adhesion and proliferation of osteoblast-like cells on anodic porous alumina substrates with...

106 IEEE TRANSACTIONS ON NANOBIOSCIENCE, VOL. 12, NO. 2, JUNE 2013

Adhesion and Proliferation of Osteoblast-LikeCells on Anodic Porous Alumina Substrates

With Different MorphologyMarco Salerno , Federico Caneva-Soumetz, Laura Pastorino, Niranjan Patra, Alberto Diaspro, and

Carmelina Ruggiero

Abstract—We have fabricated nanoporous alumina surfaces bymeans of anodization in oxalic acid in different conditions and usedthem as the substrates for the growth of cells from a human os-teoblast-like cell line. The rough nanoporous alumina substrateshave been compared both with smooth standard Petri dishes usedas the control and with commercial substrates of similar material.The viability of the cells has been assessed at different culture timesof 4, 11, 18, and 25 days in vitro. It turned out that the porous side ofthe galvanostatically fabricated alumina performed similar to thecontrol and better than the commercial porous alumina, whereasthe potentiostatically fabricated porous alumina performed betterthan all the other substrates at all times, and in particular at thetwo shortest time periods of 4 and 11 days in vitro. The best per-formance of the substrates is associated with intermediate surfaceroughness and feature spacing.

Index Terms—Biocompatibility, nanoporous alumina, os-teoblast, surface roughness.

I. INTRODUCTION

T HE EFFECT ON living cells of the chemical, physical,and morphological properties of the surfaces in contact

with them is crucial in tissue engineering and fabrication of scaf-folds, for example in the field of bone replacement and regener-ation [1]–[3]. Anodic porous alumina (APA) is a nanostructuredmaterial with very low or no cytotoxicity and high thermal andchemical stability [4], [5], which makes of it a promising im-plant coating candidate for osseointegration and reduction ofimplant loosening. The control over pivotal geometrical param-eters of APA such as nano-pore diameter and wall thicknessis particularly interesting for the development of tailored solu-tions in the field of bone tissue engineering [6]. Additionally,

Manuscript received July 13, 2012; revised April 03, 2013; accepted April 08,2013. Date of current version May 29, 2013. Asterisk indicates correspondingauthor.*M. Salerno is with Istituto Italiano di Tecnologia, Nanophysics Department,

I-16163, Genova, Italy (e-mail: [email protected]).N. Patra, and A. Diaspro are with Istituto Italiano di Tecnologia, Nanophysics

Department, I-16163, Genova, Italy.F. Caneva-Soumetz, L. Pastorino, and C. Ruggiero are with the University

of Genoa, Department of Communication, Computer and System Sciences,I-16145, Genova, Italy.N. Patra is with the Centre for Nanomaterials, Advanced Technologies and

Innovation, Technical University of Liberec, Studentská 1402/2, 461 17, CzechRepublic.Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TNB.2013.2257835

the APA nano-pores can be loaded with bioactive agents to im-prove fixation to surrounding bone tissue and/or elicit specificcell responses [5]–[7]. Furthermore, a morphological or chem-ical modification of APA can also be envisaged during the fab-rication itself, by proper modification of the used anodizationvoltage or electrolyte, respectively [8]–[10].In this study different types of APA surfaces have been pre-

pared and tested as a cell culture substrate for human osteoblast-like model cells. We have experimented with both commercialand home-made APA, at culture times of 4, 11, 18, and 25 daysin vitro (DIV), and the resulting viability of cells has been eval-uated and compared.

II. MATERIALS AND METHODS

A. APA Substrates Preparation

We have used both commercial APA membranes commonlyused for filtration purposes (Whatman Anopore, purchasedfrom SPI Supplies, West Chester, USA) and APA surfacesfabricated in our laboratory. The commercial APA came intwo different types, with nominal pores diameters of(sample “SPI20”) and (sample “SPI200”), respec-tively. The APA made in the laboratory was prepared in anelectrochemical cell at room temperature, by controlled modifi-cation of a 250 thick foil of ultrapure (99.999%) aluminiumas the anode. The cathode consisted of a Platinum foil ofsimilar purity and geometry. The aluminium foil was first elec-tropolished in a 1:5 v/v aqueous solution ofmixture, run for 4 min without stirring at constant currentdensity . Then, anodization was carriedout in a 0.3 M aqueous solution of oxalic acid ,typically overnight for a time period of . For fourspecimens the anodization was carried out in galvanostaticregime at current density , whereas fortwo specimens it was carried out in potentiostatic regime at

and with two-step process (sample “Pot”).In this case, a first anodization is performed for a shorter time

, which aims to pre-pattern the aluminium substratefor the second, longer anodization. In between, the first APAlayer is chemically removed by etching in a 3:1 v/v aqueoussolution of concentrated mixture atfor . The potentiostatically prepared specimens wereused as anodized, with the porous (outer) oxide surface asthe culture substrate. On the contrary, two of the galvanos-tatically prepared specimens were also used as anodized, onthe outer porous side (sample “Galv”), whereas the other two

1536-1241/$31.00 © 2013 IEEE

SALERNO et al.: ADHESION AND PROLIFERATION OF OSTEOBLAST-LIKE CELLS ON ANODIC POROUS ALUMINA 107

Fig. 1. Microscopic images of the different APA used. (a) and (b) SEM ofthe commercial APA, with 200 nm pore diameter (sample “SPI200,” imagesize ) and 20 nm pore diameter (sample “SPI20,” imagesize ), respectively. (c)–(e): AFM of the home-made APA.(c) and (d) Galvanostatically fabricated APA, outer (sample “Galv”) and innersurface (sample “GalvIn”), respectively. (e) Potentiostatically fabricated APA,outer side (sample “Pot”).

were further treated by chemical removal of the aluminiumsubstrate, by means of incubation for 10-20 min in saturatedaqueous solution of , assisted by occasional sonicationand delicate manual removal of the resulting copper overlayerwith tweezers. After this treatment the resulting APA substratesresulted in free-standing alumina layers, which allowed usto test as the cell culture substrate also the opposite (inner)oxide surface (sample “GalvIn”), originally buried under thealuminium support. This side has no open pores, but the bottomof the closed pores appears as hemispherical alumina caps in acontinuous oxide, called the barrier layer [8].

B. Morphological Characterization of the Substrates

The commercial APA surfaces were first inspected by scan-ning electron microscopy (SEM), carried out in a JSM-7500FA(Jeol, Japan) at 15 kV primary beam, using secondary elec-trons [see Fig. 1(a) and (b)]. Additionally, after fabrication ofthe home-made APA, all the substrates were imaged by atomicforce microscopy (AFM), carried out in aMFP-3D (AsylumRe-search, USA) working in tapping mode with silicon cantileversof resonant frequency. The AFM images have beentaken with 256 256 pixel sampling frequency, on scan areasof 3 3 and 10 10 .As the main figure describing the surface roughness, we have

considered the root mean square (RMS) of the distribution ofheights of the 2-D surface above the plane resulting from leastsquare fit, a parameter conventionally termed as [11]. Ad-ditionally, two more roughness parameters have been selected,namely the dominant spatial wavelength and the relative in-crease in surface area . The first has been obtained in the fre-quency space from the fast Fourier transform (FFT) of the AFMimages. All FFT maps had roughly central symmetry, showingno preferential orientation in space, and showed a ring withhigher intensity than the background at a distance fromthe origin. By inverting this spatial frequency we obtained the

. The second additional roughness parameter, ,is defined here as the relative increment in surface area nor-malized to the projected image area A in the xy scan plane, i.e.,

.

All the three parameters, , , and have been calculatedby means of the standard AFM acquisition and postprocessingsoftware coming with the instrument, based on Igor 6.22 soft-ware environment (Wavemetrics, USA).

C. Cell Cultures

Before seeding the cells, the APA substrates were sterilizedovernight under UV and subsequently under pure ethanol for1 h. Commercial human osteoblast-like MG63 cells were used(Interlab Cell Line Collection bank, Genova, Italy) as the cellmodel, which we have previously characterized extensively interms of attachment behavior to flat surfaces [12]. The cellswere first cultured in Petri dishes in a fully humidified atmos-phere in the presence of 5% (v/v) at 37 in Dulbecco’sModified Eagle’s Medium supplemented with 10% (v/v) fetalbovine serum, 0.5% antibiotics (diluted from a stock solutioncontaining 5000 U/mL penicillin, 5000 mg/mL Streptomycin)and 2 mM glutamine. Cells at the third passage in culture havebeen used to carry out the experiments. To this purpose 3800cell/ have been seeded on both standard Petri dishes as thenegative control and sterilized APA substrates. The cell viabilityhas been evaluated after 4, 11, 18, and 25 DIV by means of aTox-2 toxicology assay kit (Sigma, Milan). This kit is basedon the capability of mitochondrial dehydrogenases of viablecells to cleave the sodium salt of 2,3-bis[2-Methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxyanilide (XTT), yieldingorange water soluble formazan crystals that can be monitoredspectrophotometrically at 450 nm. At the end of each time in-terval, cell cultures were removed from the incubator into a lam-inar flow hood and XTT reagent added to fresh culture mediumin an amount equal to 20% of its volume. Cultures were re-turned to the incubator for 2.5 h and then aliquots of culturemedium from each sample were analyzed spectrophotometri-cally by means of a micro-plate reader. Cell viability has beenevaluated as the optical density at 450 nm wavelength of theculture medium to which the reagent had been added, normal-ized to the controls (standard Petri dish substrates) treated in thesame way for the respective culture period. On some samples,representative SEM images of cultured cells were also taken,after coating the specimens with a thick Cr layer.

D. Statistical Analysis

Where possible we performed analysis of vari-ance (ANOVA) on our experimental data of both APA sub-strates morphology and cell cultures viability. All pairs of sam-ples were compared for statistical significance of the differentmean values, using Bonferroni test and subsequently increasingsignificance level of 0.05, 0.01, and 0.001. For the ANOVA pro-cessing we used the built in functions of the plotting programOrigin 8.0 (Originlab, USA).

III. RESULTS AND DISCUSSION

A. Morphology of the APA Substrates

Representative microscopic images of all the different sam-ples used are presented in Fig. 1, which allows to evaluate thepeculiar geometrical properties of the different substrates. It ap-pears that the pores of the commercial APA [Fig. 1(a) and (b)]


Fig. 2. RMS roughness values for the different APA substrates, at 3 and 10scan size (void and filled circles, respectively). The lines are just guides to

the eyes. For the 10 scan size, pairs marked with , and representstatistically different values at increasing significance levels of 0.05, 0.01, and0.001, respectively.

are rather irregular in both shape and position, with thesmall ones [Fig. 1(b)] in particular showing not round butrather polygonal edges. The pores of the home-made APA[Fig. 1(c) and (e)] exhibit intermediate diameter (mean value

in both cases) and are predominantly round. Addition-ally, for the potentiostatic APA [Fig. 1(e)] the size distributionis narrower (standard deviation vs. ofFig. 1(c)), and micrometer scale domains appear with poresthat are arranged in an hexagonal lattice. This higher regularityis due to the two-step anodization [13], in combination with theconstant anodization potential, which is a prerequisite for a con-stant pore cell size and density, after the action of partial APAdissolution simultaneously occurring with new oxide growthfrom the aluminium-oxide interface [8]. A special case isgiven by the APA inner side face appearing in Fig. 1(d), whichpresents a pattern of hemispherical caps given by the closedpore bottoms. In this surface the pits occur at the crossing of thepore walls, and are not deep pores but onlyrelatively smooth depressions (depth ). Additionally,the pore spacing is larger than those of Fig. 2(c) and (e), being

vs. . The reason was the constant increase inanodization voltage required to maintain the current constantduring the APA layer growth, resulting in a final voltage of

.AFM is particularly suitable for a complete morphological

analysis of the APA substrates, since it allows one to obtain a3-D image of the surface topography, making it possible to eval-uate not only the texture but also the roughness of the respectivesurfaces. The most common roughness parameter inspected for2-D surfaces is [11], which is the RMS of the distribution ofsurface heights. In Fig. 2 the of the different APA surfaceshas been plotted, as measured by AFM on images of either 3 or10 scan size, averaged over three different regions on twodifferent specimens for each sample type . The datapoints are the mean values, with standard deviation for theerror bars.In Fig. 2 one can see that for the 10 10 scan areas

(filled circles) the is higher than for the 3 3 scan areas(void circles), as expected. Actually, for an increase in scan area

of almost a factor 10, a sublinear increase in of a factoris observed for all samples. In general, the same trend is ob-served for both scan areas, with the roughest sample being theone with the largest pores (“SPI200”), and the smoothest samplebeing the one with the smallest pores (“SPI20”). This was ex-pected, due to the finite sharpness of the AFM tip that limitsthe depth resolution. In fact, our tip had a nominal diameter of

and a cone full angle . As a consequence, thesmallest pores of sample “SPI20” could be hardly resolved bythe AFM. In fact, even the largest pores of “SPI200” (with di-ameter ) could be probed down to the negative tipshape inside each pore, corresponding to a maximum apparentdepth of . If one considers that an idealsinusoidal surface varying and wouldresult in a , which is an upper limit in the case ofbest tip, our value of for “SPI200” in Fig. 2appears reasonable.Very close in to the smallest pores sample (“SPI20,”

) is also the galvanostatic APA used on the closedpore side (“GalvIn,” ). Indeed, as alreadypointed out previously, on that surface no pores are present butrather only smooth depressions. Finally, the outer (porous) sur-faces of the oxalic acid APA (“Galv” and “Pot”) show similar( and , respectively), at an intermediatelevel between the commercial APA with small and large pores.This is also as expected, due to the intermediate and similar porediameter of . The slightly lower observed for theporous side of the potentiostatic APA (“Pot”) may be associatedwith the more regular surface, generally showing less bumpsand protruding defects than the galvanostatic APA (“Galv”), inaddition to the more regular 2-D texture.For the set of values at 10 scan size, which are more likely

to represent the surface area of attached cell in the cultures,ANOVA has been carried out on the . This analysis revealedthat almost all pairs of samples have statistically significant dif-ference in (at various significance levels, see legend of Fig. 2)but “SPI20”-“GalvIn” and “Galv”-“Pot.” Accordingly, the sam-ples have been divided into three groups identified with lettersA, B, and C, (see Fig. 2).In order to provide a more detailed characterization of the

substrate morphology, in addition to we have considered thetwo quantities defined in Section II as and . These quanti-ties have been selected out of more than 20 parameters that canbe used to quantitatively describe a 3-D surface, as describedin [14] where and correspond to and , respec-tively. The reason is that a 3-D surface can only be character-ized by a combination of different properties: i) the deviationof heights vertically above the horizontal baseplane (describedby so-called “amplitude parameters,” such as ); ii) the pat-tern of features as projected on the horizontal xy baseplane (de-scribed by so-called “spatial” parameters); and the combinationbetween the above mentioned spatial dimensions (described byso-called “hybrid” parameters). Indeed, is a “spatial” param-eter describing the surface texture, as it represents the domi-nating feature distance in the xy plane, whereas is a “hy-brid” parameter, as it represents the percent enhancement of ac-tual surface area due to the surface asperities, normalized to theprojected scan area.


Fig. 3. Additional roughness parameters and (green and blue bars), cal-culated from the AFM images with 10 scan size. Together with these,is also plotted once again (red bars).

The means and standard deviations of and have beenplotted in Fig. 3, for the 10 scan size data only, together withthe respective values of repeated from Fig. 2. represents thepore spacing, which is larger than the pore diameter itself, as itsums it up with the thickness of the wall between adjacent pores,and is expected to be up to twice the pore size, depending on theanodization conditions [8]. In fact, the ratio between and thepore diameter varies between for “SPI200” and forall the other porous sides substrates. The lower value for thelargest pores substrate is reasonable, since in the case of largestpores the cumulative effect of the wall thickness in forming thentotal cell size is relatively less important, as it can be seen forexample in Fig. 1(a) and (b).It is interesting to note that for this parameter the ANOVA

shows that the only pair of substrates not statistically different is“Galv”-“Pot.” In fact, in the same group of substrates (labeledC) also “GalvIn” has been included (see Fig. 3), despite its dif-ference with both samples “Galv” (significance level 0.05) and“Pot” (significance level 0.01). Actually, most of the other re-lationships of difference in Fig. 3 are of the 0.001 significancelevel, so obviously “GalvIn” is less different from “Galv” and“Pot” than the other substrates are, andmuch closer to those sub-strates than to “SPI20.” This situation is inverted with respect to, for which the substrate “GalvIn” could be grouped together

with “SPI20,” instead (see data points in Fig. 2 and red bars inFig. 3).Regarding the last parameter , the same behavior thancan be seen instead, where “GalvIn” can rather be grouped

together with “SPI20” (no significant difference even at theweakest level of 0.05), whereas “Galv” and “Pot” form adifferent group on their own (again no difference down to0.05 level). Probably, in the hybrid parameter a dominatinginfluence of the vertical dimension occurs, same as in ,whereas clearly describes a different behavior of the textureas compared to that of the height values.

B. Viability of the MG63 Cells

In Fig. 4 the main experimental result of this work is pre-sented. The XTT measured viability of the cells cultured on

Fig. 4. Viability of MG63 cells cultured on the APA substrates normalised togrowth in tissue culture Petri dishes monitored at different culture times of 4,11, 18, and 25 DIV.

the different types of APA substrates and at different times hasbeen plotted, normalized to the negative control (Petri dish),i.e., expressed as a percent value relative to the control with therespective culture time. The viability at the first culture time,DIV4, can be considered as an indication of the good cell adhe-sion to the substrates, whereas the viability at longer times is anindication of cell proliferation. Since the experiment has beenperformed on a limited number of APA specimens per sample

, for the error bars extension the semi-dispersion ofeach pair of measurements has been used, and no statistical con-siderations can be made comparing the different time periods.However, the apparent trend can be commented.At DIV4 the viability of cells cultured on the potentiostatic

APA substrates (“Pot”) was higher than the one mea-sured on the control (100% level). For most of the other APAsubstrates, namely the commercial APA with smaller pores(“SPI20”) and the galvanostatic APA, both porous (“Galv”)and inner surface (“GalvIn”), the viability was comparable tocontrol, but it was lower on the commercial APA with largerpores (“SPI200”). Subsequently, from DIV4 to DIV11 a quitesteep decrease of cell viability was observed for all types ofAPA substrates except for the galvanostatic APA porous surface(“Galv”). From DIV11 to DIV18 viability decreased further(though to a lesser extent than for the previous time interval)for several substrates but the commercial ones with smallerpores (“SPI20”) and both “Galv” and “GalvIn” galvanostaticones. Actually, for “GalvIn” the decrease was already so largefor the previous time period step (DIV4 to DIV11) that a furtherdecrease was somewhat unlikely. The viability on “Galv” is ofinterest, since it always remained on rather high level (close to100%). From DIV18 to DIV25 cell viability remained stableon all substrates, even with a small positive change in the caseof the commercial APA with larger pores (“SPI200”).Overall, all substrates but the galvanostatic APA (“Galv”)

showed a decrease in viability, which was quite marked for thefirst period of proliferation (from DIV4 to DIV11), and thenslowed. If an exponential profile is fit, thecharacteristic decay time found varies between 3.2 days for


Fig. 5. Same data as in Fig. 4, grouped according to different substrate type.All the datapoints are shown, to give an idea of the distribution, along with thestatistical boxes with median, 25 and 75 percentile values, in gray. Stars anddouble stars represent pairs of values that are statistically different at signifi-cance levels of 0.05 and 0.01, respectively.

“GalvIn” and 5.0 days for “SPI200.” The “Galv” substrates re-mained constant in viability, and at all times close to the con-trol level. Finally, the potentiostatic APA (“Pot”) showed a re-markable enhancement of viability at DIV4 with respect to thecontrol, and were the only substrates that remained at all timesabove the respective control values . Therefore, theseseem to be the most promising APA surfaces as they sustainedcell adhesion, growth, and proliferation for the considered celltype.This conclusion is also supported by Fig. 5, where the same

data as in Fig. 4 has been replotted, after grouping together allthe time periods for a given APA sample type. On these groupeddata we carried out an ANOVA test . This analysisrevealed that cell survival on “Pot” was significantly different(0.05 level) from all other samples. A significant difference(0.05 level) was also observed between “Galv” and “SPI200.”The difference between “Pot” and the other samples was alsosignificant at the 0.01 level with all samples but “Galv.” Thisanalysis supported the trend evident in Fig. 4 that the substrate“Pot” performed best among all, and “Galv” scored second,and was especially remarkable at the good long-time behavior(Fig. 5).Looking back at the roughness parameters described in

Figs. 2 and 3, it can be observed that the substrates most suitedfor culturing the selected cells were those with intermediatevalues of all the three roughness parameters selected. In partic-ular, the roughness and the feature spacing seem to be theparameters most suited for their identification. In particular, thesubstrate with lowest average viability (“SPI200”) was also theone with the highest and . However, the substrates withlowest and of “SPI20” and “GalvIn” did not correspondto the highest viability, and were close to the negative control(Petri dish 100% level) and to sample “Galv,” but lower than“Pot.” The highest viability of “Pot” was associated, on thecontrary, with intermediate values of both and . Obviously,optimum values of and exist, around which viability

decreases in both directions. These results are somewhatunexpected, since one could imagine that the larger pores of“SPI200” could favor the anchoring of the cell filopodia, whichtypically have a lateral size of [15], whereas thisis on the contrary the substrate with the lowest viability. Theoccurrence of best substrate-cell interaction at intermediateroughness was also observed in another recent study carriedout on mesoporous silicon substrates [16]. However, in thatwork the authors limited the culture time to 60 h, which in ouropinion is a too short time for verification of stable, mid tolong-term cell-substrate integration.

C. Conclusion

We have investigated the long-term viability of human os-teoblast-like cells cultured on substrates of APA fabricated inaqueous oxalic acid according to different recipes. We com-pared our APA substrates to standard Petri dishes as the con-trol and to commercial APA membranes with different poresize ( and ). Some of the substrates fabricatedin our laboratory, in particular those grown potentiostaticallyat 45 V, performed significantly better than the commerciallyavailable ones. As the analysis of cell growth is based only onthe XTT test, other important parameters that indicate how wellthe cells interact with the surfaces such as cell morphology, thedegree of spreading, dynamic behavior, have not been consid-ered here. However, within the limitations of the present work,the best substrates had both intermediate roughness (50-70nm on a 10 10 area) and intermediate pore size and thusspatial feature spacing ( and , respectively).The occurrence of best performance at intermediate roughnessvalues is qualitatively in agreement with other recent studiescarried out on porous silicon substrates [15]. Here we extendedthe culture period to significant times for a real assessment ofa healthy cell-substrate interaction (DIV 21), which would bethe requirement for e.g., osseointegration of APA-coated or-thopedical and dental implants. The APA surfaces represent apromising material for biomedical implants, thanks to the goodtenability of their geometrical parameters (pore and cell size de-termined by the applied anodization potential) and to the rela-tively mono-dispersed feature size (uncertainty in pore diam-eter of ). Additionally, the deep APA pores (up to some10 length) could be loaded with adhesion or growth factorsfor augmented integration as permanent implants in regenera-tive medicine.

REFERENCES

[1] T. A. Desai, “Micro- and nanoscale structures for tissue engineeringconstructs,” Med. Eng. Phys., vol. 22, pp. 595–606, 2000.

[2] K. Anselme, “Osteoblast adhesion on biomaterials,” Biomaterials, vol.21, pp. 667–681, 2000.

[3] L. F. Cooper, T. Masuda, P. K. Yliheikkila, and D. A. Felton, “Gener-alizations regarding the process and phenomenon of osseointegration,Part II: In vitro studies,” Int. J. Oral Maxillofacial Surgery, vol. 13, pp.163–174, 1998.

[4] A. R. Walpole, Z. Xia, C. W. Wilson, J. T. Triffitt, and P. R. Wilshaw,“A novel nano-porous alumina biomaterial with potential for loadingwith bioactive materials,” J. Biomed. Mater. Res., vol. 90A, pp. 46–54,2008.

[5] M. Karlsson, E. Palsgard, P. R. Wilshaw, and L. D. Silvio, “Initial invitro interaction of osteoblasts with nano-porous alumina,” Biomate-rials, vol. 24, pp. 3039–3046, 2003.


[6] S. Thorat, A. Diaspro, A. Scarpellini, M. Povia, andM. Salerno, “Com-parative study of loading of anodic porous alumina with silver nanopar-ticles using different methods,” Materials, vol. 6, pp. 206–216, 2013.

[7] E. E. L. Swan, K. C. Popat, and T. A. Desai, “Peptide-immobilizednanoporous alumina membranes for enhanced osteoblast adhesion,”Biomaterials, vol. 26, pp. 1969–1976, 2005.

[8] J. W. Diggle, T. C. Downie, and C. W. Goulding, “Anodic oxide filmson aluminum,” Chem. Rev., vol. 69, pp. 365–405, 1969.

[9] W. Lee, K. Schwirn, M. Steinhart, E. Pippel, R. Scholz, and U. Gösele,“Structural engineering of nanoporous anodic aluminium oxide bypulse anodization of aluminium,” Nature Nanotechnol., vol. 3, pp.234–239, 2008.

[10] M. Salerno, N. Patra, and R. Cingolani, “Use of ionic liquid in fab-rication, characterization, and processing of anodic porous alumina,”Nanoscale Res. Lett., vol. 4, pp. 865–872, 2009.

[11] Geometrical Product Specifications (GPS) Surface Texture: ProfileMethod—Terms, Definitions and Surface Texture Parameters, ISO4287, International Organization for Standardization, 1997.

[12] F. Caneva-Soumetz, J. Saenz, L. Pastorino, C. Ruggiero, D. Nosi, andR. Raiteri, “Investigation of integrin expression on the surface of os-teoblast-like cells by atomic force microscopy,” Ultramicroscopy, vol.110, pp. 330–338, 2010.

[13] H. Masuda and K. Fukuda, “Ordered metal nanohole arrays made bya two-step replication of honeycomb structures of anodic alumina,”Science, vol. 268, pp. 1466–1468, 1995.

[14] J. Loeberg, I. Mattisson, S. Hansson, and E. Ahlberg, “Characterizationof titanium dental implants I: Critical assessment of surface roughnessparameters,” Open Biomater. J., vol. 2, pp. 18–35, 2010.

[15] A. Mogilner and B. Rubinstein, “The physics of filopodial protrusion,”Biophys. J., vol. 89, pp. 782–795, 2005.

[16] F. Gentile, L. Tirinato, E. Battista, F. Causa, C. Liberale, E. Di Fab-rizio, and P. Decuzzi, “Cells preferentially grow on rough substrates,”Biomaterials, vol. 31, pp. 7205–7212, 2010.

Authors’ photographs and biographies not available at the time of publication.

adhesion and proliferation of osteoblast-like cells on anodic porous alumina substrates with...

Documents