Small surface nanotopography encourages fibroblast and osteoblast cell adhesion

Cite this: RSC Advances, 2013, 3,10309

Small surface nanotopography encourages fibroblastand osteoblast cell adhesion3

Received 5th December 2012,Accepted 7th May 2013

DOI: 10.1039/c3ra23193c

www.rsc.org/advances

Renee V. Goreham,a Agnieszka Mierczynska,ab Louise E. Smith,a Rossen Sedevb

and Krasimir Vasilev*a

In this paper, we report the initial response of 3T3 fibroblast and MG63 osteoblast cells to engineered

nanotopography gradients of three nanoparticle diameters (16 nm, 38 nm and 68 nm). These

nanoengineered surfaces were designed to provide a range of nanoparticle densities and comparable

surface area across the gradients of different nanoparticle sizes. Importantly, we provided a uniform

surface chemistry in order to be able to examine the effect of pure surface nanotopography. We found

that nanotopography features of 16 nm encourage the adhesion of both cell types and that there is a

critical nanoparticle density between 50 and 140 particles per mm2 where cells adhered in the greatest

numbers. When nanotopography features increased to 38 nm the 3T3 cells adhered and spread well,

however, the MG63 cells adhered and spread poorly. Both cell types adhered in lower numbers when the

nanotopography feature size increased to 68 nm. This work demonstrates that there is a specific

nanotopography scale that encourages cell adhesion and spreading, however, the preferential lateral

spacing and height of the nanotopography is different for different cell types.

1. Introduction

The influence of surface topography on the adhesion,proliferation and differentiation of biological cells hasgenerated significant research activities over the last dec-ade.1–4 This emerging research niche stems from the fact thatin vitro and in vivo cells encounter various topographicalfeatures ranging from micrometre to nanoscale dimensionsand thus understanding how these features affect cellresponse is important. In vitro the material surface used forcell attachment can be topographically designed to induceparticular cell behaviour such as controlled adhesion andproliferation, and can also influence differentiation in the caseof stem cells.5 Furthermore, all biomedical implantabledevices carry some degree of surface roughness that can beintentionally introduced but in most cases it results simplyfrom material processing procedures.4,6 Revealing the effect ofsurface topography (regardless its origin) on relevant cells canguide the design of a new generation of implantable devicesresulting in better biocompatibility and tissue integration.1,6,7

In the body, cells experience various topographies, such asother cells, particles (i.e. hydroxyapatite crystals in bonematrix) and fibres (i.e. extracellular matrix proteins) andhence, understanding the influence of these structures at themicron and nanoscale can be important for unravelling manyphysiological phenomena.8–10

Despite the field being still in its infancy and little evidenceis available about the effect of surface nanotopography oncellular response in vivo, the work of various researchers in cellculture demonstrates that surface nanotopography is animportant parameter that needs to be taken into considerationby biomedical device manufacturers when designing implan-table constructs because it can significantly affect cellfunction. The impact of surface topography at the micron,sub-micron and nanoscales on various cell types has beeninvestigated by employing different topographical scales andstructures and several instructive reviews have been pub-lished.6,11–18 Various micro- and nano-structures ranging fromrandom to well- controlled and organised pores,19 grooves,14

pillars20 and colloidal particles3 have been used to investigatethe adhesion, proliferation and differentiation of several typesof cells. Although much of the data is contradictory, somehypotheses have been proposed in an attempt to explain thephenomena underlying the influence of surface nanotopogra-phy on cellular behaviour. Many of the current concepts arebased on the RGD (Arginine-Glycine-Aspartic acid) signallingpathways.2,21 This is not surprising as it is well accepted thatproteins are the first to adsorb to the surface of a materialexposed to a biological fluid.22 Subsequent cell adhesion to

aMawson Institute, University of South Australia, Mawson Lakes 5095, Australia.

E-mail: [email protected] Wark Research Institute, University of South Australia, Mawson Lakes 5095,

Australia

3 Electronic supplementary information (ESI) available: TEM images of thenanoparticles used for generation of nanotopography gradients; AFM imagesacross the gradients of different nanoparticle sizes; XPS characterisation beforeand after deposition of a plasma polymer overlayer; cross sections across theAFM image. See DOI: 10.1039/c3ra23193c

RSC Advances

PAPER

This journal is � The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 10309–10317 | 10309

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article OnlineView Journal | View Issue

surfaces is mediated by surface adsorbed cell adhesionproteins such as fibronectin, and vitronectin.23,24 Overall, thefocal adhesion based concepts stipulate that surface nanoto-pography influences cellular behaviour through altering thenumber of available integrin binding sites (through theincreased surface area owing to the nanotopography) but alsointegrin orientation and clustering.3,21

This paper aims to contribute to the knowledge in thisemerging field by determining cellular responses to nanoto-pography as a function of both height and lateral spacing ofthe nanoscale features. To achieve this goal, we fabricatednumber density gradients of gold nanoparticles of threedifferent sizes. Then, we converted these nanoparticle densitygradients to gradients of pure surface topography throughdeposition of a 5 nm thin plasma polymer film over thesurface bound nanoparticles. The last step is vital (but oftenignored) in order to place under control other surfaceproperties such as wettability and chemistry which also affectcell behaviour. The benefit of using surface gradients is that arange of parameters (in the present work lateral spacing andsurface density of the nanoparticles) relevant to cellularbehaviour can be studied with a single surface. This isimportant since issues such as sample heterogeneity andexperimental errors associated with the use of multiplesamples can be eliminated.25,26 To the best of our knowledgenanotopography gradients have been employed on only oneother occasion, by Kunzler et al.,3 however only one nano-particle size was employed, 73 nm. Substrata with controlledspacing between the nanoparticles have also been employed byHuang et al.,2 small (10–15 nm) gold nanoparticles were usedfor docking RGD peptides in order to examine the effect ofpeptide density on osteoblast adhesion.

The focus of this paper is on the initial stages of adhesionof 3T3 fibroblast and MG63 osteoblast cells as a function ofthe lateral spacing and vertical amplitude of surface nanoto-pography. These cells were chosen for this first study on thegrounds that they are well-established and characterisedsubstrate dependent cell lines, and have been extensivelyemployed when investigating cellular responses to sur-faces.7,27–29

2. Experimental

Materials

Allylamine (AA) (98%, Aldrich), hydrogen tetrachloroaurate(HAuCl4) (99.9985%, ProSciTech), trisodium citrate (99%, BHDChemicals, Australia Pty. Ltd.), and 2-mercaptosuccinic acid(97%, Aldrich) were used as received. For solution preparationand glassware cleaning, high purity water was used, producedby the sequential treatments of reverse osmosis, two stages ofmix bed ion exchange, two stages of active carbon treatment,and a final filtering step through a 0.22 mm filter. The finalconductivity was less than 0.5 mS cm21 with a surface tension72.8 mN m21 at 20 uC.

Plasma polymerization

Plasma polymerization was carried out in a custom-builtreactor described previously using a 13.56 MHz plasmagenerator.30 Deposition of plasma polymer allylamine wascarried out at a flow rate of 10 sccm, power of 10 W and time ofdeposition of 4 min (underlying coating) or 1 min (topcoating), resulting in nitrogen-rich films with thicknesses ofy20 nm and y5 nm respectively. Before deposition, allsubstrates were cleaned by oxygen plasma for 2 min using apower of 20 W.

Synthesis of gold nanoparticles (AuNPs)

AuNPs were synthesized by citrate reduction of HAuCl4.Particles of 16, 38, 68 nm in diameter were synthesized from50 ml of a 0.01% boiling solution of HAuCl4, to which 1, 0.5and 0.3 ml (respectively) of a 1% solution of sodium citratewas added under vigorous stirring.31 The solution was left toboil for 20 min and then allowed to cool to ambienttemperature. The AuNPs were then surface modified with2-mercaprosuccinic acid as in Zhu et al.32 The particlediameters were confirmed via AFM imaging.

Gradients preparation

The nanoparticles were gradually absorbed onto the functio-nalized amine substrates by controlled immersion into thesolution of AuNPs. The substrates were dipped gradually witha linear motion drive (Zaber T-LSR series), using Zabersoftware. After the required length of 10 mm immersed forthe required exposure time of 2, 4 and 6 h for the 16, 38 and 68nm AuNPs respectively, the substrate was immediatelyremoved and thoroughly washed with Milli-Q water to removeall weakly bound species

Characterisation of nanotopography

The number of nanoparticles per area was calculated from theAFM images. RMS was determined from the AFM images usingthe NT-MDT software. The surface coverage was calculated bymultiplying the number of nanoparticles times the diameter ofthe corresponding nanoparticles. The increase in surface areawas calculated by multiplying the number of nanoparticlestimes the surface area of a sphere having a diameter of thecorresponding nanoparticle. The total number of particles is Nand the total area is A. The interparticle distance is calculatedas follows: the (average) number of particles per unit area is n= N/A. The (average) area per particle is L2 (assumed to be asquare; L is the average distance between two particles).Therefore n = N/a = 1/L2. The coverage is N*d2/A = nd2 = d2/L2 (dis the size of the particle, A is the unit area, N is the particlesand n is the average number of particles per unit area).

Atomic force microscopy (AFM)

An NT-MDT NTEGRA SPM atomic force microscope (AFM) wasused in non-contact mode to provide topographical images.Silicon nitride non-contact tips coated with Au on thereflective side (NT-MDT, NSG03) were used and had resonancefrequencies between 65 and 100 kHz. The amplitude ofoscillation was 10 nm, and the scan rate for 5 mm 6 5 mmimages was 0.5 Hz. The scanner used had a maximum range of

10310 | RSC Adv., 2013, 3, 10309–10317 This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

100 mm and was calibrated using 1.5 mm standard grids with aheight of 22 nm.

Cell culture

MG63 osteoblast and 3T3 fibroblasts cells (purchased fromATCC) were cultured in Dulbecco’s modified Eagle medium(DMEM) (Invitrogen, Gibco Australia) supplemented with 10%(v/v) foetal bovine serum (FBS) (Sigma Aldrich), 100 IUpenicillin and 100 mg streptomycin (Invitrogen), and weregrown in a humidified atmosphere (95% air, 5% CO2 at 37 uC)Cells were subcultured using a solution of trypsin-EDTA (1%).The substrates were UV-sterilised for 30 min in a Topsafe ClassII Bio-Cabinet and further rinsed with sterile PBS. Thistreatment did not appear to significantly change the physicalor chemical properties of the material. The sterile sampleswere placed in sterile 24-well microplates and 7.5 6 103 cellsper well were seeded on top of the substrates in 200 ml ofmedia and the control samples. Cell adhesion and motilitywere examined after 6 h.

Cell staining

Cells grown on the substrates after 6 h were washed with pre-warmed PBS and fixed in a 4% paraformaldehyde solution.After rinsing with PBS, the cells were permeabilised byexposure to a 0.5% Triton X-100 in PBS solution for 3 min atroom temperature. The substrates were then incubated withAlexa Fluor 488 phalloidin (Invitrogen, lex 495 nm, lem 518nm) (1 : 100 dilution in PBS) to stain the f-actin and the nucleiwere stained with DAPI (Invitrogen, lex 350 nm, lem 470 nm)(1 : 1000 dilution in PBS) for 30 min at room temperature andrinsed 3 times with PBS. Subsequently, the substrates weresealed on a coverslip to be viewed using an inverted Nikon A1Rlaser scanning confocal microscope. Image J software(National Institutes of Health, Bethesda, MD) was used tocalculate the cell number and area by first calibrating imagesusing an embedded scale bar. The cell number was countedevery 2 mm across the gradient (i.e.: 1, 3, 5, 7, 9 and 11 mm intriplicate) and was repeated three times. Similarly, using thecell circumference the cell area was drawn manually and theaverage cell area was calculated.

Kruskal–Wallis statistical values were calculated using IBMSPSS statistics 20 software. Comparisons between measure-ment series were made using non-parametric tests becausevariances were not homogenous and observed values were notdistributed normally, even after appropriate transformations.

3. Results

In order to comprehensively examine the effect of surfacenanotopography on cellular behaviour we generated numberdensity gradients of near monodispersed gold nanoparticleswith diameters of 16 nm, 38 nm and 68 nm (TEM Images areshown in Fig. S1 in the ESI3). Our goal was to present to thecells substrates having similar surface coverage, surface areaand wettability. We consider important having these surfaceproperties comparable because of their influence on proteinswhich rapidly adsorb to the surface before cells do. It is well

known that surface wetting affects the conformation ofadsorbed proteins and the amount of absorbed protein isdependent on the available surface area.33–35 For this reasonwe chose to tune the nanoparticle surface density in a mannerthat allows wetting, surface coverage and surface area onseveral positions across the gradients of the three nanoparticlediameters to be comparable.

Fig. 1 shows a schematic of the strategy for gradientpreparation and a set of AFM images (5 mm 6 5 mm) across thegradient of 38 nm nanoparticles (a), nanoparticle numberacross gradients of 16 nm, 38 nm, and 68 nm diameternanoparticles (b), the calculated surface coverage as apercentage of total surface area (c), the increased percentagesurface area which results from the increased nanotopography(d), the root mean squared roughness (RMS) (e) and thecalculated average interparticle distance (f). In order to achievecomparable surface coverage and increased surface area due tointroduction of nanotopography, we adjusted the nanoparticledensity across the gradient of different nanoparticles dia-meters. Fig. 1b shows that the number of nanoparticles ofeach three sizes increases linearly across gradients. However,the number of nanoparticles at any given point across thegradients was significantly different, except at position 1 mmwhere there are no nanoparticles. The number of nanoparti-cles at the highest-density end of the gradient (position 11mm) was approximately 160 per mm2 in the case of 16 nmdiameter particles, while this number was 35 and 8 per mm2 fornanoparticles with diameters of 38 and 68 nm, respectively.The surface coverage (Fig. 1c) across the gradients increaseslinearly for the three nanoparticles and at the higher densityend of the gradients is between 2.7% and 4% which allowscomparison between several positions across the surface. Theincrease in surface area (Fig. 1d), which is calculated by takinginto account the surface area of a sphere having a diameter ofthe corresponding nanoparticle, also increases linearly acrossthe gradients and reaches values between 11% and 16% at thehigh density side of the substrates. The RMS roughnessderived from the AFM images are presented in Fig. 1e. Asanticipated, the larger nanoparticles contribute to a higherRMS values. AFM images across the surface of the gradients ofnanoparticles are shown in the ESI3 (Fig. S2). The images showthe increased nanoparticle density across the surface. Thenanoparticles are randomly adsorbed and no aggregations arepresent.

The number density gradients provide controlled nanoto-pography but these substrates are also gradients of chemistry.The gold nanoparticles used for gradient generation carrycarboxylic acid groups which are necessary for electrostaticbinding to the amine plasma polymer modified surfaces.32

The result of this is an increased concentration of these groupswith increasing nanoparticles concentration. In order toachieve gradients of pure surface nanotopography we applieda thin plasma polymer layer of about 5 nm on top of thenanoparticles. The additional plasma polymer layer wasdeposited using allylamine as a precursor for plasma deposi-tion. This resulted in the same amine functionalised surface

This journal is � The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 10309–10317 | 10311

RSC Advances Paper

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

chemistry as before nanoparticle immobilisation (XPS analysisis shown in Fig. S3 of the ESI3). We aimed for a 5 nm thickpolymer layer because previous work has shown that filmsprepared from allylamine are continuous36,37 yet thin enoughnot to alter the magnitude of the surface nanotopography. Toconfirm this, cross sections of AFM images of surfaceimmobilised 38 nm nanoparticles before and after deposition

of the additional plasma polymer films were produced and areshown in the ESI3 (Fig. S4).

The importance of ensuring that pure surface nanotopo-graphy gradients were generated, water contact angle mea-surements were done as shown in Fig. 2. The water contactangle of the ‘‘as prepared’’ gradients decreases with increasingnanoparticles density. This is because of the increasing

Fig. 1 Quantitative analysis of nanoparticle gradients: (a) schematic of gradient preparation and set of AFM images (5 mm 6 5 mm) across the gradient of 38 nmnanoparticles; (b) number of nanoparticles across the gradients; (c) surface coverage (%); (d) increase in surface area (%) due to nanotopography; (e) RMS roughness(root mean squared); (f) average interparticle distance. In all graphs squares (&) represent nanoparticles of 16 nm; circles ($) represents nanoparticles of 38 nm andtriangles (m) represents nanoparticles of 68 nm.

Fig. 2 Sessile drop water contact angle (a) across the gradients of nanoparticles before and (b) after coating with additional plasma polymer overlayer.

10312 | RSC Adv., 2013, 3, 10309–10317 This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

density of carboxylic acid groups which are more hydrophilicthan amine groups at pH 7.38 Once a 5 nm thin allylamineplasma polymer film is deposited on top of the immobilisednanoparticles the water contact angle increases towards thehigher nanoparticle densities across the gradients. This is thetrend expected if surface nanotopography is the only factorthat comes into play.39

3T3 mouse fibroblasts and MG63 osteoblast cell lines werecultured on the gradients of the three nanoparticles sizes for 6h to examine initial cell adhesion. We found that this timeframe is optimal to allow for the cells to adhere strongly to thesurface but is insufficient for significant cell proliferation tooccur. The cell lines selected for this study are known for their‘‘surface responsiveness’’ and widely used in studies focussedon understanding the effects of surface properties on cellularbehaviour.7,27–29

Fig. 3 and 4 shows laser scanning confocal microscopeimages of 3T3 (Fig. 3) and MG63 (Fig. 4) cells cultured from sixpositions with increasing nanoparticle density (1, 3, 5, 7, 9 and11 mm) across the nanotopography gradients of the threenanoparticle sizes. Staining of the nucleus (blue) allows cellnumber quantification. Staining of the f-actin in the cytoplasm(green) allows evaluation of cell spreading, which is acceptedas an indication of the health and metabolism of the cells. Theimages show an overall larger number of 3T3 and MG63 cellson the gradient of 16 nm nanoparticles. However, there arealso differences across the surface. It appears that both typesof cells adhere in larger numbers to moderate nanoparticlesdensities relative to the smooth end (position 1 mm) and theend with highest density of nanoparticles (position 11 mm).The f-actin staining shows that both types of cells are well

spread on the medium density of the 16 nm nanoparticlesgradient. On the gradient of 38 nm nanoparticles both cellslines adhered in overall lower numbers, but also showeddifferent behaviour. The 3T3 cells adhered in higher numbersand had an increased surface area compared to the MG63cells. Both cell types adhered in low numbers and hadrelatively rounded morphology on the gradients of 68 nmnanoparticles.

Quantitative evaluation of the cell numbers and cellspreading area for both cells lines across the gradient of thethree nanoparticles sizes are shown in Fig. 5. As supported bythe images in Fig. 3 the number of 3T3 cells (Fig. 5a) washighest between positions 3 mm and 9 mm across the gradientof nanotopographical features of 16 nm, which corresponds tonanoparticles densities between 50 and 140 particles per mm2.The highest number of cells was found on position 5 mmwhere nanoparticles density was 90 particles per mm2. Thenumber of cells was the lowest on position 1 mm where therewere no nanoparticles and on position 11 mm wherenanoparticles density was 160 particles per mm2 (averageinterparticles distance of 107 nm). Across the gradients of 38nm nanoparticles the 3T3 cells also adhered in largestnumbers in the medium nanoparticles density, however theoverall number of cells was lower compared to the surfaceswith immobilised 16 nm nanoparticles. The cell numbersacross the gradient of 68 nm nanoparticles was the lowestcompared to the other two nanoparticles sizes with a slighttrend of decreasing cell numbers with increasing the numberof nanoparticles. It is notable that upon the introduction ofnanotopography of all three sizes the number of cells waslarger than on position 1 mm where there are no nanoparti-

Fig. 3 Laser scanning confocal microscope images of 3T3 fibroblasts 6 h post-seeding, stained for f-actin (green) and nuclei (blue), where position 1 mm representsthe no nanoparticle density and 11 mm the highest AuNP density. Images for each row are taken from the same sample. Scale 50 mm.

This journal is � The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 10309–10317 | 10313

RSC Advances Paper

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

cles. Similarly the MG63 cells adhere in greatest number in themiddle density of the nanotopography of 16 nm (Fig. 5b).However, on the nanotopography of 38 nm and 68 nm the

MG63 cells adhere in relatively low numbers and do not seemto be influenced by surface nanotopography.

Evaluation of cell spreading area (Fig. 5c and 5d) shows thatthe 3T3 cells are well spread on all nanotopography sizes. The

Fig. 4 Laser scanning confocal microscope images of MG63 cells 6 h post-seeding, stained for f-actin (green) and nuclei (blue), where position 1 mm represents nonanoparticle density and 11 mm the highest AuNP density. Images for each row are taken from the same sample. Scale 50 mm.

Fig. 5 Analysis of the average cell number per 665 6 665 mm image and average individual cell area across the gradients of different nanotopographies. Dataexpressed as mean ¡ standard deviation 3 replicates per experiment, comparison by Kruskal–Wallis. *p , 0.05, **p , 0.01, n = 9, where position represented thecontrol with no nanoparticles.

10314 | RSC Adv., 2013, 3, 10309–10317 This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

largest spreading area was determined to be in the middle ofthe gradients and was larger than on the ends of the surfaceswithout nanoparticles. The MG63 cells showed similar trendand degree of spreading as the 3T3 cells on the 16 nmnanoparticle gradients, however the cell area was noticeablysmaller on the 38 nm and 68 nm nanotopographies.

4. Discussion

These results demonstrate that surface topography at thenanoscale significantly influences the adhesion and spreadingof 3T3 and MG63 cells. Although surface nanotopography isknown to influence the behaviour of different types of cells, tothe best of our knowledge, this is the first study to examine ina systematic manner cell adhesion on surfaces of controllednanotopography in lateral and vertical direction and uniformchemistry. This was achieved through employing numberdensity gradients of three different nanoparticle sizes and thesubsequent application of a 5 nm thin plasma polymeroverlayer. The data shows that for both cell lines involved inthis study there are specific nanotopographical characteristicsthat lead to preferential adhesion and spreading of the cells.On nanoparticles of 16 nm diameters both cell lines adhere ingreater numbers on the medium nanoparticles densitieswhere the nanoparticles densities are between 50 and 140particles per mm2 and the average interparticle distances arebetween 140 nm and 90 nm (Kruskal–Wallis: comparison ofposition 1 mm and position 9 mm for 3T3 x2 = 11.627, p =0.001; for MG63 x2 = 10.834, p = 0.001). On these areas the cellsalso show greatest spreading. When the size of nanotopogra-phy increases to 38 nm the 3T3 cells again adhere in largenumbers and are well spread in the medium nanoparticledensity range. In contrast, the MG63 cell adhesion andspreading was not significantly different from that of thesmooth surface. When the magnitude of nanotopography wasincreased to 68 nm both cell lines did not show a statisticallysignificant difference (see Fig. 5) in adhered cell numberscompared to smooth surface, although the 3T3 cells seemedbetter spread. One mechanistic explanation of the effect ofsurface nanotopography on cell behaviour is sought in theresultant integrin spacing and clustering.21 It is now wellestablished that once a substrate is placed in a biological fluid,within few minutes it is covered by an adsorbed proteinlayer.23,24 Some of these adsorbed proteins such as fibronectinand vitronectin present binding sites that are recognised bycell integrin’s, which are known to be important in celladhesion.21,40 Surface nanotopography, as demonstrated inthis work, leads to an increased surface area allowingincreased adsorption of proteins. If we assume that the ratioof adsorbed cell adhesion proteins on rough and smoothsurface is the same, then increasing surface roughness wouldprovide a higher density of integrin binding sites andpotentially encourage cell adhesion. This hypothesis seemsto hold for the 16 nm nanotopography. Both types of cellsadhere in higher numbers, and have larger spreading areas,

compared to smooth surfaces. However, after a criticalnanoparticle density (about 90 particles per mm2) the numberof adhered cells decreases, and at the highest nanoparticledensity becomes similar to that of the smooth surface. Wespeculate that there could be at least two reasons for theseobservations: Although the surface area increases withincreasing nanotopography, high nanoparticle density maylead to altering the orientation of cell binding sites and thusmaking them inaccessible for integrin binding events. In otherwords, despite a higher number of RGD domains, and othercell binding sites being available through the higher amountof adsorbed cell adhesion proteins; they are not presented inan appropriate configuration for integrin binding. Anotherpossibility may be related to integrin clustering which isknown to be important for strong cell adhesion.3,21 It might bepossible that the surface nanotopography features are hinder-ing such integrin clustering which is affecting cell adhesionand spreading.

As discussed above, both cell types have similar behaviouron the nanotopography features of 16 nm. However, when thenanoparticles sizes increases to 38 nm the two cell lines showdifferent behaviour. The 3T3 cells show a similar pattern tothat seen on the 16 nm nanoparticles: cell number increasesuntil a critical nanoparticle density is reached and thendecreases. The adhesion of MG63 cells does not seem to beencouraged by this magnitude of nanotopography and theirnumbers are not statistically different than on smooth surface(position 1 mm). Although the behaviour of the 3T3 cell maybe explained in the same way as on the nanotopography of 16nm, this hypothesis clearly does not hold for the MG63 cells.There could be another factor that comes into play whennanotopography increases to 38 nm. It is known that theelastic modulus of the osteoblast cell cytoskeleton is greaterthan that of the fibroblast.41,42 The greater cytoskeletonrigidity MG63 may obstruct the cell membrane of conformingto the surface nanotopography. As a consequence, the celladhesion machinery is not able to establish a secure andstrong contact with the underlying surface.42,43 This hypoth-esis, which is schematically depicted in Fig. 6, is reinforced bythe relatively poor adhesion of both cell types on thenanotopography of 68 nm. However, the images in Fig. 3ashow that the 3T3 cells spread better than the MG63 cells. Thismay be interpreted as more flexible and adaptive cell adhesionmechanics, hence allowing the 3T3 cells to benefit from theincreased surface area on nanoparticles of 38 nm.

Another reason that may explain the influence of surfacenanotopography on cellular behaviour might be related to theconformation of proteins adsorbed on nanostructures. Studiesconducted with nanoparticles in solution demonstrate thatproteins bind to nanoparticles forming ‘‘soft’’ and ‘‘hard’’protein corona.44 The latter consist of tightly bound proteinswhich are unfolded.44,45 In this fashion, amino acid sequencesthat are normally hidden within the folded protein structureare presented to the cells and may lead to signalling pathwaysthat alter cell adhesion dynamics. To date, this possibility has

This journal is � The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 10309–10317 | 10315

RSC Advances Paper

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

been insufficiently explored when surface nanotopography isconcerned and requires further attention.

The results presented in this work show that there is nosimple relationship between the increased surface area causedby the increased surface density of nanotopography featuresand cellular adhesion. By using number density gradients ofnanoparticles we clearly show that a similar surface area butdifferent height of the nanotopography features leads to acompletely different cell adhesion behaviour. When the heightof surface nanotopography is small, cell adhesion andspreading benefit from the increase in surface area.However, when a certain surface area, caused by thenanotopography is reached, a reduction of the number ofadhering cells is observed. When the height of the nanotopo-graphy features increases some cell types may benefit from theincreased surface area, however this may not be the case forother cells. However, the different behaviour of different cellson similar topographical features should not be surprisingsince different cell types reside in tissue with differentproperties and architectures.

Due to the various substrata being employed by others instudying cell adhesion,13,14 it is not straightforward to placethis work in the context of existing literature. The only similarwork that we are aware of is the work of Kunzler and co-workers who used number density gradients of 73 nmnanoparticles (note that surface chemistry is different toours).3 Their results using osteoblast cells are comparable withour observations on nanoparticles of 68 nm. Kunzler et al.3

suggest that the increased surface density of nanoparticleshinders integrin clustering. Work by others using differentsubstrata suggest contradicting cell behaviour.13,21,46 This isnot surprising because different methods and types of surfacenanotopography were employed. In addition, different sub-strata materials were used. The latter leads to different surfacechemistries and yet another factor that makes it complicating

to fully unravel the relationship between surface nanotopo-graphy and cellular behaviour. In addition, most of the studiespublished up to now did not systematically explore variationsin the lateral spacing between the nanotopographical features.In addition, in many studies the nanotopography was notuniform in vertical directions but rather stochastic. We shouldalso stress that the nanotopography used in this work shouldnot be confused with nanotopographies produced by pores,pits and grooves because such features at the nanoscalesrather reduce the surface area available for cell adhesionwhich is contrary to our case.

The number density gradients of nanoparticles of differentdiameters used in this work combined with a thin plasmapolymer overlayer provide a platform to study nanotopographydependent cellular behaviour in a controlled manner. Suchplatforms provide in a unique manner not only controlledlateral spacing and height of nanotopographical features butalso allow the role of surface nanotopography to be investi-gated as a function of surface chemistry. In this paper, we usedamine containing films, however plasma polymerisationallows an overlayer to be produced that can have a rich varietyof chemistries and functionalities as demonstrated by us andothers.47–49 Studies in our lab are currently underway inter-rogating the role of surface nanotopography as a function ofsurface chemistry on various types of cells. As the resultspresented in this work suggest, different cell types behavedistinctly on nanotopographies of different dimensions. Thisdemonstrates the need for further research in this area usingdifferent cell types to fully reveal the underlying phenomenabut also the need to establish a more universal nanotopo-graphy platform which allows comparative analysis of the datafrom different research groups.

5. Conclusion

In summary, we have examined the effect of nanoscale surfacetopography on the initial stages of 3T3 fibroblast and MG63osteoblast cell lines. Number density gradients of nanoparti-cles of sizes of 16, 38 and 68 nm were employed to provide wellcontrolled nanotopography in lateral and vertical directions.We demonstrated that there are certain nanotopographicalcharacteristics that benefit cell adhesion and spreading. On 16nm nanotopography the number of adhered cells of both celltypes increases with increasing nanoparticle surface density.This also corresponds to an increase in cell spreading.However, after a critical nanoparticle density is reached, thenumber of adhered cells decreases and at the highestnanoparticle density the cell numbers are similar to these ona smooth surface. When the nanotopography increases to 38nm, the 3T3 cells show similar behaviour as on the 16 nmnanoparticles but the number of adhered MG63 cells and theirspreading is not significantly different to that seen on smoothsurfaces. Both cell types show relatively poor adhesion on the68 nm nanoparticles compared to the nanotopography of 16nm. Our results show that increased surface area due to

Fig. 6 Schematic depicting the effect of cytoskeleton rigidity on cell-surfacenanotopography interactions. Drawing not to scale and for illustration purposesonly. Caption.

10316 | RSC Adv., 2013, 3, 10309–10317 This journal is � The Royal Society of Chemistry 2013

Paper RSC Advances

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online

nanotopography is not sufficient to explain cell adhesion. Weagree with the common view that surface nanotopographyaffects integrin binding cites orientation and clustering but wealso suggest that the cell membrane rigidity and the flexibilityof cell adhesion machinery, which may be different for varyingcell types, is another factor that may need to be taken intoconsideration.

Notes and references

1 T. Dvir, B. P. Timko, D. S. Kohane and R. Langer, Nat.Nanotechnol., 2011, 6, 13–22.

2 J. Huang, S. V. Grater, F. Corbellini, S. Rinck, E. Bock,R. Kemkemer, H. Kessler, J. Ding and J. P. Spatz, Nano Lett.,2009, 9, 1111–1116.

3 T. P. Kunzler, T. Drobek, M. Schuler and N. D. Spencer,Biomaterials, 2007, 28, 2175–2182.

4 J. Shi, A. R. Votruba, O. C. Farokhzad and R. Langer, NanoLett., 2010, 10, 3223–3230.

5 A. M. Ross, Z. Jiang, M. Bastmeyer and J. Lahann, Small,2012, 8, 336–355.

6 F. Variola, J. B. Brunski, G. Orsini, P. T. de Oliveira,R. Wazen and A. Nanci, Nanoscale, 2011, 3, 335–353.

7 J. H. Park, R. Olivares-Navarrete, C. E. Wasilewski, B.D. Boyan, R. Tannenbaum and Z. Schartz, Biomaterials,2012, 33, 5267–5277.

8 V. Vogel and M. Sheetz, Nat. Rev. Mol. Cell Biol., 2006, 7,265–275.

9 A. Curtis and M. Riehle, Phys. Med. Biol., 2001, 46,R47–R65.

10 N. W. Karuri, S. Liliensiek, A. I. Teixeira, G. Abrams,S. Campbell, P. F. Nealey and C. J. Murphy, J. Cell Sci.,2004, 117, 3153–3164.

11 K. Anselme and M. Bigerelle, Int. Mater. Rev., 2011, 56,243–266.

12 K. Anselme, L. Ploux and A. Ponche, J. Adhes. Sci. Technol.,2010, 24, 831–852.

13 C. J. Bettinger, R. Langer and J. T. Borenstein, Angew.Chem., Int. Ed., 2009, 48, 5406–5412.

14 E. Lamers, X. F. Walboomers, M. Domanski, L. Prodanov,J. Melis, R. Luttge, L. Winnubst, J. M. Anderson,H. Gardiniers and J. A. Jansen, Nanomed.: Nanotechnol.,Biol. Med., 2012, 8, 308–317.

15 C. Y. Tay, S. A. Irvine, F. Y. C. Boey, L. P. Tan andS. Venkatraman, Small, 2011, 7, 1361–1378.

1 6 M . N i k k h a h , F . E d a l a t , S . M a n o u c h e r i a n dA. Khademhosseini, Biomaterials, 2012, 33, 5230–5246.

17 D. H. Kim, P. P. Provenzano, C. L. Smith and A. Levchenko,J. Cell Biol., 2012, 197, 351–360.

18 J. P. Spatz and B. Geiger, Cell Mechanics, Elsevier AcademicPress Inc., San Diego, 2007.

19 K. S. Brammer, C. J. Frandsen and S. Jin, Trends Biotechnol.,2012, 30, 315–322.

20 C. Padeste, H. Ozcelik, J. Ziegler, A. Schleunitz,M. Bednarzik, D. Yucel and V. Hasirci, Microelectron.Eng., 2011, 88, 1836–1839.

21 M. J. P. Biggs, R. G. Richards and M. J. Dalby, Nanomed.:Nanotechnol., Biol. Med., 2010, 6, 619–633.

22 A. Alberts, A. Johnson, J. Lewis, M. Raff, K. Roberts andP. Walter, Molecular Biology of a Cell, Garland Science, NewYork, 2002.

23 B. D. Ratner, A. S. Hoffman, F. J. Schoen and J. E. Lemons,Biomaterials Science: An Introduction to Materials Medicine,Elsevier, California, 2004.

24 P. Roach, D. Eglin and K. Rohde, J. Mater. Sci.: Mater. Med.,2007, 18, 1263–1277.

25 J. Genzer and R. R. Bhat, Langmuir, 2008, 24, 2294–2317.26 R. V. Goreham, R. D. Short and K. Vasilev, J. Phys. Chem. C,

2011, 115, 3429–3433.27 G. Weder, O. Guillaume-Gentil, N. Matthey, F. Montagne,

H. Heinzelmann, J. Voeroes and M. Liley, Biomaterials,2010, 31, 6436–6443.

28 Y.-H. Chen, Y.-C. Chung, I. J. Wang and T.-H. Young,Biomaterials, 2012, 33, 1336–1342.

29 Y. H. Lin, P. W. Peng and K. L. Ou, Implant Dent., 2012, 21,344–349.

30 J. D. Whittle, D. Barton, M. R. Alexander and R. D. Short,Chem. Commun., 2003(14), 1766–1767.

31 J. Turkevich, P. C. Stevenson and J. Hillier, Discuss. FaradaySoc., 1951, 11, 55–75.

32 T. Zhu, K. Vasilev, M. Kreiter, S. Mittler and W. Knoll,Langmuir, 2003, 19, 9518–9525.

33 J. Wei, T. Igarashi, N. Okumori, T. Igarashi, T. Maetani,B. Liu and M. Yoshinari, Biomed. Mater., 2009, 4, 1–7.

34 L.-C. Xu and C. A. Siedlecki, Biomaterials, 2007, 28,3273–3283.

35 A. Sethuraman, M. Han, R. S. Kane and G. Belfort,Langmuir, 2004, 20, 7779–7788.

36 K. Vasilev, A. Mierczynska, A. L. Hook, J. Chan, N.H. Voelcker and R. D. Short, Biomaterials, 2010, 31, 392–397.

37 A. Michelmore, A. Mierczynska, Z. Poh, R. V. Goreham,D. Losic, R. D. Short and K. Vasilev, Nanotechnology, 2011,22, 415601.

38 A. Mierczynska, A. Michelmore, T. Abhishek, R.V. Goreham, R. Sedev and K. Vasilev, Soft Matter, 2012, 8,8399–8404.

39 A. B. D. Cassie and S. Baxter, Trans. Faraday Soc., 1944, 40,546–551.

40 B. Geiger, J. P. Spatz and A. D. Bershadsky, Nat. Rev. Mol.Cell Biol., 2009, 10, 21–33.

41 E. K. F. Yim, E. M. Darling, K. Kulangara, F. Guilak and K.W. Leong, Biomaterials, 2010, 31, 1299–306.

42 T. G. Kuznetsova, M. N. Starodubtseva, N. I. Yegorenkov, S.A. Chizhik and R. I. Zhadanov, Micron, 2007, 38, 824–833.

43 K. Wyss, C. Y. Y. Yip, Z. Mirzaei, X. Jin, J.-H. Chen and C.A. Simmons, J. Biomech., 2012, 45, 882–887.

44 Z. J. Deng, M. Liang, M. Monteiro, I. Toth and R.F. Minchin, Nat. Nanotechnol., 2011, 6, 39–44.

45 E. Casals, T. Pfaller, A. Duschl, G. J. Oostingh andV. Puntes, ACS Nano, 2010, 4, 3623–3632.

46 C. P. Pennisi, A. Dolatshashi-Pirouz, M. Foss, J. Chevallier,T. Fink, V. Zachar, F. Besenbacher and K. Yoshida, ColloidsSurf., B, 2011, 85, 189–197.

47 K. S. Siow, L. Britcher, S. Kumar and H. J. Griesser, PlasmaProcesses Polym., 2006, 3, 382–418.

48 K. Vasilev, A. Michelmore, H. J. Griesser and R. D. Short,Chem. Commun., 2009(24), 3600–3602.

49 K. Vasilev, A. Casanal and H. Chalougui, et al., J. Phys.Chem. B, 2009, 113, 7059–7063.

This journal is � The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 10309–10317 | 10317

RSC Advances Paper

Publ

ishe

d on

07

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TY

OF

SOU

TH

AU

STR

AL

IA o

n 21

/02/

2014

01:

54:2

7.

View Article Online