The effect of different surface morphology and roughness on osteoblast-like cells

The effect of different surface morphology and roughnesson osteoblast-like cells

Michael Ball,1 David M. Grant,2 Wei-Jen Lo,2 Colin A. Scotchford2

1Department of Materials, Imperial College, London SW7 2AZ2School of Mechanical Materials, Manufacturing Engineering and Management, University of Nottingham,United Kingdom

Received 4 June 2007; accepted 14 June 2007Published online 16 November 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31652

Abstract: Increased magnitude of biomaterial surfaceroughness and micromachined-grooved surfaces has bothbeen shown to stimulate osteoblast activity, but have notbeen compared in the same study quantitatively. A seriesof titanium alloy (Ti6Al4V) samples were prepared usingsimple machining techniques to undertake such a compari-son. Samples were either grit blasted (Gb) or shot peened(Sp) to give random discontinuities, or silicon carbideground (SiC) to produce ordered grooves. These werecompared with micropolished samples (Mp). The sampleswere coated with a 1 lm continuous coating of hydroxyap-atite to remove differences in surface chemistry. Humanosteoblast-like cells were seeded onto the materials andmetabolic activity, proliferation, alkaline phosphatase ac-tivity, and osteocalcin production assessed. Cell responseswere highly dependent on the substrate that they werecultured on. Cells cultured on the smooth and ordered

(Mp and SiC, respectively) samples had higher metabolicactivity and a more elongated morphology than those cul-tured on the randomly structured Gb or Sp samples. Over21 days, cell metabolic activity peaked relative to the con-trol between 7 and 14 days on the Mp sample, andbetween 14 and 21 days on the Gb, Sp, and SiC samples.In common with other researchers, we note that micronscale topography may have potential for influencingosseointegration. More importantly, as the magnitude ofthe discontinuities on SiC, Gb, and Sp were similar, thedifferences in cell responses does not appear to lie withthe size of the features, but whether the features showedan ordered or disordered structure. � 2007 Wiley Periodi-cals, Inc. J Biomed Mater Res 86A: 637–647, 2008

Key words: surface morphology; surface roughness; osteo-blast-like cell; cell adhesion; cell differentiation

INTRODUCTION

The early interactions, between bone and a bioma-terial implant, exert a strong influence on the successor failure of the implant.1 The use and design ofmaterials that could control bone formation onimplants, to stimulate osseointegration is thereforean important goal.

Changes in cell morphology, metabolic activityand expression of markers of cell differentiation thatcorrelate with changes in surface texture or topogra-phy have been widely reported in the literature.2

Cell responses to different surface topographies ofmaterials used in orthopaedic applications have beenstudied using cultured bone-derived cells.3–8 Thesurfaces examined can be broadly separated intothose studies examining differences between surfaceswith varying magnitudes of roughness (including

smooth control surfaces) and those examining regu-

lar or precisely machined surface features, such as

microgrooves. The former studies often involve

quantitative measurement of cell activity and differ-

entiation, while the latter have primarily focused on

microscopical examination of cell morphology.

Rough and smooth surfaces described by the magni-

tude of undefined discontinuities often report Ra,

which describes the average deviation from an arbi-

trary mean line.9 However, this does not quantify

the topography in detail. Cell culture experiments

have indicated that osteoblast-like cells cultured on

rough surfaces show stronger cell adhesion10 and

high production of both differentiation-associated

growth factors11,12 and extracellular matrix pro-

teins.13 Additionally, osteoblast-like cells cultured on

materials with micron scale grooves or channels

become elongated parallel to the direction of the

grooves.14,15 Surfaces with grooves or channels have

been reported as encouraging bone formation, both

in vitro16,17 and in vivo.18

Correspondence to: M. Ball; e-mail: [email protected]

� 2007 Wiley Periodicals, Inc.

A number of questions relating to cell responsesto topography remain. Firstly, the term ‘‘roughness’’does little to define the surface shape. Methods ofcreating surface roughness have used such relativelysimple methods as acid etching,3,11,19–23 grit blast-ing,7,24 and simple sandpaper grinding.3,25 Thesemethods generate different and distinct surface top-ographies, which might provoke different responsesin cells seeded on them. While metabolic activityand differentiation have been assessed on surfacesvarying surface roughness and grooved surfaces sep-arately, they have not been directly compared withinthe same study. It is therefore not clear whether cellswould demonstrate similar responses to the groovedtopography as they would to surfaces of similar Ravalues but with random roughness. In this presentstudy, two surfaces with similar roughnesses, butdifferent surface features are compared, along with aflat mirror polished sample and a rougher samplewith random surface features.

A grooved topography can be generated simplyand consistently by grinding the surface of a mate-rial with sandpaper.3,25 Although the grinding actionis circular, the grooves produced are parallel. Theexperiments reported here were designed to investi-gate the relative influences of different surface fea-tures and roughness of hydroxyapatite (HA) coatedtitanium alloy on metabolic activity of osteoblast-likecells and differentiation, with a view to evaluatingan optimal surface for orthopaedic applications. Themethods used were simple machining techniques:shot peening, grit blasting, and sandpaper grinding.These surfaces create variations in both surface to-pography and roughness, but may also show differ-ences in crystal surface and surface chemistry due tothe energy generated by the method. For example, ithas been shown that shot peening may alter the sur-face energy of a metal.26 To reduce any differencesin crystallinity or surface chemistry, the surfaceswere subsequently coated with hydroxyapatite, a ce-ramic analogous to the inorganic mineral phase ofbone. The cells used for the in vitro assessmentswere human bone-derived osteoblast-like cells.

MATERIALS AND METHODS

Fabrication of the Ti-alloy samples

The titanium alloy, obtained in its annealed conditionhad a composition 6.48 wt % Al, 4.13 wt % V, with tita-nium as balance (Duckworth and Kent, UK). The test sam-ples were punched from a 1 mm thick sheet into discs of6 mm diameter.

The discs were then subjected to four different treatmentsto produce the different surface topographies. One set of sam-ples was bombarded with a series of 150–300 lm diameterglass beads. This set of samples was termed ‘‘shot peened’’(Sp). A second set was bombarded with silicon carbide pow-der, �100 lm in size. These samples were termed ‘‘gritblasted’’ (Gb). Both these procedures were carried out for 30 s,at a distance of 15 cm from the target, at a pressure of 0.5–1.0MPa. A further set was ground on a polishing wheel using 240grade silicon carbide sandpaper, for 3 min. These sampleswere termed ‘‘silicon carbide ground’’ (SiC). The final set wasinitially ground using the 240 grade sandpaper, then groundfor 5 min using a 1 lm grade diamond powder paper. This setwas termed the ‘‘mirrorpolished sample’’ (Mp). The surfacemodificationmethods are listed in Table I.

Hydroxyapatite coating by plasma sputtering

After fabrication, the samples were coated with hy-droxyapatite (HA) (Plasma Biotal, Tideswell, England) byplasma sputtering, to create a �1 lm thick, HA coatingthat previous work has shown to be continuous and con-forming to the contours of the surface.27

The HA coated samples were produced by a capacitivelycoupled RF plasma sputtering apparatus as reported else-where.28 Briefly, using argon as the working gas, the flowwas controlled by two 100 standard cubic centimeters perminute (sccm) mass flow controllers accurate to 60.1% offull-scale deflection. The continuously pumped reactionchamber pressure was maintained by feedback from a capac-itance manometer fed to the mass flow controller viaa controller. The Ti-alloy samples were ultrasonically cleanedin absolute ethanol for 30 min before insertion in the reactionchamber and then in situ sputter cleaned in the reactionchamber with argon as the source gas for 5 min before depo-sition. The coating coefficients are listed in Table II.

TABLE IManufacturing Parameters of the Samples

Micropolishing Shot Peening Grit Blasting Grinding

Number of samples 12 12 12 12Abrasive media 1 lm diamond

powder150�300 lm white

glass beads100 lm brownSiC powder

240 SiC paper

Time 5 min 30 s 30 sec 3 minNozzle angle – 908 908 –Nozzle distance – 15 cm 15 cm –Air pressure 80�150 lb/inch2 80�150 lb/inch2 –

638 BALL ET AL.

Journal of Biomedical Materials Research Part A

The HA sputtering target was fabricated as previouslydescribed.28 Briefly, medical grade HA powder was madeby slip casting using 160 g of HA powder, 70 g water, and2 mL ammonium polyacrylate as the dispersant, mixed to-gether to make 115 mL slurry. The slurry was poured intoa plaster mould and dried at 508C for 2 days. The driedtarget was sintered in air at 13008C for 2 h. The resultingdisc was 10 cm diameter and 5 mm thick and had a den-sity of 87% of its theoretical value. X-ray analysis was pre-viously performed on these materials.28

X-ray diffraction

X-ray diffraction was performed to assess coating crys-tallinity, using a Siemens D500 diffractometer with mono-chromated Cu Ka radiation at 20 mA and 40 kV. Scans for2u angles between 108 and 808 were made with a 0.028step size and 2.5 s dwell time.

Topography of the coated samples

The roughness of the HA coated samples was measuredby means of a SURFACOM surface texture measuringinstrument (profilometer) equipped with a 0.1 mm diame-ter tip pyramidal stylus. Roughness values of the sampleswere measured by determining the average peak to valleyheights over an area of 1 mm 3 1 mm.

Cell culture methods

Human osteoblasts used were obtained from femoralhead trabecular bone and characterized as described previ-ously.29 After isolation, cells were maintained in Dulbec-co’s Modified Eagles’ Media (DMEM) supplemented with1% penicillin/streptomycin, 1% glutamine, 10% fetal calfserum (Gibco Life Technologies, Paisley, Scotland), and50 mg/L ascorbic acid (BDH, Poole England). Cells weremaintained at 378C and 5% CO2 in sterile tissue cultureflasks (Falcon). Once confluent, the cells were dissociatedfrom the flask using 0.05 mg/mL trypsin in 10 mM HEPESin Ca, Mg free PBS, and resuspended in fresh media. Cellconcentration was assessed using a haemocytometer, via-ble cells being identified using trypan blue exclusion dye.

All HA coated titanium samples were sterilized byrepeated washings in a solution of 1% penicillin/streptomy-cin in PBS, then rinsed in PBS, and finally washed in doubledistilled H2O before use in tissue culture experiments. Thesamples were used in triplicate, and tissue culture plasticwas used as the positive control for cell growth.

Cell morphology

The samples were placed in 96-well Falcon tissue cul-ture plates (Becton-Dickinson Biosciences, Oxford, UK)and cells seeded at a density of 2 3 104 cells per well andincubated as described above. After 24 h, the cells wereremoved from media and fixed using 3% glutaraldehydein 0.1M sodium cacodylate buffer. Samples were thenwashed five times in distilled water and taken through agraded alcohol series before a final 100% dried ethanolwash, for 15 min. Cells were dried via hexamethyldisila-zane (HMDS) (Sigma) before sputter coating with gold.The cells were examined using a Jeol scanning electronmicroscope operating at 15 kV.

Shape factor

Electron micrographs were examined using Scion Image(Scion Corp, USA). Cells were outlined using a perimetertools, and the perimeter and the area calculated. The shapefactor was calculated according to the formula: shape fac-tor (sf) 5 [(area/perimeter2) 3 4p]. The closer the resultantvalue is to one, the more the cell shape tends toward to acircle. Conversely, the closer the shape factor tends tozero, the more the cells shape tends to a straight line.

Alkaline phosphatase assay

An alkaline phosphatase test kit was used (Merck, Darm-stadt, Germany). The end point being the conversion of p-nitrophenol phosphate a yellow product. Cells were culturedas for the alamar Blue assay. After 48 h the media wasremoved, the cells rinsed with PBS and 100 lL ddH2Oadded to each well. The plate was placed at 2808C for20 min, then removed and thawed at 378C for 15 min. Thiswas repeated twice. The resulting cell suspension was thendiluted 50:50 with the p-nitrophenol dye in a pH 10.3 buffer,and the plate read colorimetricaly on an Anthos plate reader(Anthos Labtec, Salzburg, Austria) after 3 min using a 405nm measurement filter and a 620 nm reference filter.

Immunocytochemistry

The Ki-67 antibody detects an epitope expressed by thecells during all sections of the cell cycle except for Go;hence, this can be used as a marker for proliferation incell-biomaterial interactions.30 Cells were cultured for 48 hon triplicate samples prepared as described for the alamarBlue assay. After 48 h, the samples were removed andrinsed using PBS, then fixed using 100% dried acetone.This was removed, the samples rinsed with PBS, thenstained using a 1 in five dilution from stock of mouse anti-

TABLE IISputtering Coefficients Used for the Deposition ofHA by Capacitively Coupled RF Plasma Sputtering

Apparatus with Magnetron

Sputtering Coefficient HAP Sputtering

Substrate temperature (8C) 120 6 10Target substrate distance (cm) 5.0 6 0.4Target self bias (V) 2300 6 30Chamber pressure (mTorr) 3 6 0.5Sputtering time (h) 4Working gas Pure argonCoating layer thickness (lm) 1.2 6 0.3Power density (W/cm2) 5.0 6 0.1

EFFECT OF SURFACE MORPHOLOGY AND ROUGHNESS ON OSTEOBLAST-LIKE CELLS 639


human Ki-67 antibody (Dako). The samples were placedinto an incubator at 378C for 1 h. After incubation, theantibody was removed, the samples rinsed again withPBS, and then stained using a FITC-conjugated rabbit anti-mouse IgG for 1 h. The secondary antibody was removed,the cells rinsed for a final time and the cells mountedusing a PBS/glycerol mountant containing 1,4-diazobicyclo[2.2.2] octane and viewed using a fluorescence microscopeequipped. Ki-67 positive nuclei were counted using a mor-phometric counting technique. This involved dividing thearea of each disc into a grid using the graticule then count-ing and recording the number of positive nuclei in the firstsquare, omitting the second square, counting the numberof nuclei in the third square, omitting the fourth square,counting the fifth, and so on. Similarly every other columnwas also omitted. The mean number of positive cells ineach square was calculated.

DNA measurement

The DNA content of cell layers on surfaces was meas-ured using the binding of bis-Benzamide 33258 to DNA.31

A cell lysate was prepared by freeze-thaw as described forthe alkaline phosphatase assay, 50 lL of the lysate wasadded to an equal volume of bis-Benzamide 33258 (Sigma,UK). Fluorescence was measured at 360 nm excitationwavelength and 460 nm emission using a Cytofluor fluori-metric plate reader. Values were determined against astandard curve constructed using calf thymus DNA(Sigma).

Collagen staining

Immunocytochemistry was performed on samples forcollagen after 7 days. The media was removed from thesamples and the cells on the samples were washed inPhosphate Buffered Solution(PBS)/1% Bovine Serum Albu-min (BSA) and fixed with 4% paraformaldehyde and 2%sucrose solution in PBS, for 5 min at room temperature.The samples were washed again in PBS/BSA, and then thecells permeabilized for 5 min at 08C in buffered Triton x-100 and further washed two to three times in PBS/BSA.The samples were then rinsed with PBS with 1% BSA, andthe primary antibody for 1 h at 378C. The samples werethen rinsed in PBS/BSA as before and the secondary anti-body added for 30 min at 378C. The antibody used wasrabbit anti-mouse TRITC-conjugated IgG (Dako). Afterlabeling with the secondary antibody, the cells were rinsedagain with PBS/BSA, removed from the wells andmounted in glycerol/PBS mountant containing DABCO.All samples were performed in duplicate with a negativecontrol, which consisted of a sample which had the pri-mary antibody substituted with normal mouse serum. Pri-mary antibodies were obtained from Sigma.

Alamar blue assay

The samples were removed from the incubator after 48 hculture, and the media removed from the wells. Hundredmicroliter Alamar Blue dye (Serotec, UK) at a one in 10 dilu-

tion in Hanks’ Balanced Salt Solution (HBSS, Sigma) wasadded to each well, and the plate incubated for 1 h at 378C.After this time the dye was removed from the cells, placedin fresh wells, and fluorescence measured using a Cytofluorfluorimetric plate reader (Perceptive Biosciences) at an exci-tation wavelength of 560 nm and an emission wavelength of590 nm. This process was repeated at 7, 14, and 21 days.

Osteocalcin assay

Samples were prepared as previously described. Cellswere seeded onto the surfaces at a concentration of 5 3

105 cells/mL on samples in a 96-well plate, resulting in afinal seeding concentration of 5 3 104 cells per sample.Cells were maintained in standard tissue culture environ-ment as described above (378C, 5% CO2), with the mediaon the cells being changed every 48 h. Media sampleswere stored at 2208C before osteocalcin measurement.Osteocalcin measurements were performed using anELISA assay (Biosource, UK) as directed in the manufac-turers instructions.

All samples were used in triplicate, experiments wererepeated three times. Statistics were one or two-wayANOVA (as appropriate) performed using the PRISM sta-tistical package (Graphpad, USA), using n 2 1 degrees offreedom for all samples. Significant differences betweenthe means were assessed using the Tukey post-test.

RESULTS

Surface characterization

Surface roughness parameters

The profilometry measurements in Table III showthat the Mp surface has the lowest Ra value, with Ravalues rising through the SiC and Sp surfaces to thehighest on the, Gb surface. The Rt and Rz values aremeasurements of max peak-valley height and meanroughness depth, respectively. The SiC sample hasan Rt measurement of 5.24 and Rz value of 5.11,indicating a surface with features of regular depth asshown in Figure 2. The Sp surface has a similar Rtvalue to the SiC surface of 5.55 lm, but a lower Rzvalue, of 4.62 lm. This agrees with visual observa-tion from Figure 2 that the Sp and Gb surfaces havefeatures of varying depth. The Gb surface has an Rtof 13.15 and an Rz of 11.65 lm.

TABLE IIIRoughness Values of Coated Samples

HAP Coated SamplesRa(lm)

Rt(lm)

Rz(lm)

Micropolished <0.05 <1.01 �0.0150�300 lm glass beads shot peening 0.54 5.55 4.62100 lm SiC particles grit blasting 1.75 13.15 11.64240 SiC paper ground 0.49 5.24 5.11

640 BALL ET AL.


Surface profiles of the samples

X-axis plots using the SURFACOM were used todemonstrate surface topography. The results showthat the four samples have characteristic topogra-phies. The micropolished surface [Fig. 1(D)] is flat,with occasional small pits or defects caused by thepolishing or coating process. The shot peened sam-ple [Fig. 1(A)] is also flat, with occasional craters,caused by the impact of the glass beads. The gritblasted surface [Fig. 1(C)] presented a sharp randomprofile of pits and troughs, while the silicon carbidegrooved sample [Fig. 1(B)] had an apparent groovedtopography with a variety of depths having eitherrounded or v-shaped profiles.

Osteoblast responses

Cell morphology

After 48 h, cell coverage was visible on all of thesamples. The cells on the micropolished hydroxyapatitecoated titanium [Fig. 2(A)] appeared slightly elongatedwith numerous processes that appeared to form attach-ments with the surface. On the grit blasted and theshot peened surfaces, the cells were flattened overthe surface topography and were closely adhered to theunderlying surface [Fig. 2(B,C)]. The cells on the siliconcarbide sample appeared flattened, elongating alongthe axis of the larger and deeper grooves [Fig. 2(D)].

Shape factor measurements

The shape factor measurements indicated that thecells cultured on the silicon carbide sample showedthe lowest shape factor (shape factor 0.19, standarddeviation of 0.047), which indicated the cells on

these materials were the most elongated of the sam-ples. The cells cultured on the micropolished samplehad a similar shape factor (0.2027, sd 0.096). Osteo-blast-like cells cultured on both the shot peened andgrit blasted surfaces were more rounded than thoseon the Mp and SiC surfaces (p < 0.001) (Fig. 3).

Total DNA measurement

There was no significant difference in the DNAcontent of cell lysates from any of the surfaces (Fig.4). This showed that there were similar numbers ofcells on all surfaces after 48 h.

Alkaline phosphatase activity

The alkaline phosphatase activity, again expressedas a percentage of the control, was not significantlydifferent for any of the samples (Fig. 5).

Cell proliferation

The cell proliferation was assessed by Ki-67 stain-ing of the cells. The proliferative activity of cells, asindicated by the number of positively stained cellsper microscope field, cultured on silicon carbide andmicropolished samples were significantly greaterthan those cultured on the shot peened and gritblasted surfaces (p < 0.001) (Fig. 6).

Collagen staining

The matrix protein type I collagen was alsoimaged on the surfaces at 7 days [Fig. 7(A–D)]. Col-lagen appears present on all samples, with similarisolated distribution on the grit blasted and shot

Figure 1. 3D representations of the surfaces of the samples, generated by profilometry. A: Sample peened with 100–300lm glass beads (Sp). B: Sample ground with 240 grade silicon carbide paper (SiC). C: Grit blasted sample (Gb). D: Micro-polished sample (Mp). Z-scale is 1 mm 3 1 mm, scale bars show 1 lm.



peened samples, and more widely spread on the sili-con carbide and micropolished samples.

Long-term cell metabolic activity

The metabolic activity of the osteoblasts seeded onthe surfaces was measured using an alamar blueassay, and expressed as a fraction of the tissueculture plastic control. Cell activity on micropolishedand silicon carbide samples were both significantlyhigher after 2 days than that on the grit blasted sam-ple and the shot peened sample (p > 0.05, p > 0.001,respectively, measured using ANOVA), with the sili-con carbide sample being significantly higher thanthe shot peened sample (p > 0.01) (Fig. 8). For themicropolished sample, cell activity decreasedbetween 7 and 21 days while for Sp and Gb, activityincreased at 14 days, then decreased by day 21. Cellmetabolic activity on the SiC sample decreased after

2 days, and then decreased again after 14 days.There was no significant difference between cell met-abolic activities at 21 days.

Osteocalcin production

Detectable amounts of osteocalcin in the mediawere observed for all samples at 7 days showing allcells were maintaining an osteoblastic phenotype. Cellson all the hydroxyapatite coated samples showed sig-nificantly higher osteocalcin production than the cellscultured on tissue culture plastic (Fig. 9).

DISCUSSION

This study has demonstrated that osteoblast-likecell behavior is dependent on the physical nature of

Figure 2. Scanning electron micrographs of osteoblasts cultured on HA coated titanium alloy surfaces for 24 h. A:Micropolished surface (Mp): cells on these materials are spread with numerous cell processes which are attached tothe surface. B: Shot peened surface (Sp): numerous small, rounded surface scars are visible, caused by the impactof small glass beads during the machining process. Cells are attached but appear more compact than on the Mpsurface, with fewer long processes. C: Grit blasted surface (Gb): the surface is very irregular with features close tothe size of the osteoblast. Cells are again compact, similar in morphology to the Sp surface. D: Silicon carbideground surface (SiC): the grooved structure is immediately obvious. At 3 1000 magnification, the grooves appear tobe straight. The groove width is 2–6 lm. Cells have spread and elongated along these features parallel to thegroove axis.

642 BALL ET AL.


the surface that they are grown on. Potential differ-ences in chemical affects caused by the surfacemachining were removed by coating the surfaceswith a thin 1 lm coating of HA which followed thecontours of the surface. Furthermore, it is clear thatthe response shown by these cells is not simplyinfluenced by the magnitude of the surface features.

The alamar blue assay used in this study is a sim-ple, nonradioactive assay that monitors oxygen con-

sumption by virtue of the ability of the dye toreplace oxygen as the terminal electron receptor inthe electron transport chain.32 The initial cell activitymeasurements using both alamar blue and Ki-67staining showed that cell metabolic activity on theSp and Gb surfaces are lower than on the SiC andMp surfaces. No significant differences in DNA con-tent of the cells were noted, so the differences in ac-tivity were not due to different cell numbers. It isreasonable to conclude that the subsequent decreasein activity over time is due to the cells reaching con-fluence on the Mp and SiC samples, but not the Gbor Sp surfaces. For osteoblasts, cell metabolic activitydecreases over time as the cells increase their stateof differentiation.33 This is thought to be linked with

Figure 3. Shape factors for osteoblasts cultured on experi-mental surfaces for 24 h. A shape factor of 0 represents astraight line, while a shape factor of 1 indicates a circle.Values are means and standard deviations, (n 5 10). Mp,micropolished; Sp, shot peened; Gb, grit blasted; SiC, sili-con carbide ground. Shape factors of cells on the Sp andGb samples are significantly greater than SiC and Mp (p <0.001), indicating that cells on the Sp and Gb materials aremore rounded. [Color figure can be viewed in the onlineissue, which is available at www.interscience.wiley.com.]

Figure 4. Total DNA content of cells cultured on surfacesof different topography after 48 h. Values are mean andSEM of repeated experiments, n 5 5. Mp, micropolished;Sp, shot peened; Gb, grit blasted; SiC, silicon carbideground. No significant difference between samples found.[Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com.]

Figure 5. Alkaline phosphatase activity of osteoblasts onsurfaces of different topography after 48 h. Values aremean and SEM of repeated experiments, n 5 5. Mp, micro-polished; Sp, shot peened; Gb, grit blasted; SiC, silicon car-bide ground. No significant difference between samples.[Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com.]

Figure 6. Number of Ki-67 positive cells on surfaces ofdifferent topography after 48 h. Values are mean and SEM,n 5 30. Mp, micropolished; Sp, shot peened; Gb, gritblasted; SiC, silicon carbide ground. Mp and SiC higherthan Gb and Sp (p < 0.001). [Color figure can be viewed inthe online issue, which is available at www.interscience.wiley.com.]



cell withdrawal from the cell cycle during the G1

phase to differentiate during the quiescent phaseG0.

34 This pattern of a steady increase in activity to apeak, followed by a decrease is characteristic of dif-ferentiating osteoblasts in culture.35 Given thisassumption it may be inferred from the cell activitydata that the cells on the micropolished sample shiftto a more mature state around 7 days earlier thancells on the grit blasted sample. The presence of col-lagen I staining at 7 days on the Gb and Sp samplesindicates that those cells which have accumulated inthe discontinuities have begun to elaborate anextracellular matrix. At 14 days, the cells on the Spand Gb random surfaces are more active than on the

micropolished smooth and silicon carbide orderedsurfaces. It is likely that, given the lower prolifera-tion rate as indicated by Ki67, cells are still prolifer-ating on these surfaces having not reached conflu-ence. Further, although alkaline phosphatase produc-tion showed a similar pattern to the cell activity, nosignificant differences were noted. It may beinferred, therefore, that cells on the Gb and Sp surfa-ces proliferate slowly in the discontinuties, and elab-orate an ECM before continuing proliferation oncethey have filled these areas.

Morphologically, cells cultured on the Mp surfaceappear elongated with numerous filopodial exten-sions. These cells appear similar to those cultured on

Figure 7. Osteoblasts cultured on hydroxyapatite coated titanium samples of different topography for 7 days; then,stained using a primary antibody to human collagen I. Cell nuclei were stained with propidium iodide. A: Micropolishedsample: general cell coverage, little collagen staining, widely dispersed over the sample. B: Shot peened sample: cell cover-age more isolated, but collagen matrix evident between the cells (arrows). C: Grit blasted sample: good cell coverage, evi-dence of extensive collagen I staining (arrows). D: Silicon carbide ground sample. Few cells, isolated staining, but notextensive.

644 BALL ET AL.


the SiC surfaces, although those cells exhibited fewerfilopodia. Cells on the Gb and Sp surfaces weremore rounded than the cells on the SiC and Mp sur-faces, as reflected in their lower shape factor values.That these cells were less active was not altogethersurprising therefore, as cell adhesion and spreadingare required for entry to the cell cycle.36

The pattern of cell behavior on the grooved sur-face topography was comparable with that seen inother studies on surfaces with precisely machinedgrooves.37 while increased cell activity and alkalinephosphatase production have been noted on surfacesground using sandpaper.12 This is suggestive of factthat the underlying cell responses to machinedgrooves and ground surfaces are similar.

One possible explanation for the differences in ac-tivity on different surfaces may lie in the pattern ofattachment. Anselme and coworkers4,38 analyzed a se-ries of osteoblast interactions with titanium surfacesby relating the cell activity to the overall surface disor-der. They found that increased surface disorderresulted in less cell contact that resulted in decreasedcell activity. In this study, numerical analysis of cellshape has indicated that cells cultured on grooved(SiC) or flat (Mp) surfaces are more spread than thoseon shot-peened (Sp) or grit blasted (SiC), despite theSp and SiC having similar magnitudes of roughness(Ra). This indicates that cells show increased spread-ing on ordered or flat surfaces compared with those

with random discontinuity. The data presented here isa clear indication that the surface morphology playsan important role in influencing cell responses.

On recent work studying surface order, Riehle et alcultured cells on materials with random arrange-ments of nanometer scale features.39 Cells adhering tothese structures showed similar patterns of adhesionto random as to flat surfaces, after initially increasingadhesion. This apparent conflict between findingsmay be a consequence of the materials and the scaleexamined used. Both this study and those of Anselmeand coworkers used titanium as the template foraltering the surface, while the Riehle study used apolymeric substrate. The scale examined in this studywas an order of magnitude higher than that of Riehle.It is apparent therefore, that there may be an influ-ence of the material used and the scale of the featuresin determining the nature of the cell response.

The observed changes in cell responses are due tothe physical changes to the surface topography asthis study used a smooth PVD deposited hydroxyap-atite coating used to ameliorate the effect of anychemical changes caused by the surfaces modifyingof the titanium alloy. A further observation is thatHA supported greater osteocalcin production thantissue polystyrene. This is encouraging for the pros-pects of using this coating material for hard tissueimplants, as previous studies have shown little dif-ference between osteocalcin production on tissue cul-ture plastic and HA at early time points.5,40

A number of studies have indicated correlationswith osteoblast activity and the roughness of a sub-

Figure 8. Cell metabolic activity of osteoblasts, as a frac-tion of the tissue culture plastic control, measured up to 21days. Values are mean and SEM, n 5 5. Mp, micropol-ished; Sp, shot peened; Gb, grit blasted; SiC, silicon car-bide ground. Cell activity on Mp and SiC samples weresignificantly higher than that on Gb (p > 0.05, p > 0.001,respectively, at 2 days. SiC was higher than Sp (p > 0.01).Two-way ANOVA shows decrease in activity on themicropolished sample, between 7 and 21 days, while forSp and Gb, activity was similar at 7 and 14 days, but haddecreased by day 21. Activity on the SiC sample decreasedafter 2 days and 14 days.

Figure 9. Osteocalcin production by osteoblasts culturedthe experimental surfaces for 7 days. Control, tissue cul-ture polystyrene; Mp, micropolished; Sp, shot peened; Gb,grit blasted; SiC, silicon carbide ground. Values are meanand SEM, n 5 3. Cells on all the hydroxyapatite coatedsamples showed significantly higher osteocalcin produc-tion than the cells cultured on tissue culture plastic control(p < 0.01). [Color figure can be viewed in the online issue,which is available at www.interscience.wiley.com.]



strate.3,8,11–13,19–22 However, research is emergingthat the magnitude of the roughness alone might notbe that only influencing factor.4,14 These data indi-cate that osteoblasts cultured, in vitro, on surfaceswith a random discontinuity spread less and are lessmetabolically active than those on a more organizedsurface, or a flat surface. These influences persistover a 28 day period, although further study isrequired to assess the extent of mineralization of theosteoblasts. These data are a clear indication thatresearch in this area should not concentrate solelyon the magnitude of surfaces roughness, but alsoconsider that morphology of the surface will play arole in directing cell responses.

CONCLUSIONS

The methods used to modify the titanium the sur-faces produced consistent micron scale changes insurface roughness and topography. Further coatingwith a thin HA coating allowed the effects of thesurface roughness to be isolated from differences insurface chemistry. Osteoblasts responded to orderedor smooth surface topography by increasing cellmetabolic activity relative to the disordered surfaces.It is suggested that the cell responses to the surfaceswere influenced primarily by the topography of thesurfaces, and that cells on the ordered surfaces couldspread or elongate successfully whilst those on therough surfaces were constrained. The use of orderedmicron scale topographical surface modifications forbone contacting devices may prove to be a usefultarget for research to influence osseointegration ofbone-contacting materials.

The authors would like to acknowledge the help ofAction Research, who provided a grant to fund this work,and Professor Sandra Downes for helpful comments andideas that lead to this study.

References

1. Kieswetter K, Schwartz Z, Dean DD, Boyan BD. The role ofimplant surface characteristics in the healing of bone. CritRev Oral Biol Med 1996;7:329–345.

2. Curtis A, Wilkinson CDW. Topographical control of cells.Biomaterials 1997;18:1573–1583.

3. Deligianni D, Katsala N, Ladas S, Sotiropoulou D, Amedee J,Missrelis Y. Effect of surface roughness of the titanium alloyTi-6Al-4V on human bone marrow cell response and on pro-tein adsorption. Biomaterials 2001;22:1241–1251.

4. Anselme K, Bigerelle M, Noel B, Dufresne E, Judas D, Iost A,Hardouin P. Qualitative and quantitative study of humanosteoblast adhesion on materials with various surface rough-nesses. J Biomed Mater Res 2000;49:155–166.

5. Zreiqat H, Evans P, Howlett CR. Effect of surface chemicalmodification of bioceramic on phenotype of human bone-derived cells. J Biomed Mater Res 1999;44:389–396.

6. Zreiqat H, Howlett CR. Titanium substrata composition influ-ences osteoblastic phenotype: In vitro study. J Biomed MaterRes 1999;47:360–366.

7. Lauer G, Wiedmann-al-Ahmed M, Otten J, Huber U. The tita-nium surface texture effects adherence and growth of humangingival keratinocytes and human maxillar osteoblast-likecells in vitro. Biomaterials 2001;22:2799–2809.

8. de Santis D, Guerriero C, Nocini P, Ungersbock A, RichardsG, Gotte P, Armato U. Adult human bone cells from jawbones cultured on plasma-sprayed or polished surfaces of ti-tanium or hydroxylapatite disks. J Mater Sci Mater Med1996;7:21–28.

9. Ungersbock A, Rahn B. Methods to characterise the surfaceroughness of metallic implants. J Mater Sci Mater Med1995;5:434–440.

10. Ranucci CS, Moghe PV. Substrate microtopography canenhance cell adhesive and migratory responsiveness to ma-trix ligand density. J Biomed Mater Res 2001;54:149–161.

11. Martin JY, Schwartz Z, Hummert TW, Schraub DM, SimpsonJ, Lankford J, Dean DD, Cochran DL, Boyan BD. Effect of ti-tanium surface-roughness on proliferation. Differentiationand protein-synthesis of human osteoblast-like cells (MG63).J Biomed Mater Res 1995;29:389–401.

12. Hatano K, Inoue H, Kojo T, Matsunaga T, Tsujisawa T,Uchiyama C, Uchida Y. Effect of surface roughness on prolif-eration and alkaline phosphatase expression of rat calvarialcells cultured on polystyrene. Bone 1999;25:439–445.

13. Boyan BD, Hummert TW, Dean DD, Schwartz Z. Role of ma-terial surfaces in regulating bone and cartilage cell response.Biomaterials 1996;17:137–146.

14. Anselme K, Bigerelle M, Noel B, Iost A, Hardouin P. Effect ofgrooved titanium substratum on human osteoblastic cellgrowth. J Biomed Mater Res 2002;60:529–540.

15. Chesmel KD, Clark CC, Brighton CT, Black J. Cellular-responses to chemical and morphologic aspects of biomaterialsurfaces. II. The biosynthetic and migratory response of bonecell-populations. J Biomed Mater Res 1995;29:1101–1110.

16. Gray C, Boyde A, Jones SJ. Topographically induced boneformation in vitro: Implications for bone implants and bonegrafts. Bone 1996;18:115–123.

17. Chehroudi B, Ratkay J, Brunette DM. The role of implant sur-face geometry on mineralization in vivo and in vitro—Atransmission and scanning electron-microscopic study. CellsMater 1992;2:89–104.

18. Chehroudi B, McDonnell D, Brunette DM. The effects ofmicromachined surfaces on formation of bonelike tissue onsubcutaneous implants as assessed by radiography and com-puter image processing. J Biomed Mater Res 1997;34:279–290.

19. Boyan BD, Batzer R, Kieswetter K, Liu Y, Cochran DL,Szmuckler-Moncler S, Dean DD, Schwartz Z. Titanium sur-face roughness alters responsiveness of MG63 osteoblast-likecells to 1 a,25-(OH)(2)D-3. J Biomed Mater Res 1998;39:77–85.

20. Batzer R, Liu Y, Cochran DL, Szmuckler-Moncler S, Dean

DD, Boyan BD, Schwartz Z. Prostaglandins mediate the

effects of titanium surface roughness on MG63 osteoblast-like

cells and alter cell responsiveness to 1 a,25-(OH)(2)D-3.

J Biomed Mater Res 1998;41:489–496.

21. Lohmann CH, Sagun R, Sylvia VL, Cochran DL, Dean DD,

Boyan BD, Schwartz Z. Surface roughness modulates the

response of MG63 osteoblast-like cells to 1,25-(OH)(2)D-3

through regulation of phospholipase A(2) activity and activa-

tion of protein kinase A. J Biomed Mater Res 1999;47:139–151.

22. Boyan BD, Lohmann CH, Sisk M, Liu Y, Sylvia VL, Cochran

DL, Dean DD, Schwartz Z. Both cyclooxygenase-1 and cyclo-

oxygenase-2 mediate osteoblast response to titanium surface

roughness. J Biomed Mater Res 2001;55:350–359.23. Schwartz Z, Lohmann CH, Vocke AK, Sylvia VL, Cochran

DL, Dean DD, Boyan BD. Osteoblast response to titanium

646 BALL ET AL.


surface roughness and 1 a,25-(OH)(2)D-3 is mediatedthrough the mitogen-activated protein kinase (MAPK) path-way. J Biomed Mater Res 2001;56:417–426.

24. Lampin M, WarocquierClerout R, Legris C, Degrange M,SigotLuizard MF. Correlation between substratum roughnessand wettability, cell adhesion, and cell migration. J BiomedMater Res 1997;36:99–108.

25. Anselme K, Linez P, Bigerelle M, Le Maguer D, Le Maguer A,Hardouin P, Hildebrand HF, Iost A, JM L. The relative influ-ence of the topography and chemistry of TiAl6V4 surfaces onosteoblastic cell behaviour. Biomaterials 2000;21:1567–1577.

26. Pypen C, Plenk H, Ebel MF, Svagera R, Wernisch J. Charac-terization of microblasted and reactive ion etched surfaces onthe commercially pure metals niobium, tantalum and tita-nium. J Mater Sci Mater Med 1997;8:781–784.

27. Lo WJ, Grant DM. Hydroxyapatite thin films deposited ontouncoated and (Ti,Al,V)N-coated Ti alloys. J Biomed MaterRes 1999;46:408–417.

28. Grant DM, Lo WJ, Parker KG, Parker TL. Biocompatible andmechanical properties of low temperature deposited quater-nary (Ti,Al,V)N coatings on Ti6Al4V titanium alloy sub-strates. J Mater Sci Mater Med 1996;7:579–584.

29. Ball MD, Downes S, Scotchford CA, Antonov EN, Bagratash-vili VN, Popov VK, Lo W-J, Grant DM, Howdle SM. Osteo-blast growth on titanium foils coated with hydroxyapatite bypulsed laser ablation. Biomaterials 2001;22:337–347.

30. VanKooten TG, Klein CL, Kohler H, Kirkpatrick CJ, WilliamsDF, Eloy R. From cytotoxicity to biocompatibility testing invitro: Cell adhesion molecule expression defines a new set ofparameters. J Mater Sci Mater Med 1997;8:835–841.

31. Rago R, Mitchen J, Wilding G. DNA Fluorometric assay in 96-well tissue-culture plates using Hoechst-33258 after cell-lysis byfreezing in distilled water. Anal Biochem 1990;191:31–34.

32. Ahmed SA, Gogal RM, Walsh JE. A new rapid and simplenonradioactive assay to monitor and determine the pro-liferation of lymphocytes—An alternative to H-3 thymi-dine incorporation assay. J Immunol Methods 1994;170:211–224.

33. Basle MF, Rebel A, Grizon F, Daculsi G, Passuti N, Filmon R.Cellular-response to calcium-phosphate ceramics implantedin rabbit bone. J Mater Sci Mater Med 1993;4:273–280.

34. Lodish H, Berk A, Zipursky S, Matsudaira P, Baltimore D,Darnell J. Molecular Cell Biology. New York: W.H. Freeman;2000.

35. Aubin JE, Turkson K, Heersche JNM. The osteoblastic cell lin-eage. In: Noda M, editor. Cellular and Molecular Biology ofBone. San Diego: Academic Press; 1993. p 1–29.

36. Assoian RK. Anchorage-dependent cell cycle progression.J Cell Biol 1997;136:1–4.

37. Perizzolo D, Lacefield WR, Brunette DM. Interactionbetween topography and coating in the formation of bonenodules in culture for hydroxyapatite- and titanium-coatedmicromachined surfaces. J Biomed Mater Res 2001;56:494–503.

38. Bigerelle M, Anselme K, Noel B, Ruderman I, Hardouin P,Iost A. Improvement in the morphology of Ti-based surfaces:A new process to increase in vitro human osteoblastresponse. Biomaterials 2002;23:1563–1577.

39. Riehle MO, Dalby MJ, Johnstone H, MacIntosh A, Affross-

man S. Cell behaviour of rat calvaria bone cells on surfaces

with random nanometric features. Mater Sci Eng C

2003;23:337–340.40. Lopes MA, Knowles JC, Santos JD, Monteiro FJ, Olsen I.

Direct and indirect effects of P2O5 glass reinforced-hydroxy-apatite composites on the growth and function of osteoblast-like cells. Biomaterials 2000;21:1165–1172.



The effect of different surface morphology and roughness on osteoblast-like cells

Documents

Transcript of The effect of different surface morphology and roughness on osteoblast-like cells