An accelerated buoyancy adhesion assay combined with 3-D morphometric analysis for assessing osteoblast adhesion on microgrooved substrata

An accelerated negative buoyancy method has been developed to assess cell adhesion strength. This method has been used in conjunction with 3-D morphometric analysis to understand the effects of surface topology on cell response. Aligned micro-grooved surface topographies (with a range of groove depths) were produced on stainless steel 316L substrates by laser ablation. An investigation was carried out on the effect of the microgrooved surface topography on cell adhesion strength, cell and nucleus volumes, cell phenotypic expression and attachment patterns. Increased hydrophobicity and anisotropic wettability was observed on surfaces with deeper grooves. A reduction was noted in cell volume, projected areas and adhesion sites for deeper grooves, linked to lower cell proliferation and differentiation rates and also to reduced adhesion strength. The results suggest that the centrifugation assay combined with three-dimensional cell morphometric analysis has considerable potential for obtaining improved understanding of the cell/ substrate interface. & 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC


Introduction
Over recent decades, many studies have focused on optimising the interface between implants and surrounding tissue. Failure of prostheses, particularly loosening of the bond with surrounding tissue, is a major concern (McGee et al., 2000). Successful implantation depends critically on how the bond is formed during the initial period of contact. Several methods have been developed to measure cell adhesion strength on grooved surfacessee review Garcia and Gallant (2003). The tendency towards increased adhesion strength on microgrooved substrates was corroborated by enzymatic detachment studies Chen et al., 2007;Ismail et al., 2007;Chen et al., 2009). Although these techniques are capable of providing some useful information on cell adhesion, they yield no quantitative information. Furthermore, enzyme activity can be influenced by the nature of the solid/ liquid interface (Koutsopoulos et al., 2007), which in turn can be affected by topographic effects.
To quantify cell-substrate adhesion strength, techniques that can provide well-defined mechanical-based parameters are needed. Centrifugation is a simple, reproducible technique that provides reliable quantitative measurements of cell adhesion (McClay et al., 1981;Reyes and García, 2003) by detaching a population of adherent cells through the application of normal or shear stresses. However, the centrifugation assay applies relatively low detachment forces and has therefore been limited to short adhesion times, typically less than 60 min, as at longer adhesion times, cellular attachment strength exceeds the maximum centrifugal force and the assay losses sensitivity (Gallant, 2013).
It is not clear whether cell proliferation and differentiation is intrinsically linked to cell and nucleus morphological characteristics on grooved surfaces. Arguably, cells need to be able to spread in order to enter phase S of the cell cycle during proliferation (Alberts et al., 2002) and if sufficient spreading is not achieved, their proliferation may be inhibited. The addition of topographical features, such as grooves, brings additional physical challenges for cells, probably impeding sufficient spreading essential to achieve normal or improved proliferation rates. Nevertheless, various cell types, such as mesenchymal stem cells (Gerecht et al., 2007), osteoblasts (Kenar et al., 2006;Erdoğan et al., 2011), fibroblasts (Wang et al., 2010) and epithelial cells (den Braber et al., 1996;Khang et al., 2008) have shown to increase their proliferation rates on micro-grooved surfaces. Similarly, regarding osteogenic differentiation, some studies reported increased osteogenic differentiation markers on grooved surfaces (Mata et al., 2009;Lee et al., 2010), while others suggest no significant effects or reduction of cell differentiation (Bruinink and Wintermantel, 2001;Kenar et al., 2006;Holthaus et al., 2012). While the wide range of materials, cell types and experimental conditions again makes it difficult to draw clear universal conclusions, these studies do suggest that the presence of grooves can affect cell morphology, gene expression, proliferation and the strength of adhesion.
Several techniques have been used previously to create well-defined surface topographic features. These include photolithography, electron beam lithography, interference lithography, soft lithography, casting, grit blasting etc. Although such techniques can create well-defined (nanoand micro-scale) topographic features, most of them do suffer from certain limitations and in some cases they are not well-suited to the range of materials commonly used for implantsi.e. relatively hard metals and ceramics. The use of lasers, on the other hand, introduces considerable versatility, since there is a lot of scope for their accurate control and they can melt/ablate virtually any material. In fact, lasers have already been used for surface topography modification in implants and other biomedical devices (e.g. Duncan et al., 2002;Lawrence et al., 2006;Wang et al., 2008;Qin et al., 2014;Zheng et al., 2015).
In this study, an accelerated-buoyancy adhesion test (up to 112,000 g) in conjunction with 3D morphometric analysis was used on micro-grooved surfaces to measure cell adhesion strength and study topology effects on cell attachment/ morphology. Aligned micro-grooved surface topographies with a range of groove depths and fixed width were produced on stainless steel 316L substrates by laser ablation. The use of lasers introduces considerable versatility, since there is a lot of scope for their accurate control and they can melt/ablate virtually any material. Unlike most previous studies, cell adhesion experiments were performed 3 days after seeding, allowing for the cell/substrate contact areas to increase.

Material and sample preparation
Mirror-finish 316L (austenitic) stainless steel sheets, supplied by Aalco (Roundtree Way, Mousehold Lane, Norwich, NR7 8SR, UK), were used for this study. The sheets were 0.9 mm in thickness. Circular discs of 11 mm diameter were cut from these sheets, using a water-jet cutter.

Laser ablation treatment
Samples were surface-treated using a pulsed Ytterbium (Yb) fibre laser, G3 SP-20P-0011, supplied by SPI lasers. Details of the laser system are described elsewhere (O'Neill and Kun, 2009). The laser processing conditions employed are listed in Table 1. Sets of parallel grooves were produced on disc surfaces, as illustrated in Fig. 1. They were designed to generate both relatively coarse undulations and finer scale topographic features. In order to minimise oxidation, laser ablation was carried out using argon as a shrouding atmosphere. Discs polished down to a 0.3 mm finish served as control surfaces. For cell culture studies and material characterisation, the surfaces were cleaned ultrasonically for 5 min, sequentially, in acetone, ethanol, and isopropanol, and then left to dry overnight submerged in ethanol, inside a flow cabinet. Between the acetone and ethanol washing steps, the surfaces were wiped with an ultra-soft cloth (Metprep, UK) soaked in 100% ethanol. After this, the samples were sterilised at 160 1C for 2 h.

2.3.
Topographic and chemical characterisation 2.3.1. Atomic force microscopy (AFM) Topographic features were studied using a Digital Instruments Nanoscope (Veeco Instruments, UK) atomic force microscope. The laser-ablated surfaces were investigated in tapping mode, with silicon nitride tips (Veeco Instruments, UK). Scanning areas were selected according to the type of analysis: (a) 5 Â 5 mm 2 for average roughness (R a ) measurements and (b) 100 Â 100 mm 2 for 3D renderings. Scanning was carried out at 451 to the groove long axis, in order to reduce errors associated with the sharp borders of the grooves. The frequency was slightly adapted, depending on the scanning area and type of surface analysed, but in most cases it was between 0.3 and 0.5 Hz. Two samples of each type were analysed (n ¼2). The cross-sections were characterised in terms of groove depth (d g ) and width (w g ). The software packages employed were Digital Instruments v.7.30 (Veeco Instruments, UK) and Scanning Probe Imaging Processor (SPIP) (Image Metrology A/S, Denmark).

Interferometric optical profilometry
The surface topography was also investigated using a Veeco Wyko NT3300 Interferometer. A 20 Â objective lens was used.
Three scans, with a 200 Â 200 mm 2 scan area, were taken randomly from each sample. Four samples of each type were analysed (n¼ 4). Surface maps and R a measurements were obtained using the Veeco Vision 3.43 and the SPIP imaging softwares (Image Metrology A/S, Denmark).

Contact angle (CA) measurement
Static contact angles were measured using Millipore water (R¼ 18 MΩ) with a KSV CAM 200 goniometer (KSV Instruments), using the sessile drop method. A drop was released onto the surface and imaged from the side using a planar CCD camera. Measurements were performed at room temperature, using 1 mL droplets. Photographs of the water drop were taken, after stabilisation (for 5 s), parallel and transverse to the grooves. Six samples were studied for each type (n¼ 6) and three measurements were carried out on each sample. The results were analysed using the KSV Instruments software.

Accelerated buoyancy adhesion characterisation test
To quantify the adhesive strength of cells to substrates, buoyancy forces were generated under enhanced acceleration, in a swing-bucket ultra-centrifuge. The procedure is described in detail elsewhere (Robinson et al., 2011). In this study, in order to ensure similar seeding efficiencies between samples, customised holders were designed to hold the specimens inside 24-well plates, leaving only a circular area of 10 mm in diameter available for the cells to attachsee Fig. 2. A total of about 4000 cells were seeded to each sample (5093 cells/cm 2 ). This relatively low cell density was chosen so that individual cells were not in contact with their neighbours and showed little or no expression of extracellular matrix. Forces acted normal to the plane of the substrate. The rotational speed was 25,000 rpm and the radial distance of the substrates was approximately 160 mm, generating accelerations of around 112,000 g. The samples were centrifuged for a fixed period of 7 min. Six samples (n¼ 6) were used for each specimen type and the tests were performed 3 days after seeding. The proportion of cells remaining adhered to the substrates was estimated by calculating the number of cells before and after centrifuging using the CyQUANT assay (Section 2.4.5).

Scanning electron microscopy (SEM)
Cell morphology was studied using a JEOL 5800LV, with an accelerating voltage of 10 kV and a working distance of about 10 mm. Cell-seeded samples were rinsed twice with warm Table 1 -Processing conditions, average groove width and depth and roughness data, obtained using both optical interferometry and atomic force microscopy (AFM) techniques. LT surfaces were laser-treated with an average power of 10 W and a pulse frequency of 125 kHz. No AFM roughness values are shown for LT-22, since the depth was out of the z-range of the equipment. Nomenclature: "LT-d g ", where LT stands for laser treated and d g is the groove depth.  buffer solution and placed in a cold 24-well culture plate, with 500 mL of primary fixative (2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4)). After 1 h, the fixative was removed and the samples were washed twice in PBS. The cells were then dehydrated, by soaking in a series of ethanol solutions (30%, 50%, 70%, 90% and 100%) for 20 min each. Finally, the samples were soaked for 5 min in hexamethydisilazane (HMDS), before allowing air-drying overnight. Once dry, the discs were attached to SEM stubs with doublesided carbon tape and then sputter-coated with Pt for 2 min at 40 keV.

Confocal laser scanning microscopy (CLSM)
After 3 days in culture, the cells were fixed in 4% paraformaldehyde/phosphate buffer solution (PBS), with 1% sucrose, for 15 min at room temperature. The cell-seeded samples were then washed with PBS and permeabilised in 0.15% Tritons X-100 (Sigma, UK) in PBS for 15 min at room temperature. Samples were then subjected to the following immuno-labelling protocols.
2.4.4.1. 2D and 3D morphometric analysis. Samples were incubated in the dark in FITC conjugated phalloidin (P5282, Sigma, UK) at 5 mg mL À 1 (1:100 in 1% BSA/PBS, Sigma, UK) for 40 min at 37 1C. They were then washed and incubated in the dark with 2 mg/mL HCS CellMask red cytoplasmatic/nuclear stain (H32711, Invitrogen, UK) in PBS for 30 min at room temperature. Samples were then washed twice with PBS before mounting on the Vectorshield fluorescent mountant containing DAPI for nuclear labelling (Vector Laboratories, UK). Approximately 100-150 individual cells per type were analysed using the Volocity 5.5.1 software (PerkinElmer, USA). 2D and 3D image z-stacks were acquired with a Nikon 20 Â / 0.75 N.A. Plan Fluor Multi-Immersion lens. The channels for DAPI, FTIC conjugated phalloidin and HCS were merged using a brightest point algorithm. All cells and nuclei were automatically identified, thresholded and reconstructed in the merged channel using Volocity standard measurement protocols. The area and orientation of cells and nuclei were obtained using CLSM imagessee Fig. A.1 (Appendix). A box was fitted to each cell/nucleus using the Volocity software and the diagonal overlapping the cell/nucleus was traced. The cell orientation was obtained from the angle between the traced diagonal and the groove long axis (laser-patterned samples) or a random line (polished surfaces). The nucleus elongation factor was the ratio of the shortest axis of the

Focal adhesion analysis. Samples were incubated in
1% BSA in PBS for 30 min at room temperature, to block unspecific binding sites of antibodies. They were then incubated at room temperature (in the dark) for 1.5 h in 1:150 primary antibody monoclonal Anti-Vinculin antibody in 1% BSA/PBS (V9131, Sigma, UK). Anti-vinculin labelling (orange) was used to detect the adhesions. Once this step was complete, the samples were rinsed three times with PBS containing 0.05% of Tweens 20 (P2287, Sigma, UK) and then incubated in 1:200 AlexaFluor 594 goat anti-mouse secondary antibody in 1% BSA/PBS (A11005, Invitrogen, UK) for 1 h at room temperature (in the dark) . After washing the samples three times with PBS containing 0.05% of Tweens 20, they were incubated in the dark in FITC conjugated phalloidin (P5282, Sigma, UK) at 5 mg mL À 1 (1:100 in 1% BSA/PBS, Sigma, UK) for 40 min at 37 1C. Samples were washed twice with PBS, before mounting on the Vectorshield fluorescent mountant containing DAPI for nuclear labelling. The Volocity software was used to automatically find and analyse adhesion sites. 3D image z-stacks were acquired using a Nikon 60 Â /1.3 N.A. Plan Apo VC Oil Immersion lens. The adhesion frequency was calculated as the mean number of adhesions per cell. The criteria for adhesion classification were taken from Bershadsky et al. (2006), where adhesions in the range 0.5-2 mm are designated as focal complexes (FXs), adhesions between 2 and 5 mm as focal adhesions (FAs) and adhesions larger than 5 mm as super-mature adhesions (SMAs). This analysis was combined with the FTIC phalloidin and DAPI channels to qualitatively describe the relatively position of the adhesion sites according to the morphology of the cell, configuration of the actin filaments and surface topography. About 10-15 cells were tested for each type of surface topography. Samples were imaged using a Nikon C1 scanning confocal system, with an upright microscope (Nikon Instruments Europe). A UV diode laser, argon ion laser and a red diode laser were used to provide excitation for the chosen fluorophores at wavelengths of 405, 488 and 633 nm respectively. The emission signal produced by each fluorophore upon excitation was collected on separate photo-multiplier tube detectors. To prevent spectral cross-talk between emission signals, a sequential scan mode was employed. All images were acquired using an XY scan resolution of 1024 Â 1024 pixels, which provided scan areas of 1.24 and 0.414 mm 2 , whilst using the 20 Â and 60 Â objective lenses respectively. The smallest pinhole setting allowable on the system was used for scanning through the sample in Z. A suitable Z-step size was employed to satisfy the Nyquist sampling theorem for both objective lenses.

Cell proliferation
Cell proliferation was assessed using the CyQUANT Cell Proliferation Assay (Invitrogen, UK). At each time point (1, 4, 7, 10, 15 and 21 days), the medium was aspirated and six samples (n¼ 6) from each type were frozen at À80 1C until required for measurement. After thawing, a CyQUANT lysis buffer was added to each sample and the solution was agitated. A volume of 50 mL of the cell lysate was transferred to a black 96-well plate and 50 mL of the CyQUANT dye was added. DNA content was measured fluorimetrically (excitation 480 nm, emission 520 nm), using a BMG Labtech FLUOstar Optima plate reader. Fluorescence readings were converted to cell numbers through a standard curve. For this, a range of cell concentrations were seeded onto 24-well plates and prepared in the same manner as the samples. Cells were seeded at a density of 10,000 cells per sample ( $12,732 cells cm −2 ). To prevent cell growth, the cells were frozen 24 h after seeding.

Cell differentiation
Alkaline phosphatase (ALP) enzymatic activity was used as a marker for early osteoblastic differentiation, using 6,8difluoro-4-methylumbelliferyl phosphate (DiFMUP) as a fluorogenic substrate. The reaction of ALP with DiFMUP can be followed at 358 nm, as 6,8-difluoro-7-hydroxy-4-methylcoumarin (DiFMU) is produced. ALP activity was assessed fluorimetrically, following the mixing 50 mL of the cell lysate prepared for the CyQUANT assay with 50 mL ALP assay buffer containing DiFMUP and incubating at 37 1C for 15 min in the absence of light. The reaction was measured twice in each sample. The releasing fluorescent DifMU was measured using a BMG Labtech FLUOstar Optima plate reader at 358 nm excitation and 455 nm emission. The signal was measured in fluorescence emission intensity units and converted to quantity of DiFMU produced, using a DifMU standard curve. This experiment was performed on the same samples used for the proliferation analysis (n¼ 6).

Statistical analysis
For cell adhesion experiments, averages of two independent studies, with six replicates per surface, were expressed as the arithmetic mean7SD. For cell proliferation and differentiation, three independent experiments, with six replicates per surface, were conducted. Results are presented and expressed as the arithmetic7standard deviation (SD). Data were analysed using Tukey analysis of variance (ANOVA), with Graphpad Prism software (Graphpad Software Inc., USA). Differences were considered statistically significant at p values of o0.05.

Surface characterisation
3.1. Surface topography 3D surface maps and section profiles, obtained using optical interferometry, are shown in Fig. 1. Section profiles were taken parallel (blue lines) and perpendicular (green lines) to the groove direction. Table 1 lists the laser processing conditions employed to produce micro-grooved surfaces, the groove dimensions (depth (d g ) and width (w g )) and the average surface roughness, R a . While the polished surface is featureless and flat, the laser treated (LT) surfaces exhibit parallel grooves, containing a series of sub-micron "lips" ( $1.1 mmsee Fig. 1) within. These "lips" are produced by surface melting and refreezing phenomena associated with individual laser pulses (Kurella and Dahotre, 2005). The spacing between these lips, and also the groove depth, are determined by the scanning speed and the number of irradiation steps (i.e. the number of repeated scans in the same area). All patterns have the same groove width ( $40 mm), since the laser beam spot size was fixed, but varied in groove depth (Table 1). The LT surfaces are designated, on the basis of their groove depth, "3.5 μm", "7 μm" and "22 μm". R a values (Table 1), taken at 451 to the groove long axis, increase from a small fraction of a micron ( $50 nmpolished surface) to about 5 μm (LT-22 surface). This increase is attributed to the increasing groove depth, whereas the section profiles (blue lines of Fig. 1) show that the roughness of the bottom of the grooves is very similar for all LT surfaces.

3.2.
Wetting characteristics Fig. 3 shows the water drop contact angles measured perpendicular (θ ? ) and parallel (θ // ) to the axis of the grooves. Also shown are images of the 1 mL water droplets, taken along both directions. It can be seen that the contact angles measured parallel (θ x ) to the groove axis increase with groove depth from $ 801 (slightly hydrophilic) on polished surfaces to $1341 (strongly hydrophobic) on LT-22 surfaces. These angles were higher than the corresponding ones measured perpendicular to the groove axis, suggesting a wetting anisotropy for the laser-treated surfaces.

Cell adhesion
The results from the accelerated buoyancy adhesion characterisation test are shown in Fig. 4, as cell removal fractions after centrifuging. It can be seen that the proportion of removed cells is significantly higher on the laser-treated samples with deeper grooves (LT-7 and LT-22) compared to polished or shallow-grooved samples (LT-3.5). No significant differences were detected between the LT-7 and LT-22 surfaces, and also between the polished and LT-3.5 surfaces.

4.2.
Cell morphology Fig. 5 shows cell morphologies after 4 days in culture, as obtained by SEM and CLSM. It can be seen that, on the polished surfaces, cells exhibit a polygonal morphology typical of an osteoblast in culture. On the laser-treated surfaces, some differences in the attachment pattern were observed between different groove depths. On the LT-3.5 surface, the cell attachment pattern was similar to that observed on the polished surfaces. Cells commonly straddled two or more parallel grooves, residing inside the grooves. On the LT-7 and LT-22 surfaces, on the other hand, the majority of cells were elongated in the direction of the groove axis. Cells were bridging the grooves and on the LT-22 surfaces often did not conform to the concave shape of the underlying grooves.

2D and 3D morphometric analysis
Cell areas and orientation angles are presented in Fig. 6 (a) and (b) respectively. It can be seen that the LT-7 and LT-22 surfaces exhibit lower projected areas compared to polished and LT-3.5 surfaces ( Fig. 6(a)). Cell orientation measurements ( Fig. 6(b)) were categorized into 9 bins of 101. On polished surfaces, no preferred orientation was observed. On LT-3.5 surfaces, a very small proportion of cells exhibit intermediate orientation angles (301-601) with respect to the groove long axis. The majority of cells are oriented either perpendicular (701-901) or parallel (01-301) to the groove direction. On LT-7 surfaces, most hFOBs ( E40%) were oriented between 01 and 401, with the largest proportion of cells found in the 01-101 group. On sample LT-22 surfaces, the vast majority of hFOBs ( E30%) were found in the 01-101 group. Nucleus areas and elongation factors are shown in Fig. 6 (c) and (d) respectively. It can be seen that the nucleus area becomes statistically smaller with increasing groove depth. On both LT-7 and LT-22 surfaces, the nucleus is more elongated compared to polished and LT-3.5 surfaces. Fig. 7 shows typical 3D images of individual hFOBs attached onto the different surfaces. It can be seen that LT-7 surfaces seem to promote a highly elongated morphology, compared with the other two laser-ablated surfaces. The results from the volumetric analysis of cells and nuclei are shown in Fig. 8. No statistically significant differences (p40.05) were observed between polished and LT-3.5 surfaces. However, significantly lower cell and nuclei volumes were measured on the LT-22 surfaces, compared with polished and LT-3.5 surfaces.

Focal adhesions
Quantification of the focal adhesion sites for each surface topography, and type of adhesion (FX, FA or SMA), is presented in Fig. 9(a). It can be seen that the total number of focal adhesion sites, and also the number of focal adhesion sites per type, were lower on the laser-ablated surfaces, compared to polished surfaces. Focal complexes (FXs) and focal adhesions (FAs) were the mostly prominent adhesions. Fig. 9(b) shows correlations between the number of adhesion sites per cell and groove depth.  Fig. 9(a), to promote readability. However, all time points and conditions where standard deviations are not overlapping can be considered statistically significant (po0.01). It can be seen that polished surfaces exhibit higher cell proliferation rates than laser-ablated surfaces until day 9. From day 9 onwards, on all surfaces, except LT-22, a decrease in proliferation rate was observed as the surfaces started to reach confluence. After day 14, a plateau was reached. Cell numbers were lower on grooved surfaces compared to polished, with numbers decreasing with groove depth until about day 12, after which no effect of groove depth was observed. The ALP activity of the osteoblasts cultured on the different surfaces ( Fig. 10(b)) shows that statistically significant differences were only observed from day 9 onwards, with polished surfaces showing higher ALP activity when compared to grooved surfaces.

Discussion
In this study, an accelerated-buoyancy adhesion test is used, in conjunction with 3D morphometric analysis, to assess cell adhesion strength and topology effects on cell attachment and morphology. A laser ablation technique was employed to generate micro-grooves on austenitic stainless steel 316L surfaces. To ensure that the effect of surface topography is truly distinguished from that of surface chemistry, XPS analysis was carried out (Fig. A.3 -Appendix). No differences in surface chemistry between laser-ablated and polished surfaces were detected suggesting that any differences in cell response are likely to be attributed to surface topography orthogonal to the groove decreased with groove depth, whereas along the grooves, spreading was somewhat similar. This suggests that, at high groove depths, the fluid cannot occupy all the available surface area resulting in an increase in the contact angle (higher hydrophobicity). A similar anisotropic wetting behaviour, associated with groove depths, has been reported previously (Zhao et al., 2007) which was attributed to an increasing energy barrier. The wetting state of the water droplet is mainly Wenzel, in which the droplet reaches the bottom surface of the groovessee e.g. Belaud et al. (2015) for the different wetting regimes. On the LT-22 surface though, the surface is hydrophobic (θ // $ 1341) and the drop resides mostly on the top of the surface grooves (i.e. the drop does not advance much down on the vertical walls of the groove) suggesting that the wetting mode is Cassie-Baxter.
No significant differences in cell morphology were detected between polished surfaces and surfaces containing shallow grooves (LT-3.5). On surfaces with deeper grooves (LT-7 and LT-22), on the other hand, a contact guidance effect was observed. The suggestion that deeper grooves (up to 25 μm) can be more effective in guiding cell orientation is in agreement with previous studies (Clark et al., 1990;Chehroudi et al., 1992;Curtis and Wilkinson, 1997). Fig. 6 (a) shows that the projected cell areas on these surfaces are lower than those on polished and LT-3.5 surfaces, which are more hydrophilic, in agreement with a previous study (Liu et al., 2007). This observation corroborates previous findings  ( Anselme and Bigerelle, 2005) in which human osteoblasts were found to spread more on surfaces with low roughness than on rough ones. Cell orientation measurements depicted in Fig. 6(b) show that on polished and LT-3.5 surfaces there is no preferred orientation, whereas on the LT-7 and LT-22 surfaces, cells tend to align along the long axis of the groove, which is in agreement with SEM and CLSM observations. Nuclei exhibited lower projected areas and higher elongation factors on deeper micro-grooved surfaces as shown in Fig. 6 (c) and (d) respectively. Cell and nucleus elongation with increasing groove depth has been reported previously on nano-scale grooves (Davidson et al., 2010(Davidson et al., , 2015. SEM observations suggest that cells tend to reside within the grooves on LT-3.5 mm surfaces, but on deeper grooves , cells often attached to the groove rims and spread from one rim to another, bridging the groove. Bridging is more likely to occur on surfaces with deeper grooves. Osteoblast bridging onto micro-grooved surfaces has been reported previously (Jayaraman et al., 2004;Cao et al., 2009).
Immunolabeling of focal adhesions suggests that fewer adhesion sites per cell were observed on grooved substrates.
Focal complexes (FXs) and focal adhesions (FAs) were the mostly prominent types of adhesion cite. The distribution of adhesion sites exhibits a peak between 1.5 and 2 μm adhesion lengths, which is in agreement with Biggs et al. (2007Biggs et al. ( , 2008aBiggs et al. ( , 2008bBiggs et al. ( and 2009   6.

Conclusions
In this study, an accelerated buoyancy technique was used to assess how strongly foetal human osteoblast cells attach on stainless steel substrates, with and without sets of parallel grooves (created by laser processing, which left the surface chemistry largely unaffected). Three-dimensional cell morphometric analysis was used to evaluate the effect of surface topography on cell attachment. A reduction was noted in cell and nucleus volume, and cell adhesion strength on grooved, more hydrophobic surfaces, particularly with deep grooves. This is consistent with the lower number of adhesion sites and projected cell areas observed for these cases.
The XPS data were acquired using a Thermo Scientific K-Alpha instrument (Thermo Fisher Scientific, UK), with a base pressure in the analysis chamber of 5 Â 10 À 9 mbar. The source of radiation was a monochromated Al-Ka X-ray gun (1486.7 eV), operated at a voltage of 12 kV with an emission current of 1.16 mA and a 200 mm X-ray spot. For depth profiling, an Ar þ ion gun was used, with a 500 eV ion energy, 1 mA current and an etch rate 0.21 nm s À 1 (based on a calibration on a known Ta 2 O 5 standard). During depth profiling, the system base pressure was 8 Â 10 À 8 mbar. Compucentric rotation was used during the etch cycle. Data were collected using rapid snapshot acquisition (128 channels, 2 s per element, 150 eV pass energy) and processed using the Thermo Scientific Avantage software.

Surface chemistry
Typical XPS depth profiles for as-received and laser-ablated surfaces are shown in Fig