between cells should facilitatecell–cell interaction and communication.

Fig. 6. Images of water droplets placed on: (a) the roughest patterned Si surface (A4) and (b) on the same surface coated with a hydrophilic silicon oxide layer (B4). Thecorresponding confocal microscopy pictures of fibroblast cells cultured for three days on these surfaces are also shown. After thermal oxidation, the super-hydrophobicsubstrate becomes super-hydrophilic. As a direct consequence, the culture substrate switches from fibroblast-phobic to fibroblast-philic.

Fig. 7. (a) Relation between cell adhesion and surface wettability for the flat andpatterned Si surfaces of series A and series B. All experiments were done ontriplicates and the cell density values plotted are the calculated mean values. Thelines serve as a guide for the eye.

2716 A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720

In order to evaluate the ability of fibroblasts to adhere on 3Dstructured surfaces, the expression of vinculin protein was exam-ined. The membrane expression of vinculin, which is a memberof focal adhesions molecules, indicates the existence of strongcell–substrate adhesion. As shown in Fig. 9, vinculin was expressedin fibroblasts cultured on A1 surfaces (Fig. 9a) and co-localizedwith actin at the edges of filopodias (Fig. 9c). Similar results wereobtained after double staining of cells cultured on the highestroughness ratio oxidised substrates (B4). On the other hand theweak cell–substrate adhesion on super-hydrophobic surfaces isevidenced by the low vinculin expression (A4; Fig. 9d).

4. Discussion

The correlation between the substrate characteristics and celladhesion highly depends on the level of surface free energy [35].Originally, Baier suggested that the amount of bioadhesion does

not correlate well with the surface energy of the substrate [36].He reported the existence of a surface energy window with mini-mal adhesion, while substrates exhibiting surface energies outsidethe defined range could absorb considerable amounts of biomass.In contrast, Schakenraad et al. found a sigmoidal dependence ofcell spreading on surface energy, indicating a sharp transition atlow surface energies between poor and good adherence [37]. How-ever, in these studies the adhesion dependence was measured on alarge number of different polymers and glasses and not on onesubstratum type, modified to display different WAs. Thus, it cannotbe excluded that in some cases, specific substrate chemical effectsmay determine adhesion and not the wettability itself.

The physical behaviour of hydrolyzed living cells may be re-garded as a drop of liquid. The adhesion of this cell liquid can beaffected by the surface wettability due to the increased or de-creased contact area, which is proportional to the solid–liquidinterfacial adhesive force. Alternatively, a living cell can be de-scribed by tensegrity models [38] which consider that cellularshape and adhesion are largely influenced by the cytoskeleton;the cell tends to form focal adhesions in locations that balancecytoskeletal forces. In this respect surface wettability is crucial asit describes to what extend the surface is exposed to culture med-ium and subsequent protein adsorption. A reduction in adhesiveprotein adsorption due to a decreased liquid–surface interfacialarea may detrimentally affect the ability of cells to form adhesions.In any case, the surface energy influences the contact area of thecell membrane with the substrate, while the profile of membranecould change depending on the wettability of the adjacent solid.In order to understand the cell response on substrates with differ-ent roughness, one has to consider the effect of the macroscopicroughness on wettability, which has been theoretically approachedby two different models. In the Wenzel model [39], the liquid is as-sumed to wet the entire rough surface, without leaving any airpockets underneath it. The apparent WA, hw, is given by the follow-ing equation:

cos hw ¼ rw cos ho; ð1Þ

where rw is the ratio of the unfolded surface to the apparent area ofcontact under the droplet, and ho is the contact angle on a flatsurface of the same nature as the rough. Since rw is always greater

Fig. 8. Fluorescent microscopy images of fibroblasts, showing actin cytoskeletalnetworks on flat silicon surface (a) and on patterned surfaces (c, e, g, and i). Thecorresponding SEM micrographs of fibroblast cell adhering to flat (b) and topatterned surfaces (d, f, h, and j) are also shown; (c and d: A1 surfaces; e and f: A2surfaces; g and h A3 surfaces; l and j: A4 surfaces).

A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720 2717

than unity, this model predicts that the contact angle will decrease/increase with surface roughness for an initially hydrophilic(ho < 90�)/hydrophobic (ho > 90�) surface. In contrast, Cassie andBaxter (CB) assumed [40] that the liquid does not completely per-meate the rough surface because air pockets get trapped under-neath it. The liquid is thus said to be in a ‘fakir’ or the Cassie–Baxter state and the apparent WA, hCB, is an average of the flat sur-face, ho, and the value for full hover over the flat surface (that is,180�) and is given by:

cos hCB ¼ �1þ f ð1þ cos#oÞ; ð2Þ

where f defines the fraction of the projected solid surface that is wetby the liquid. As f is always lower than unity, this model always pre-dicts enhancement of hydrophobicity, independently of the value ofthe initial contact angle ho. The lower the value of f, the smaller thesolid–liquid contact area and the higher the increase in the mea-sured contact angle.

Following the results presented in Fig. 2a, it is obvious that theWA values measured for the structured substrates are consistentwith the CB model because, in contrast to the Wenzel model, it pre-dicts a rise in the WA upon enhancement of the roughness of aninitially hydrophilic (h < 90�) surface. Hence, for the super-hydro-phobic substrates, where the solid–liquid contact area is minimal,water cannot penetrate the roughness elements and therefore anintervening air layer persists. This resembles the case of manyaquatic and semi-aquatic arthropods (insects and spiders) whichare rendered water repellent due to a rough, waxy exterior fes-tooned with hairs [41]. Owing this super-hydrophobic integument,the respiratory demands of these species are facilitated by a thinintervening layer of trapped air, which is called ‘plastron’ andmaintained along their body surface. This air layer is visible atnon-zero reflection angles and is responsible for the silvery under-water reflections from aquatic species. Indeed, as shown in Fig. 10,when a water repellent spike-substrate is immersed in water orcell culture liquid, it glistens with a silvery sheen, indicating thata sheathing film of air remains on the submerged surface. Con-versely, for more hydrophilic substrates, the fraction of the wettedarea increases and the surface glistening disappears, as a result ofincreasing liquid penetration. Finally for super-hydrophilic sub-strates, the contact area is maximized and water completely pene-trates the roughness elements. Therefore, taking into account thatcell culture medium is aqueous, the interaction of cell membranewith the underlying substrate may be governed by the degree ofsurface wettability and thus surface energy. Fibroblast spreadingis promoted on hydrophilic or high surface energy rough substratesdue to the permeation of the culture liquid in the structures, allow-ing cells to take advantage of the high surface-area-to-volume ratiooffered by the structured substrates. On the contrary, fibroblastadhesion is almost impossible on ultrahydrophobic or low surfaceenergy rough substrates, as the penetration of the culture liquid onthe structures is inhibited.

The results presented in Fig. 5 suggest the existence of a switch-ing function for the structured Si surface characteristics, fromfibroblast adhesive to fibroblast repulsive, when a critical combi-nation of roughness and wettabilty, and thus surface energy, is at-tained. This switching effect agrees with the sigmoidal dependencesuggested by Baier and may be attributed to the critical transitionfrom the Wenzel to Cassie–Baxter states [42]. Further support tothis statement is coming form the results obtained after substrateoxidation, where the highly rough substrate is converted from ultr-ahydrophobic to ultrahydrophilic and accordingly the wettabillityswitches from Cassie–Baxter to Wenzel states. This transition isagain followed by a corresponding enhancement in cell spreading,where the surface switches from cell-phobic to cell-philic. We canthus conclude that the sharp transition from superhydrophobicityto superhydrophilicity is accompanied by a similar transition incell adherence as well. This sharp transition between the two ex-treme wetting states may be potentially a useful tool towards con-trolling cell behaviour on culture substrates.

Furthermore, a non-monotonic dependence of fibroblast adhe-sion on wettability could be detected (Fig. 5), which is qualitativelyin accordance to the Baier model. Optimum cell adhesion was ob-tained for small roughness ratios, independently of the WA valuesand surface chemistry (Fig. 7). This observation was also supportedby the SEM images of Fig. 8, where the total number of filopodialprocesses was much higher in surfaces with low roughness.The proliferation and immigration of adhesion depended cells

Fig. 9. Confocal images showing the distribution of actin (red) and vinculin (green) in fibrobalsts cultured for 3 days: (a–c) on the low-rough patterned surface (A1), (d–f) onthe highly-rough patterned surface (A4). Double stained images are also shown.

2718 A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720

was realised though exertion of contact strain [43]. The filopodia ofgrowth cones play an important role in feeling the environment.On a surface exhibiting a suitable roughness, the cell soma andthe processes would adopt a shape complementary to the surfaceprofile, achieving maximum contact area, and therefore, interfacialforce. This force will benefit cell adhesion and spreading on thesubstrate surface, so that cells recognize and migrate to areas ofoptimum roughness.

There have been a lot of reports on studying the effect of rough-ness and surface energy on cellular responses. Both discrete spec-imens and surface energy gradients were used in these studies andmany different outcomes have been observed. In some cases, cellfunctions are enhanced on hydrophilic surfaces [37,44–48]whereas, in other cases, cell functions are enhanced on hydropho-bic surfaces [49–53]. While in other cases, surface energy has noeffect on cell functions [46,53] or cell functions have a maximumat an intermediate surface energy [48,53–55]. This broad range ofoutcomes is possibly a result of the wide variability in experimen-tal conditions such as cell types, incubation times, culture condi-tions, surface chemistries and topographies. In this study, it isattempted to discriminate among the different effects by studyingthe effect of surface energy for the whole range of contact angles,from superhydrophobicity to superhydrophilicity, while keepingthe same micro/nano morphology or vice versa. In this respect,the main outcome of the current study is that cell response showsa non-monotonous dependence on the surface energy, in accor-dance to some previous observations [48,53–55]. Besides this,there are various studies on the effect of surface roughness on cellresponse. Most studies investigate the separate effect of nano- ormicro-roughness respectively [56,57]. On the contrary, due todemanding design requirements, the synergistic effect of rough-ness at micro and nano-scales on cell response has been limitedlystudied [58]. The present study focuses on cell response on sur-faces possessing micro/nano topography (i.e. microstructures dec-orated by secondary nano-features), thus investigating thesynergistic effect of micro- and nano-scale roughness on cell adhe-sion and viability. It is shown that a dual rough surface amplifieshydrophilicity towards superhydrophilicity or hydrophobicity to-wards water repellency and accordingly enhances cell-phobicityor cell-philicity for the culture substrate.

The cell spreading dependence on surface energy has an impor-tant heuristic impact in the area of biomaterials research and it iscrucial to understand this dependence on a molecular basis. Mech-anistically, one has to go back to the first reaction between a bio-material and the organism, which is protein adsorption. At highWAs, i.e. in the hydrophobic range different proteins are adsorbedcompared to the hydrophilic range. The adsorbed protein layerscomprising different proteins on the surface could therefore resultin differential tissue reactions, which would lead to stronger bioad-hesion. The investigation of protein adsorption to these artificialsurfaces needs further exploration.

5. Conclusions

In summary it is shown that laser structured Si micro andnano rough spike scaffolds with controllability of roughness ratioand surface chemistry can serve as a novel means to elucidate the3D cell–biomaterials interactions in vivo. It is demonstrated thatthe wettability of such artificial substrates can be preferentiallytuned from super-hydrophobic to super-hydrophilic throughindependently controlling roughness ratio and surface chemistry.The dependency of fibroblast cell response on the artificial struc-tures was systematically investigated and clarified that a funda-mental parameter that determines cell adhesion on 3Dsubstrates is not solely the degree of roughness or surface chem-istry but the synergy of both, which determines the wettability orsurface energy of the culture substrate. Indeed, a proper changein the surface energy for the same degree of roughness can switchthe behaviour from cell-phobic to cell-philic and vice versa andthis transition is always accompanied by a similar sharp transi-tion in surface wettability. However, and in accordance to previ-ous theoretical models describing cell response, a non-monotonicdependence of fibroblast adhesion on wettability is found.Although it appears to be a general tendency that adhesion is fa-voured on hydrophilic substrates, it is observed that cells’ spread-ing becomes optimum on low-rough substrates, independently oftheir wettability. This indicates that cell attachment is further en-hanced and facilitated by a proper form and size of surfacetopography.

Fig. 10. Pictures of the as-prepared super-hydrophobic (sample A1, on the left) andoxidized super-hydrophilic (sample B4, on the right) patterned regions immersed inwater (a) and cell culture medium (b) respectively. The silvery shine is visible onlyon the super-hydrophobic patterned region of the A1 sample while it is absent onflat regions and the less hydrophobic patterned area of the B4 sample.

A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720 2719

The silicon scaffold material design is crucial as it affects cellu-lar attachment and provides tools for micro- and nanoscaleaddressability of material architectures which allow to preciselyposition and explore the complex interaction phenomena ofanchorage-dependent cells with their in vitro environment. Hierar-chical roughness should provide physical cues for cell orientationand spreading, while etched pores may provide room for remodel-ing of tissue structures. Further studies on the behavior of differentcell types are ongoing to better understand the interactions be-tween artificial structures and adhesion proteins, but this approachoffers a potent method for tuning cell–material interactions at themolecular level. We envisage that the multi-variant platforms of-fered by the biomimetically modified substrates described here,will find potential applications in bioengineering, especially inguiding cell adhesion, motility and differentiation as well as inthe development of tissue scaffolds and cell microarrays, in whichthe ability to switch from strong to weak cell–substrate adhesion isbeneficial.

Acknowledgements

This work was performed at the Ultraviolet Laser Facility oper-ating at IESL-FORTH and has been supported in part by the Euro-pean Commission through the Research Infrastructures activity ofFP6 (‘‘Laserlab-Europe” RII3-CT-2003-506350). The continuoussupport of Aleka Manousaki and Alexandra Siakouli with the Scan-ning Electron Microscopy is acknowledged.

Appendix A. Figures with essential colour discrimination

Certain figures in this article, particularly Figures 1, 4, 6 and 8–10, are difficult to interpret in black and white. The full colourimages can be found in the on-line version, at doi:10.1016/j.actbio.2010.01.016.

References

[1] Causa F, Netti PA, Ambrosio L. A multi-functional scaffold for tissueregeneration: the need to engineer a tissue analogue. Biomaterials2007;28:5093.

[2] Falconnet D, Csucs G, Grandin HM, Textor M. Surface engineering approachesto micropattern surfaces for cell-based assays. Biomaterials 2006;27:3044.

[3] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage.Biomaterials 2000;21:2529.

[4] Cen L, Liu W, Cui L, Zhang WJ, Cao YL. Collagen tissue engineering:development of novel biomaterials and applications. Pediatric Res2008;63:492.

[5] Kurella A, Dahotre NB. Review paper: surface modification for bioimplants: therole of laser surface engineering. J Biomater Appl 2005;20:5.

[6] Stevens MM, George JH. Exploring and engineering the cell surface interface.Science 2005;310:1135.

[7] Cyster LA, Parker KG, Parker TL, Grant DM. The effect of surface chemistry andnanotopography of titanium nitride (TiN) films on 3T3-L1 fibroblasts. J BiomedMater Res Part A 2003;67A:138.

[8] Cyster LA, Parker KG, Parker TL, Grant DM. The effect of surface chemistry andnanotopography of titanium nitride (TiN) films on primary hippocampalneurones. Biomaterials 2004;25:97.

[9] Suh JY, Jang BC, Zhu XL, Ong JL, Kim K. Effect of hydrothermally treated anodicoxide films on osteoblast attachment and proliferation. Biomaterials2003;24:347.

[10] Backman L. Shape changes of the human red-cell studied by aqueous 2-phasepartition. Febs Lett 1990;262:107.

[11] Salloum DS, Olenych SG, Keller TCS, Schlenoff JB. Vascular smooth muscle cellson polyelectrolyte multilayers: hydrophobicity-directed adhesion and growth.Biomacromolecules 2005;6:161.

[12] Flemming RG, Murphy CJ, Abrams GA, Goodman SL, Nealey PF. Effects ofsynthetic micro- and nano-structured surfaces on cell behavior. Biomaterials1999;20:573.

[13] Teixeira AI, Nealey PF, Murphy CJ. Responses of human keratocytes to micro-and nanostructured substrates. J Biomed Mater Res Part A 2004;71A:369.

[14] Reichert J, Brukner S, Bartelt H, Jandt KD. Tuning cell adhesion on PTFE surfacesby laser induced microstructures. Adv Eng Mater 2007;9:1104.

[15] Lee EJ, Cuddihy JM, Kotov N. Three-dimensional cell culture matrices: state ofthe art. Tissue Eng: Part B 2008;14:61.

[16] Seidlits SK, Lee JY, Schmidt CE. Nanostructured scaffolds for neuralapplications. Nanomedicine 2008;3:183.

[17] Choi CH, Hagvall SH, Wu BM, Dunn JCY, Beygui RE, Kim CJ. Cell interactionwith three-dimensional sharp-tip nanotopography. Biomaterials2007;28:1672.

[18] Ma ZW, Kotaki M, Inai R, Ramakrishna S. Potential of nanofiber matrix astissue-engineering scaffolds. Tissue Eng 2005;11:101.

[19] Silva GA, Czeisler C, Niece KL, Beniash E, Harrington DA, Kessler JA, et al.Selective differentiation of neural progenitor cells by high-epitope densitynanofibers. Science 2004;303:1352.

[20] Roach P, Eglin D, Rohde K, Perry CC. Modern biomaterials: a review-bulkproperties and implications of surface modifications. J Mater Sci-Mater Med2007;18:1263.

[21] Ghosh K, Ingber DE. Micromechanical control of cell and tissue development:implications for tissue engineering. Adv Drug Deliv Rev 2007;59:1306.

[22] Stratakis E, Ranella A, Farsari M, Fotakis C. Laser based micro/nano-engineering for biological applications. Progr Quant Electron 2009;33:127.

[23] Grinnell F. Cellular adhesiveness and extracellular substrate. Int Rev Cytol1978:65.

[24] Wong JY, Leach JB, Brown XQ. Balance of chemistry, topography, andmechanics at the cell–biomaterial interface. Issues and challenges forassessing the role of substrate mechanics on cell response. Surf Sci2004;570:119.

2720 A. Ranella et al. / Acta Biomaterialia 6 (2010) 2711–2720

[25] Reinhardt C, Passinger S, Zorba V, Chichkov BN, Fotakis C. Replica molding ofpicosecond laser fabricated Si microstructures. Appl Phys A-Mater Sci Process2007;87:673.

[26] Zorba V, Persano L, Pisignano D, Athanassiou A, Stratakis E, Cingolani R, et al.Making silicon hydrophobic: wettability control by two-lengthscalesimultaneous patterning with femtosecond laser irradiation. Nanotechnology2006;17:3234.

[27] Oner D, McCarthy TJ. Ultrahydrophobic surfaces. Effects of topography lengthscales on wettability. Langmuir 2000;16:7777.

[28] Zorba V, Stratakis E, Barberoglou M, Spanakis E, Tzanetakis P, Fotakis C.Tailoring the wetting response of silicon surfaces via fs laser structuring. ApplPhys A-Mater Sci Process 2008;93:819.

[29] Zorba V, Stratakis E, Barberoglou M, Spanakis E, Tzanetakis P, Anastasiadis SH,et al. Biomimetic artificial surfaces quantitatively reproduce the waterrepellency of a lotus leaf. Adv Mater 2008;20:4049.

[30] Barberoglou M, Zorba V, Stratakis E, Spanakis E, Tzanetakis P, Anastasiadis SH,et al. Bio-inspired water repellent surfaces produced by ultrafast laserstructuring of silicon. Appl Surf Sci 2009;255:5425.

[31] Lim JY, Shaughnessy MC, Zhou ZY, Noh H, Vogler EA, Donahue HJ. Surfaceenergy effects on osteoblast spatial growth and mineralization. Biomaterials2008;29:1776.

[32] Andersson AS, Backhed F, von Euler A, Richter-Dahlfors A, Sutherland D,Kasemo B. Nanoscale features influence epithelial cell morphology andcytokine production. Biomaterials 2003;24:3427.

[33] Charest JL, Garcia AJ, King WP. Myoblast alignment and differentiation on cellculture substrates with microscale topography and model chemistries.Biomaterials 2007;28:2202.

[34] Dalby MJ, Riehle MO, Sutherland DS, Agheli H, Curtis ASG. Changes infibroblast morphology in response to nano-columns produced by colloidallithography. Biomaterials 2004;25:5415.

[35] Schakenraad JM, editor. Biomaterials science. Academic Press; 1996.[36] Manly RS, editor. Adhesion in biological systems. New York: Academic Press;

1970.[37] Schakenraad JM, Busscher HJ, Wildevuur CRH, Arends J. The influence of

substratum surface free-energy on growth and spreading of human-fibroblastsin the presence and absence of serum-proteins. J Biomed Mater Res1986;20:773.

[38] Ingber DE. Cellular tensegrity: defining new rules of biological design thatgovern the cytoskeleton. J Cell Sci 1993;104:613.

[39] Wenzel RN. Ind Eng Chem 1936;28:988.[40] Cassie ABD, Baxter S. Trans Faraday Soc 1944;40:546.[41] Flynn MR, Bush JWM. Underwater breathing: the mechanics of plastron

respiration. J Fluid Mech 2008;608:275.[42] Lafuma A, Quere D. Superhydrophobic states. Nature Mater 2003;2:457.

[43] Dembo M, Wang YL. Stresses at the cell-to-substrate interface duringlocomotion of fibroblasts. Biophys J 1999;76:2307.

[44] Hallab NJ, Bundy KJ, O’Connor K, Moses RL, Jacobs JJ. Evaluation of metallic andpolymeric biomaterial surface energy and surface roughness characteristics fordirected cell adhesion. Tissue Eng 2001;7:55.

[45] Lee SJ, Khang G, Lee YM, Lee HB. The effect of surface wettability on inductionand growth of neurites from the PC-12 cell on a polymer surface. J ColloidInterf Sci 2003;259:228.

[46] Ruardy TG, Schakenraad JM, Vandermei HC, Busscher HJ. Adhesion andspreading of human skin fibroblasts on physicochemically characterizedgradient surfaces. J Biomed Mater Res 1995;29:1415.

[47] Vandervalk P, Vanpelt AWJ, Busscher HJ, Dejong HP, Wildevuur CRH, Arends J.Interaction of fibroblasts and polymer surfaces – relationship between surfacefree-energy and fibroblast spreading. J Biomed Mater Res 1983;17:807.

[48] Vanwachem PB, Beugeling T, Feijen J, Bantjes A, Detmers JP, Vanaken WG.Interaction of cultured human–endothelial cells with polymeric surfaces ofdifferent wettabilities. Biomaterials 1985;6:403.

[49] Lee JH, Jeong BJ, Lee HB. Plasma protein adsorption and platelet adhesion ontocomb-like PEO gradient surfaces. J Biomed Mater Res 1997;34:105.

[50] Lee JH, Khang G, Lee JW, Lee HB. Platelet adhesion onto chargeable functionalgroup gradient surfaces. J Biomed Mater Res 1998;40:180.

[51] Lee JH, Lee SK, Khang G, Lee HB. The effect of fluid sheer stress on endothelialcell adhesiveness to polymer surfaces with wettability gradient. J Colloid InterfSci 2000;230:84.

[52] Redey SA, Razzouk S, Rey C, Bernache-Assollant D, Leroy G, Nardin M, et al.Osteoclast adhesion and activity synthetic hydroxyapatite, carbonatedhydroxyapatite, and natural calcium carbonate: relationship to surfaceenergies. J Biomed Mater Res 1999;45:140.

[53] Ruardy TG, Moorlag HE, Schakenraad JM, VanderMei HC, Busscher HJ. Growthof fibroblasts and endothelial cells on wettability gradient surfaces. J ColloidInterf Sci 1997;188:209.

[54] Kennedy SB, Washburn NR, Simon CG, Amis EJ. Combinatorial screen of theeffect of surface energy on fibronectin-mediated osteoblast adhesion,spreading and proliferation. Biomaterials 2006;27:3817.

[55] Lee JH, Lee JW, Khang G, Lee HB. Interaction of cells on chargeable functionalgroup gradient surfaces. Biomaterials 1997;18:351.

[56] Jager M, Zilkens C, Zanger K, Krauspe R. Significance of nano- andmicrotopography for cell–surface interactions in orthopaedic implants. JBiomed Biotechnol 2007.

[57] Kim MS, Khang G, Lee HB. Gradient polymer surfaces for biomedicalapplications. Progr Polym Sci 2008;33:138.

[58] Zhao G, Raines AL, Wieland M, Schwartz Z, Boyan BD. Requirement for bothmicron- and submicron scale structure for synergistic responses of osteoblaststo substrate surface energy and topography. Biomaterials 2007;28:2821.