Enhanced Osteoblast Differentiation on Scaffolds Coated with TiO₂ Compared to SiO₂ and CaP Coatings

Abstract

The aim was to compare the protein release from normal human osteoblasts (NHO) cultured on scaffolds with similar morphology but different coatings. Different ceramic coatings; TiO₂, SiO₂ and calcium phosphate (CaP); Ca₉HPO₄(PO₄)₅OH, were applied to porous TiO₂ scaffolds prepared by polymer sponge replication. NHO were cultured on scaffolds in triplicates. The concentration of cytokines and Ca²⁺, and alkaline phosphatase (ALP) activity in the cell media was quantified. The secretion of osteopontin, osteoprotegerin, vascular endothelial growth factor and interleukin-6 was higher from NHO on TiO₂ compared to SiO₂ and CaP. The secretion from cells on the three scaffolds was, however, either similar or lower than the control cells cultured on plastic. The Ca²⁺ concentration was higher in cell media on CaP the first week, and no difference in ALP activity was observed. TiO₂ coating induced a higher secretion of factors indicating enhanced osteoblast differentiation as compared to CaP and SiO₂.

1 Introduction

Autologous bone grafts have long been considered the ‘gold standard’ for reconstruction of bone due to immunological feasibility and osteoinductive properties [1]. Disadvantages of grafting with autologous bone may be limited supply of bone suitable for harvest and potential donor site morbidity [2]. These factors have led to alternative approaches of bone substitute grafting.

A scaffold for bone repair must be biocompatible. In addition, it must have osteoconductive properties, preferably with a structure mimicking trabecular bone. Further the scaffold must provide immediate physical support in order to prevent collapse of the bony defect due to potential high loads to the reconstructed site [3, 4]. It is important that the scaffolds are porous and provide a high interconnectivity in order to facilitate angiogenesis, osteoblast ingrowth and attachment [3, 5]. Various synthetic bone graft materials have been developed to fulfill these criteria in order to serve as a scaffold to promote proliferation of host osteoblasts and bone growth.

Calcium phosphates (CaP), hydroxyapatite (HA) and β-tricalcium phosphate (β-TCP) in particular, have been extensively researched and widely used for bone repair as it represents the main inorganic constituent in hard tissues and possesses osteoconductive properties [6]. Further, certain types of calcium phosphates have demonstrated osteoinductive properties [7, 8], although these mechanisms are unclear [6]. Bioactive glasses represent another type of biomaterial with osteoinductive properties [9], and a porous Bioglass^® scaffold for bone repair has recently been developed [10]. However, common for calcium phosphates and bioactive glasses is low compressive strength in scaffolds designed with the porosity and interconnectivity required for bone ingrowth [10–13]. These mechanical limitations may impede scaffold use for bone repair in load bearing sites.

Titanium dioxide (TiO₂) is a biocompatible material reported to have bioactive properties [14, 15]. In addition, the TiO₂ scaffolds have demonstrated compressive strength sufficient for placement in moderately mechanically loaded bone [16], and the authors hypothesized that the TiO₂ scaffold may perform well in clinical use.

Sabetrasekh and co-workers [17] compared a TiO₂ scaffold to commercially available bone graft materials of natural and synthetic origin CaP, and titanium, and demonstrated different effects on cell viability and proliferation. However, the morphology of the tested graft materials in this study varied considerably and may be responsible for the variation in biological response. Few studies have compared similar morphology with different surface chemistry.

Therefore, the aim of the study was to compare the effect of SiO₂ or CaP coatings with TiO₂ on primary human osteoblasts in scaffolds of similar pore morphology.

2 Materials and Methods

2.1 Scaffold Production

Ceramic TiO₂ scaffolds were fabricated by a replication process [16]. In brief, TiO₂ slurry was prepared by dispersing 65 g of TiO₂ powder (Kronos 1171, Kronos Titan GmbH, Leverkusen, Germany) in 25 ml sterilised H₂O and the pH of the dispersion was kept at 1.5 for the entire duration of the stirring with small additions of 1 M HCl. The slurry was stirred (Dispermat CA-40, VMA-Getzmann GmbH, Germany) for 2.5 h at 5,000 rpm. Cylindrical polyurethane foam templates, 17 mm in diameter and 3.75 mm in height (60 ppi, Bulbren S, Eurofoam GmbH, Wiesbaden, Germany), were coated with the prepared slurry. Prior to sintering in 1,500 °C for 40 h, the polymer template was carefully burnt out of the green body at a lower temperature. 90 scaffolds were produced in total and divided into three groups of 30 scaffolds each. The final dimensions of the scaffolds were approximately 13.5 mm in diameter and 3 mm in height due to the shrinkage during the sintering phase. After relevant recoating procedures, all scaffold samples were steam sterilised in 121 °C for 20 min.

2.1.1 TiO₂

Sintered scaffolds were recoated with TiO₂ slurry prepared using the previously described procedure, only this time mixing 40 g of TiO₂ powder with 25 ml of water. Excess slurry was removed from the scaffold structure by centrifugation (1 min at 1,500 rpm; Biofuge 22R Heraeus Sepatech, Osterode, Germany). The recoated scaffolds were then sintered in 1,500 °C for 40 h.

2.1.2 SiO₂

SiO₂ slurry was prepared by dispersing 12 g of microcrystalline SiO₂ powder (Sigma-Aldrich, Steinheim, Germany) in 25 ml sterilised H₂O and stirred (Dispermat CA-40) for 2.5 h at 2,500 rpm. TiO₂ scaffolds were coated with SiO₂ by immersing the scaffolds in the prepared slurry. Excess slurry was centrifuged (1 min at 1,500 rpm; Biofuge 22R Heraeus Sepatech) out of the TiO₂ foam templates to ensure that only a thin layer of slurry covered the scaffold struts without blocking the pores. The SiO₂ coating layer was dried in 60 °C for 5 h and then heat treated at 1,200 °C for 5 h (LHT 02/17 LBR, Nabertherm GmbH, Lilienthal, Germany), the heating and cooling rates were set at 2.5 °C/min. The coating and heat treatment procedures were repeated once to apply two coating layers.

2.1.3 CaP

Calcium phosphate (CaP) coating was produced using sol–gel method [18]. 3 M triethyl phosphite (C₂H₅O₃PO; Sigma-Aldrich, Pittsburgh, PA, USA) dissolved in anhydrous ethanol was hydrolysed for 24 h with fixed amount of deionised water (the molar ratio between water and phosphite was fixed at 4:1) in a parafilm-sealed glass container under vigorous stirring. A stoichiometric amount of 3 M calcium nitrate (Ca(NO₃)₂×4H₂O; Sigma-Aldrich) dissolved in anhydrous ethanol was added dropwise to the hydrolysed phosphite colloidal solution (sol) (Ca-to-P ratio fixed at 1.67:1). The mixed sol solution was then continuously agitated for additional 10 min and subsequently aged for 24 h at room temperature. TiO₂ scaffolds were dip coated with the prepared CaP sol, and excess sol was removed from the scaffolds by blowing pressurised air through the coated scaffolds to ensure that only a thin layer of sol covered the scaffold struts without blocking the pores. The CaP coating layer was dried in 80 °C for 10 h and then heat treated at 1,000 °C for 1 h (LHT 02/17 LBR), the heating and cooling rates were set at 2 °C/min. The dip coating and heat treatment procedures were repeated twice to apply three coating layers. For the purpose of XRD analysis, TiO₂ ceramic discs were dip coated in prepared CaP sol. The TiO₂ discs were prepared using the same slurry and heating cycle used in the production of the TiO₂ scaffolds.

2.2 Scaffold Characterisation

2.2.1 Pore Architecture

The initial visualisation and optical observation of the pore structure of the prepared scaffolds were performed using a scanning electron microscope (SEM) (TM-1000, Hitachi High-Technologies, Tokyo, Japan), and a desktop 1172 micro-computed tomography imaging system (micro-CT) (Skyscan, Kontich, Belgium) was used to determine the three-dimensional microstructure and porosity of the scaffolds. All scans were performed with a voxel resolution of 8 μm. Otherwise all scanning and reconstruction (Nrecon, Skyscan) settings were performed as previously reported [16]. In order to eliminate potential edge effects, a cylindrical volume of interest (VOI) with a diameter of 8 mm and a height of 1.5 mm was selected in the centre of the scaffold. Pore architectural parameters were determined using the 3D analysis function in the standard Skyscan software (CTan, SkyScan).

2.2.2 Chemical Surface Composition of Scaffolds

The chemical compositions of the scaffold surfaces were investigated with energy-dispersive X-ray spectroscopy (EDX) using Quanta 200 FEG SEM (FEI, Denver, USA) equipped with EDX detector (EDAX, USA). The samples were scanned at environmental SEM mode with 15 kV acceleration voltage and 50 s acquisition time. The EDAX detector had a resolution of 132 eV.

X-ray diffraction (XRD) was used for additional chemical analysis of the CaP coating. XRD analysis was performed with D8 Discover powder diffractometer with a Ge (111) monochromator providing CuKα1 radiation and LynxEye detector (Bruker, Karlsruhe, Germany). The acquisition was done using θ − 2θ configurations.

2.2.3 Compressive Strength

The mechanical strength was investigated in a compressive test (Zwicki, ZwickRoell, Ulm, Germany). The compression tests were performed in accordance with DIN EN ISO 3386 at room temperature using a load cell of 1 kN with preloading force set to be 0.5 N. The scaffolds were compressed along their long axes at a compression speed of 100 mm/min until failure. The force and displacement were recorded throughout the compression and converted to stress and strain based on the initial scaffold dimensions.

2.3 Cell Culture

The utilized cells were commercially available normal human osteoblasts (NHO) of two different donors (Cambrex Bio Science, Walkersville, MD, USA) from tibia and femur, respectively. The cells were grown in osteoblast growth media (OGM; Cambrex Bio Science), and the media was changed three times per week until confluence.

At the time of scaffold seeding, the NHO from tibia had reached passage 5 and the NHO from femur had reached passage 6. The scaffolds were placed in 24-well plates (Multidish, Nuncleon Surface, Nunc., Roskilde, Denmark). Triplicate of empty plastic wells served as untreated controls and triplicates of each scaffold were harvested at each time point.

The cells were seeded from a uniform solution in a dropwise manner (0.5 ml containing 18–25 × 10³ cells) onto each scaffold. Further 0.5 ml of cell medium only was added to the wells after cell adhesion to ensure that all scaffolds were fully soaked. From this time the media was changed twice a week throughout the experiment.

At day 1, 7, 14, 21, 28, 35 and 42 cell culture media was sampled, and stored at −20 °C until analysis. At day 1, 14, 21, 28 and 42 the wells were washed in PBS prior to fixation in phosphate buffered formalin with a pH of 7.4.

2.4 Histological Analysis

The wells were immersed in an ascending alcohol solution prior to plastic embedding with a light-curing methacrylate resin (Technovit 7200 VLC, Heraeus Kulzer GmbH, Wehrheim, Germany). Slices containing parts of the cylindrical scaffolds were cut in either a vertical or horizontal direction using an Exakt saw (Exakt, Norderstedt, Germany). For the empty plastic wells, care was taken to include the well floor in the section. The histological sections were prepared according to the cutting-grinding technique previously described [19, 20]. The sections were grinded and polished (Polycut-S, Reichert-Jung, Leica Microsystems, Wetzlar, Switzerland) down to a thickness of approximately 50 μm and stained with hematoxylin and eosin, alizarin red or van Gieson. The histological sections were analysed to assess distribution of osteoblasts and mineralized nodules.

2.5 Cytokine and Chemokine Levels in Culture Medium: Immunoassay

Multianalyte profiling was performed using the Luminex 200^TM system and the XY Platform (Luminex, Corporation, Austin, TX, USA). Classification calibration microspheres were purchased from Luminex Corporation. Acquired fluorescence data were analysed by the 3.1 xPONENT software (Luminex). Prior to analysis, the samples were concentrated 5 times using Microsep Centrifugal tubes with 3 kDa cut-off (Pall Life Science, Ann Armor, MI, USA).

The concentrations of cytokines were determined using the 25-Milliplex Human Cytokine Immunoassay kit (Millipore, Billerica, MA, USA). Further, the level of bone markers interleukin-1β (IL-1β), interleukin-6 (IL-6), osteoprotegerin (OPG), osteocalcin (OC), leptin, osteopontin (OPN), parathyroid hormone (PTH), tumor necrosis factor-α (TNF-α), adrenocorticotropic hormone (ACTH), adiponectin and insulin were determined using Milliplex Human Bone Panel 1B Immunoassay kit (Millipore). All analyses were performed according to the manufacturers’ protocols.

2.6 Total Protein Content

The total protein content (TP) was quantified with a BCA^TM Protein Assay Kit (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer’s instructions. The original cell media sample was diluted 1:5 prior to the analysis to fit the standard curve. This was determined using a spectrophotometer (NanoDrop, ND-1000, NanoDrop Products, Wilmington, DE, USA). 200 μL of the kit working reagent was added to 25 μL of each diluted sample. The absorbance was measured at 562 nm in a plate reader (Expert 96, Asys. Hitech GmbH, Eugendorf, Austria).

2.7 Alkaline Phosphatase Activity

Alkaline phosphatase (ALP) activity was quantified by measuring the cleavage of p-nitrophenyl phosphate (pNPP) (Sigma-Aldrich, St. Louis, MO, USA) into a soluble yellow end-product that absorbs at 405 nm. A standard curve was constructed with calf intestinal alkaline phosphatase (Promega, Madison, WI, USA). 25 μL of media was taken from each sample and incubated with 100 μL pNPP solution in a 96-well plate for 30 min in the dark at room temperature. 25 μL of 2 N NaOH was then added to each well to stop the reaction. The absorbance was measured in a plate reader (Expert 96).

2.8 Calcium Atomic Absorption Spectrophotometry

The Ca²⁺ concentration in the cell medium was assessed using a PerkinElmer 2380 flame atomic absorption spectrophotometer (PerkinElmer, Norwalk, CT, USA). 0.5 ml of a 3.6 % HCl solution was added to the cell medium and left for incubation with agitation on a plate shaker for 24 h. The next day a solution of 5 % lanthanum (LaCl₃) and 1 % HCl was added. The spectrophotometer was calibrated using a calcium concentration standard curve ranging from 0 to 5 ppm Ca²⁺, created by the serial dilution of an atomic absorption calcium standard (Sigma Diagnostics, St. Louis, MO, USA).

2.9 Statistics

The data obtained by protein analyses and calcium atomic absorption spectrophotometry was compared between and within groups using Tukey’s test following a parametric one way ANOVA. Where the equal variance and/or the normality test failed, a Kruskal–Wallis one way ANOVA on ranks was performed (SigmaPlot 12, Systat Software Inc, San Jose, USA). Mean and standard deviation are presented. A probability of <0.05 was considered significant.

3 Results

3.1 Scaffold Characterisation

All fabricated scaffolds showed an open and highly interconnected pore structure formed by spherical macropores, the average size of which exceeded 400 μm (Fig. 1). The average porosity of the scaffolds was 92 ± 2 % (n = 6) and pore network showed over 95 % interconnectivity through connections as large as 160 μm. The volumetric scaffold parameters are listed in Table 1.

Fig. 1

Comparison of pore structure and surface morphology of the different scaffold groups

Table 1 Volumetric parameters (mean ± standard deviation) of the microCT morphological analysis for scaffolds coated with TiO₂ (n = 6)

The thickness of the coating layer was estimated from SEM images to be approximately 1–3 μm for both CaP and SiO₂ coatings, and thus was not considered to affect the overall porosity or the pore architectural parameters, such as pore size, surface area-to-volume ratio, and interconnectivity. However, some closed pore walls were seen close to the bottom edge on some of the SiO₂ scaffolds. Due to limited thickness of the coating layer and the more radiolucent nature of CaP and SiO₂ coatings in comparison to TiO₂ bulk material, the coating layers were scarcely visible in micro-CT images.

While the pore structure of the scaffolds from the three different groups was regarded similar to each other, the scaffolds differed greatly in their surface properties. Typical surface morphologies of each of the three scaffold groups are presented in Fig. 1. The EDX analysis of the scaffold surfaces revealed that the surface of the TiO₂ scaffolds consisted of titanium and oxygen (Ti: 52 weight percentage (wt%), O: 46 wt%) with a small amount of aluminum impurities (2 wt%), while only silicon and oxygen was detected on the surface of SiO₂ coated scaffolds (Si: 48 wt%, O: 52 wt%). XRD pattern of the CaP coating revealed the chemical composition of the coating to match with calcium hydrogen phosphate hydroxide (Ca₉HPO₄(PO₄)₅OH) as shown in Fig. 2.

Fig. 2

XRD pattern of the CaP coating. Chemical composition of the coating identified the CaP as calcium hydrogen phosphate hydroxide. Rutile (TiO₂) (times) and Ca₉HPO₄(PO₄)₅OH (plus)

The TiO₂ coated scaffolds had the highest compressive strength. The compressive strength test results are listed in Table 2.

Table 2 Compressive strength (mean ± standard deviation) (n = 5)

3.2 Histology

Several hundred sections were made in total, representing all harvest time points, all groups and both donors. Disappointingly, no mineralized nodules could be observed, and osteoblasts were scarce and only occasionally visible. The histological analysis was thus excluded from the study.

3.3 Enhanced Secretion of OPN and OPG from Cells on TiO₂

The level of IL-1β, PTH, TNF-α, ACTH, adiponectin and insulin were all below reliable detection and the level of leptin was detected in the femur donor only.

The secretion of leptin was similar for all treatment groups until day 42 where the secretion from cells on TiO₂ were higher than on CaP and SiO₂ (P = 0.016, P = 0.015, respectively). No difference was observed at any time point between TiO₂ and plastic (data not shown).

There were no differences in the secretion of OC from cell cultures in the different scaffolds (data not shown).

The secretion of OPN from both tibia and femur osteoblasts were higher on TiO₂ than from those on CaP and SiO₂ at day 7 (P = 0.013, P = 0.043 and P < 0.001, P = 0.001, respectively), day 14 (P < 0.001, P = 0.002 and P = 0.003, P = 0.014, respectively), day 21 (P = 0.003, P = 0.008 and P < 0.001, P = 0.012, respectively), and in the femur donor only at day 28 (P < 0.001, P = 0.001, respectively) and at day 42 (both P < 0.001) (Fig. 3).

Fig. 3

The mean concentration (ng/ml) of OPN and OPG in cell culture media from primary human osteoblasts from tibia (left column) and femur (right column) at each time point (days). The error bars represent the standard deviation of the mean (n = 3). CaP = Ca₉HPO₄(PO₄)₅OH

Cells on CaP had no change in the secretion of OPN, and was significantly lower than the secretion from cells on SiO₂ in femur at day 28 (P = 0.046).

The secretion of OPN from osteoblasts, both from tibia and femur, was enhanced on plastic during the study period (day 1; 0.32 ± 0.09 ng/ml and 0.31 ± 0.06 ng/ml, respectively, day 7; 1.54 ± 0.34 ng/ml and 1.1 ± 0.07 ng/ml, respectively, and day 14; 2.38 ± 0.25 ng/ml and 1.2 ± 0.08 ng/ml, respectively). The secretion of OPN from cells on TiO₂, however, was lower than the secretion from control cells at day 7 (P = 0.003 and P < 0.001, for tibia and femur osteoblasts, respectively), day 14 (P < 0.001 and P = 0.02, respectively), and day 28 (P = 0.034) in the femur donor only. After day 28 there is no significant difference between the secretions from cells on plastic compared to TiO₂.

The concentration of OPG was enhanced in all groups (Fig. 3). The secretion of OPG from tibia osteoblasts on TiO₂ was higher than from cells on SiO₂ at day 7 (P = 0.017) and day 14 (P = 0.008), and the secretion from femur osteoblasts on TiO₂ was higher than from cells on CaP and SiO₂ at day 14 (P = 0.023 and P = 0.025, respectively), day 21 (P = 0.001 and P = 0.039, respectively), and at day 28 compared to CaP only (P = 0.007). On plastic, the OPG secretion from tibia and femur osteoblasts were 0.22 ± 0.001 and 0.32 ± 0.01 ng/ml, respectively, at day 1, and 1.78 ± 0.36 and 1.11 ± 0.04 ng/ml, respectively, at day 7. The secretion of OPG on plastic, however, was significantly higher than TiO₂ only in the femur donor at day 1 (P < 0.001).

3.4 Enhanced Secretion of IL-6 and Rantes from Cells on TiO₂

The secretion of IL-6 from tibia osteoblasts on TiO₂ was higher than from cells on CaP (P < 0.001) and SiO₂ (P < 0.001) at day 28 only. The secretion of IL-6 from femur osteoblasts was higher compared to from cells on CaP at day 21 (P = 0.003) (Fig. 4).

Fig. 4

The mean concentration (pg/ml) of IL-6, IL-4 and RANTES in cell culture media from primary human osteoblasts from tibia (left column) and femur (right column) at each time point (days). The error bars represent the standard deviation of the mean (n = 3). CaP = Ca₉HPO₄(PO₄)₅OH

The secretion of IL-6 was enhanced from both tibia and femur osteoblasts on plastic (day 1; 224 ± 18 and 274 ± 34 pg/ml, respectively, day 7; 378 ± 61 and 403 ± 83 pg/ml, respectively, and day 14; 545 ± 135 and 560 ± 0 pg/ml, respectively). The IL-6 secretion from cells on plastic was higher than the secretion on TiO₂ in cells of tibia origin only at day 21 (P = 0.003) and day 28 (P < 0.001).

There were no differences in the secretion of IL-4 from cells on TiO₂ and CaP. In comparison the secretion from tibia and femur osteoblasts on SiO₂ were notably lower than on both TiO₂ and CaP the first 14 days (day 1; P = 0.002, P = 0.002 and P = 0.009, P = 0.023, respectively, day 7; P < 0.001 for all, and day 14, P < 0.001, P < 0.001 and P = 0.024, P = 0.006, respectively) (Fig. 4). There were no significant difference in the secretion from cells on TiO₂ and plastic. A similar pattern of secretion was observed for IL-8, with no differences between cells cultured on plastic, TiO₂ or CaP, but a reduced secreted level from cells on SiO₂ the first time points (data not shown).

The secretion of RANTES was enhanced from both tibia and femur osteoblasts on TiO2 from day 1; 0.6 ± 0.04 and 1.4 ± 0.6 pg/ml, respectively, to day 42; 3.1 ± 0.6 and 10.5 ± 4 pg/ml, respectively (Fig. 4). The secretion of RANTES from tibia osteoblasts was higher on TiO₂ than from cells on SiO₂ at day 7 (P = 0.004), and both CaP and SiO₂ at day 14 (P = 0.002, P = 0.022, respectively). Femur osteoblasts secreted higher levels on TiO₂ compared to CaP and SiO₂ at day 21 only (P = 0.006, P = 0.021, respectively). No difference was observed at any time point in either donor comparing TiO₂ and plastic.

3.5 Angiogenic Markers

The secreted levels of MCP-1 were higher from cells on TiO₂ and SiO₂ compared to from cells on CaP (Fig. 5). However, the difference was not significant for other time points than day 28 in the tibia donor (TiO₂: P < 0.001 and SiO₂: P < 0.001).

Fig. 5

The mean concentration (pg/ml) of MCP-1 and VEGF in cell culture media from primary human osteoblasts from tibia (left column) and femur (right column) at each time point (days). The error bars represent the standard deviation of the mean (n = 3). CaP = Ca₉HPO₄(PO₄)₅OH

The secretion of MCP-1 was enhanced for cells in both the tibia and femur donor on plastic (from day 1; 0.25 ± 0.01 and 0.7 ± 0.02 ng/ml, respectively, to day 7; 0.9 ± 0.01 and 1.1 ± 0.02 ng/ml, respectively). The MCP-1 secretion from cells on plastic was similar to that of cells on TiO₂, and a difference was observed at day 1 in the femur donor only (P = 0.008).

The secretion of VEGF was enhanced from femur osteoblasts on TiO₂ from day 1; 0.03 ± 0.01, to day 42; 0.6 ± 0.08 ng/ml (Fig. 5), and secretion was higher than from cells on SiO₂ at day 7 (P = 0.045), CaP at day 14 (P = 0.028), CaP and SiO₂ at day 21 (P < 0.001, P < 0.001, respectively), day 28 (P < 0.001, P = 0.004, respectively) and day 42 (P < 0.001, P < 0.001, respectively). No significant differences were observed in the tibia donor and no difference was observed at any time point for either donor between TiO₂ and plastic.

3.6 Alkaline Phosphatase

The ALP activity of day 1 was similar to that of day 42 for all groups in both donors, and ALP was secreted in a steady-state throughout the 6 weeks on all scaffolds and plastic (data not shown).

3.7 Total Protein Content

There was no difference in TP secretion among groups in either donor. Osteoblasts from tibia demonstrated a threefold increase of the TP concentration from day 1 to the end of the experiment, whereas osteoblasts from femur displayed a minor increase only. Plastic did not exhibit any difference compared to the test groups at any time in either donor (data not shown).

3.8 Ca²⁺

The secretion of Ca²⁺ was not altered in cells from either the tibia or femur donor on plastic, TiO₂ or SiO₂ (data not shown). An initial high concentration was, however, found in the cell culture media from tibia osteoblasts on CaP at day 1 and 7 (P < 0.001) compared to all other groups, and in osteoblasts from femur compared to TiO₂, SiO₂ and plastic at day 1 (P = 0.004, P = 0.005, P = 0.004, respectively) and at day 7 (P < 0.001, P = 0.003, P = 0.002, respectively).

4 Discussion

The TiO₂ coating induced a higher osteoblast secretion of OPN, OPG, IL-6, RANTES and VEGF compared to CaP and SiO₂ coatings. The two latter coatings did not induce significantly higher secretion of any detectable cytokine at any single time point compared to the TiO₂ coated group. Thus, TiO₂ coated scaffolds demonstrated osteoblast secretion levels closer to the control, although neither tested coating could display osteoblast secretion higher than that of the control for any cytokine at any single time point.

Webster and co-workers [21] compared rat osteoblast function on discs of TiO₂, alumina, hydroxyapatite and borosilicate. Although the study aimed to evaluate ceramic grain size, nanophase TiO₂ induced higher levels of ALP and Ca²⁺ compared to hydroxyapatite and borosilicate. The cell density was similar on TiO₂ and hydroxyapatite at 5 days, both higher than borosilicate, whereas the cells were more spread on borosilicate compared to TiO₂ and hydroxyapatite.

The results suggested that nanophase TiO₂ support rat osteoblast differentiation better than borosilicate and hydroxyapatite. Similarly, TiO₂ supported higher secretion of osteoblastic markers compared to calcium phosphate and silica in the present study. Although no differences were observed in ALP activity and the Ca²⁺ release only differed the first week, the level of other markers suggested enhanced osteoblast differentiation. Regarding cell proliferation, Webster and co-workers [21] found higher cell density in the TiO₂ and hydroxyapatite groups compared to the borosilicate control at day 3 and 5. In the present study no difference in total protein content was observed. However, these methods of evaluating cell proliferation vary considerably, and may not be comparable. Detection of significant differences in the total protein content in cell media consisting of foetal calf serum require great variation, especially if the cell number is low.

Sabetrasekh et al. [17] evaluated cell proliferation using human mesenchymal stem cells cultured on TiO₂ granules and compared it to other bone graft substitute granules, one consisting of hydroxyapatite and one of hydroxyapatite and β-tricalcium phosphate in a 60:40 ratio. The cell number was significantly higher on TiO₂ compared to the calcium phosphates at both day 1 and day 3. Again the cell proliferation methodology may not be compared to the present study. The results of these studies suggest that other analyses should be employed in further research to detect significant differences in cell proliferation.

Although the porosity and interconnectivity was similar across the differently coated scaffolds, the surface roughness was different for the three coatings (Fig. 1). It has been demonstrated that cell differentiation is sensitive to both micro- and submicroscale surface roughness, and increased roughness is associated with enhanced osteoblast differentiation and local factor production in vitro [22, 23]. Thus, other than the chemistry of the coating material, the difference in surface roughness may therefore have attributed to the results. However, the TiO₂ coated scaffolds had the smoothest surface, yet higher secretion, suggesting that the chemistry of the coating materials indeed had significant impact.

The XRD analysis demonstrated the CaP material to be Ca₉HPO₄(PO₄)₅OH, known as calcium-deficient hydroxyapatite. This material has been described previously, and has been reported to have properties comparable to β-tricalcium phosphate with regard to osteogenic induction [24]. CaP demonstrated higher concentrations of Ca²⁺ ions at the first two time points in both donors. This was probably due to passive dissolution of Ca²⁺ ions from the CaP material the first week, which has been reported previously [25, 26]. As the cells on CaP coated scaffolds reached the steady-state concentration of Ca²⁺ ions observed in the other groups by day 14, this may suggest that by then the dissolution had reached equilibrium. The dissolution of calcium did not, however, seem to enhance expression of osteoblastic markers as previously suggested [27, 28].

The histological analysis did not contribute to any results. However, the absence of mineralized nodules was in agreement with the non-detectable changes in the concentrations of OC, Ca²⁺ and ALP activity, indicating that mineralization did not take place. Alizarin red and van Gieson staining were also performed as the hematoxylin and eosin stain did not render measurable results, but with the same outcome. The histological analysis suggests there may have been too few cells in the experiment. A three-dimensional scaffold may necessitate a higher quantity of cells in order to observe proper mineralization. The lack of mineralized nodules observed on plastic may further suggest a low cell number. The chosen methodology of plastic embedding and grinding prior to histological analysis may not have been appropriate for the in vitro experiment performed.

Zhou and co-workers [29] found that a small seeding density of human alveolar osteoblasts on mPCL-TCP scaffolds could support cell attachment and proliferation, secretion of OC and ALP, as well as extracellular matrix synthesis. Further, the authors observed that seeding at higher densities did not influence the extent of cell proliferation and tissue formation over time. However, the lowest cell density tested in that study was 5 × 10⁴ million cells per scaffold, utilizing a scaffold size of 3 × 3 × 5 mm². The cell density in the present study was in comparison 1.8–2.5 × 10⁴ million, cultured on considerably larger scaffolds of 13.5 mm in diameter and 3 mm in height. This may suggest that the cell density was too small for a large three-dimensional substrate in order to achieve significant variation in OC and ALP levels, and histologically detectable mineralization. On the other hand, osteoblasts cultured on plastic did render similar results as the test groups, despite the lower total surface area. As mineralization was not evident in any scaffold or on plastic, it is tempting to suggest that the results would have been different with a higher cell seeding number. In the present study a static culture condition was employed. A dynamic culture condition in combination with a higher cell seeding number may further modify the outcome. This has previously been reported to improve cell invasion and cellular attachment inside plastic and titanium scaffolds [30, 31], but will have to be confirmed by further in vitro experiments.

5 Conclusions

A higher secretion of bone markers (OPN, OPG), immune modulators (IL-6, RANTES) and angiogenic markers (VEGF) indicate that TiO₂ support osteoblast growth and bone remodeling better than CaP and SiO₂. This study demonstrates that TiO₂ scaffolds are able to induce human osteoblast differentiation in vitro, suggesting its potential as a scaffold for hard tissue repair.