Modified Titanium Surfaces Alter Osteogenic Differentiation: A Comparative Microarray-Based Analysis of Human Mesenchymal Cell Response to Commercial Titanium Surfaces

Commercial pure titanium (Ti) and its alloys are currently used as implant materials because they exhibit excellent mechanical properties, chemical stability, and biocompatibility.¹  The biocompatibility of Ti is closely related to its surface properties: roughness, topography, and chemistry.²  Implant surface modification by altering topographical and chemical surface properties may enhance the process of osseointegration.³ 

Both in vivo and in vitro research efforts have extensively studied the effect of surface roughness to the outcome of osseointegration. In vivo, rough surfaces have demonstrated to promote bone-to-implant contact as compared to smooth surfaces.^4,5  In addition, it has been demonstrated that the removal torque of rough implants is significantly increased corresponding to a higher anchorage level than that of smooth implants.^6,7 

In vitro research on the effects of surface roughness to osseointegration has provided controversial results. Human cells with a prominent fibroblastic phenotype, such as gingival fibroblasts or human periodontal ligament cells have demonstrated a greater attachment rate to smooth rather than rough surfaces.^8,9  On the other hand, human cells with a clear osteoblastic phenotype attach preferentially to rough surfaces.^10–12 

Surface roughness regulates in vitro cellular activities; proliferation, differentiation, initial attachment, and protein production are upregulated when cells are grown on rough surfaces. Also, it has been evident that as surface roughness increases, cell differentiation is driven towards a clear osteoblastic phenotype where cells secrete alkaline phosphatase, osteocalcin, osteoprotegerin, and transforming growth factor-beta (TGF-β).^13–15  Increased osteogenic differentiation in response to rough surfaces was correlated to elevated alkaline phosphatase activity and nodule formation and simultaneously to decreased cell proliferation. It was also indicated that cell number and proliferation was not a determinant of osteogenic responses to rough Ti surfaces.^16,17  Moreover, surface modifications may also regulate osteoblast differentiation transcription factors.^18,19 

An implant surface can be modified by either additive or subtractive methods. Commercial techniques for surface treatment mainly include acid etching, sandblasting and plasma spray coating of hydroxyapatite (HA).²⁰  Acid etching is a useful technique to modify the Ti surfaces. It can create regular micro pits in the range of micron to submicron size. OSSEOTITE is a typical example of the double acid-etching method applied on machined surface.²¹  A commonly used subtractive methodology involves a simultaneous large-grit sand-blasting and an acid-etching procedure (SLA modification). The SLA implant surface has been mechanically modified (sand-blasting) where microcavities of 20–40 μm in diameter are produced and acid treated where micropits of 0.5–3 μm in diameter are resulted (acid-etched).²² 

Previous research efforts failed to demonstrate that a particular surface treatment or modification optimizes clinical success.^23,24  However, it has been evident that a certain degree of roughness increases bone-implant contact, especially in the earlier healing process.²⁵  In addition, a 1.5 μm range in surface irregularities produced a higher bone response than any smoother or rougher surface.²⁶ 

The vast majority of the in vitro research studies that examined the effect of Ti surface modifications were conducted using human primary osteoblasts or transformed cells.^27–29  So far, little evidence has been published regarding the effect of different titanium surface roughness to human mesenchymal stromal cell (hMSC) populations.³⁰  hMSCs are isolated from bone marrow, they have the ability to differentiate via different pathways into osteogenic, chondrogenic, and adipogenic lineages.^31–33  hMSCs with a prominent osteogenic potential can be harvested from the fraction of bone marrow cells expressing the stro-1 surface epitope.³⁴  These cells can form de novo bone tissue when transplanted in vivo³⁵  and are easily differentiated at an in vitro level.³⁶  These properties make hMSCs very important tools in studying the biological phenomenon of transcriptional control of osteogenic differentiation and its regulation by different implant surface modifications.

In the present investigation, the effects of the 3 commercial implant surface modifications (OSSEOTITE, SLA and HA) on hMSC initial adhesion, proliferation and primarily to its potential to induce expression of genes involved in osteogenesis were assessed. A smooth Ti surface was used as control. DNA microarray analysis was employed to create an osteoblastic gene expression profile of hMSCs cultured on different Ti surfaces. Such an analysis allows both qualitative and quantitative interpretation since it is a sensitive method detecting a change of expression level in the investigated cells when compared to normal samples.^37,38  The human DNA microarray utilized contains 109 different genes that regulate the process of early osteogenesis during osseointegration.

Early passage cell populations of were obtained from commercial sources (Biowhittaker, San Diego, Ca). Cells were allowed to grow in Minimum Essential Medium supplemented with 10% (v/v) fetal bovine serum, L-glutamine (2 mM), nonessential amino acids (1×) and penicillin/streptomycin/fungizone stock solutions (all culture media were purchased from Invitrogen/LTI, Paisley, UK). Cells were maintained in standard culture condition (37°C in a 5% CO₂ humidified atmosphere). Culture medium was changed every 2 days until confluency was achieved. All experiments were performed using cells up to the third passage. One mL of cell suspension was applied carefully to 24-well plates containing Ti discs of the examined surfaces. Each experiment was performed in triplicate.

All disc specimens were provided by commercial manufacturers. Discs were fabricated from commercially pure grade 4 Titanium sheets 1 mm in thickness and 15 mm in diameter in orders to fit into the wells of a 24-well cell culture plate. Four different titanium surfaces were prepared; (1) a relatively smooth surface (SMO) using SiO grinding paper (grit size was 15–600 μm in diameter) and polished with a 10 μm diamond paste in oil and finally with a 0.06 μm SiO₂ suspension, (2) a dual acid etched (HCl/H₂SO₄) implant surface (OSSEOTITE), (3) a standard rough surface (SLA) upon sandblasting with large grits of 0.25-0.50 mm and etched with HCl/H₂SO₄, and (4) a hydroxyapatite plasma spray coated surface (HA).

Initial hMSC cell attachment was assessed using a 3-[4,5-dimenthylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) (Sigma-Aldrich, Dorset, UK) dye reduction assay.³⁹  hMSC cells were placed and allowed to grow onto the Ti discs at a cell density of 1 × 10⁵ cells per disc in 1 mL of standard culture medium. Cells were cultured for 3 hours; nonadherent or apoptotic cells were washed off with phosphate buffer solution. Medium was then replaced and 250 mL MTT was added per well, cells were incubated for 1 h at 37°C. Following medium removal, 1 mL of extraction buffer (10% w/v sodium dodecyl sulphate/0.5 mol/L dimethylformamide; Sigma-Aldrich) was added to each disc and cells were incubated overnight at 37°C. Then, 100 μL of each solution was transferred to a 96-well plate and the absorbance of each well was assessed using an Angstrom Advanced Inc. spectrophotometer (Angstrom Advanced Inc. Technologies, Braintree, Mass) at 550 nm.

Proliferation of hMSC cells on Ti discs was determined using the MTT dye-reduction assay as previously described. Cells were allowed to grow onto Ti disks at a cell density of 5 × 10⁴ cells per disc in 1 mL of culture medium. In this way, cells were allowed to grow for the time interval of 1, 3, and 7 days. At this time point 250 mL of MTT (5 mg/mL) was added after culture media solution was discarded and plates were incubated overnight at 37°C. One hundred mL aliquots were transferred to a 96-well plate and absorbance of each well was determined.

For the total RNA (tRNA) preparation studies, cells were cultured on Ti discs at a density of 2 × 10⁵ cells per disc over a time course of 7 days. tRNA was isolated from cultured cells, which adhered to the retrieved Ti discs of different surfaces with RNAzol (RNAzol B; Biogenesis, Dorset, UK) and quantified via UV Spectrophotometry (Spectrophotometer-DU650).

cDNA probes for microarray analysis were prepared from total RNA templates as previously described⁴⁰  using Superscript II reverse transcriptase and random hexameric primers (Invitrogen/LTI, Paisley, UK). The produced cDNA was cleaned by Quick Clean Purification Resin (Clontech, Mountain View, Calif), precipitated with sodium acetate/ice-cold absolute ethanol (−20°C, 1 hour) and finally collected by centrifugation (21 000g, 20 minutes, 4°C). The cDNA quality and quantity were determined by spectophotometry. Additionally, the cDNA content was precisely quantitated using the Agilent 2100 bioanalyzer in combination with the DNA chips (DNA 12000 Assay), according to the manufacturer's instructions (Agilent Techologies, Waldbronn, Germany), and normalized cDNA quantity was used for the label reaction and microarray hybridization.

The rhodamine derivative fluorescent dye Alexa 546 was used. cDNA labeling was performed by the use of Ulisis Alexa Fluor Nucleic Acid Labeling Kit as recommended by the manufacturer (Molecular Probes, Eugene, Ore). The GEArray Q series Human Osteogenesis Gene Array, including 109 cDNAs in quadruplicates was used (cDNA GEArray^T kit for Human Osteogenesis, HS-026-4, www.superarray.com) (Table 1). The solution was applied to a microarray membrane and the hybridization was performed at 60°C, overnight and the membrane washed according to the manufacturer's instructions.

We assumed that hMSC cell counts follow a Poisson distribution. To assess differences among surfaces in cells attachment and proliferation, we used Poisson regression models with depended variable for the cell counts and independent of the type of surface. Differences in cell counts are expressed as relative % change.

DNA microarray analysis examined alterations in early gene expression using a human osteogenesis gene array, including 109 cDNAs in quadruplicates of major regulatory genes for osteogenesis. Analysis was carried out using QuantArray Analysis Software (Perkin-Elmer, Wellesley, Mass). All hybridized membranes containing the 109 human genes were scanned using the ScanArray EXPRESS microarray scanner (Perkin Elmer Life and Analytical Science, Wellesley, Mass). After membrane scanning, image analysis and data normalization, genes with a mean fluorescent intensity ratio over two compared to control in each scanned membrane were considered as significantly upregulated. With this consideration, the statistical significance of the difference between the mean ratios for each gene at treated and untreated cells, in each independent experiment as well as in the four independent experiments carried out, assessed by paired Student's t-test. Statistical significance was also evaluated using 1-way ANOVA with Dunnett's correction for multiple comparisons, comparing each group of genes (5) to control at a 5% significance level.

Using a MTT dye-reduction assay, hMSCs initial attachment was measured 3 hours after seeding with optimal density measurement used as a relative measure of absolute cell number. Cell attachment was found significantly increased when cells were seeded on the OSSEOTITE implant surface as compared to control (smooth surface-SMO) (P < .001) (Table 2, Figure 1). Meanwhile, attachment was significantly reduced when cells were grown on the SLA and the HA surfaces compared to control (Table 2, Figure 1) (P < .001). HA surface provoked the weakest initial hMSCs attachment compared to all surfaces (Table 2, Figure 1) (P < .001).

Absolute hMSCs numbers were determined using the MTT dye-reduction assay at a 1, 3, and 7 days' time interval.

At Day 1 of the experiment, all tested implant surfaces revealed an increased rate of cell proliferation as compared to control, however, only the SLA surface demonstrated a clear and statistically significant increase (P < .001) in hMSC cell proliferation as compared to both control and other implant surfaces (Table 3, Figure 2).

At Day 3, cell proliferation was found to be increased on both the SLA and the HA surface in a statistically significant manner (P < .001) when compared to the SMO surface (Table 4, Figure 3). Cell proliferation was also found increased when cells were seeded on the OSSEOTITE implant surface as compared to control but not significantly (P = 0.137) (Table 4, Figure 3). Similarly to Day 1, the SLA surface demonstrated a statistically significant increase (P < .001) in hMSC cell proliferation as compared to both control and other implant surfaces (Table 4, Figure 3).

Finally, at Day 7, hMSCs were found to proliferate on all tested surfaces in a statistically significant higher rate when compared to the control (P < .001) (Table 5, Figure 4). Again, the SLA surface demonstrated a statistically significant increase (P < .001) in proliferation as compared to both control and tested surfaces (Table 5, Figure 4).

Gene expression analysis demonstrated that; (1) 15 genes was found to be significantly upregulated when hMSCs were cultured on the OSSEOTITE surface, (2) 19 were significantly upregulated when hMSCs were cultured on the SLA surface and (3) 16 genes were significantly upregulated when hMSCs were cultured on the HA surfaces. The genes differentially upregulated in OSSEOTITE, SLA and HA implant surfaces are reported in Table 6. Upregulated genes control cell differentiation, signal transduction, cell cycle regulation, angiogenesis, cell adhesion, and extracellular matrix and bone formation.

Rough Ti surfaces have been clinically proven to provoke an accelerated bone response during osseointegration than smooth surfaces.⁴¹  At cellular and molecular level, the mechanisms that dictate the process of osseointegration in response to rough surfaces remain unclear. In this research study, the most prominent objective was to identify the transcriptional regulation of a variety of genes related to osteoblastic differentiation in response to different Ti surfaces. For this purpose, primary hMSC cultures were allowed to grow on four different current industry standard titanium surfaces; (1) a relatively smooth surface (SMO), (2) a dual acid-etched (HCl/H₂SO₄) implant surface (OSSEOTITE), (3) a standard rough surface (SLA) and (4) a hydroxyapatite plasma spray coated surface (HA). In this way, the ability to induce osteogenic pathways of differentiation in populations of hMSCs is examined. The rationale behind that is the fact that hMSCs is the main cell type recruited to sites of bone damage and repair when bone marrow is exposed to mechanical trauma or stress (ie, implant site preparation to alveolar bone).

At cellular level, initial cell attachment and proliferation was assessed using the MTT reduction assay and at molecular level, transcriptional regulation were analyzed by DNA human osteogenesis gene array composed of 109 genes critical for bone healing. Initial cell attachment was significantly increased when cells were seeded on the dual acid etched (HCl/H₂SO₄) implant surface (OSSEOTITE) (P < 0.001) compared to both control and other tested surfaces. However, cell attachment was significantly reduced on the standard rough surface (SLA) and the hydroxyapatite plasma spray coated surface (HA) as compared to control (P < .001).

Quite surprisingly, cell proliferation after the first day of incubation was significantly upregulated (P < .001) when cells were allowed to grow on the SLA surface not only compared to control but also compared to the OSSEOTITE (which initially induced a higher cell attachment) and to HA surface. After 3 and 7 days of culture, hMSCs continued to behave in a similar manner; they tended to proliferate on the SLA surface in a continuously increased rate as compared to the other surfaces. These results clearly demonstrate that hMSCs were sensitive to the Ti surface modifications. Cell attachment and proliferation are key determinants of biocompatibility as they dictate cell survival and growth prior to matrix deposition during bone healing.

Previous studies on osteoblast-like cell proliferation in response to rough Ti surfaces revealed a surface-roughness dependent mechanism.^27–29,42  Such a mechanism was closely related to osteogenic differentiation; when osteoblastic cells come in contact with roughened or chemically modified surfaces they either differentiate to mature osteoblast cells with a prominent secretory phenotype or they are unable to survive and undergo spontaneous or programmed cell death (apoptosis).⁴²  In addition, it has been evident that osteoblast-like cells can survive, proliferate and differentiate on rough Ti surfaces with little selection or cell death,⁴³  whereas nonosteoblast cell types (eg, fibroblasts) prefer smooth surfaces.⁴⁴ 

Transcriptional analysis of the major genes, which are involved in osteogenic differentiation, demonstrated a clear upregulation of the pathways related to bone induction and formation in response to roughened surfaces as compared to smooth surfaces. Interestingly, genes significantly downregulated were not reported. Upregulated genes are critical for certain functions such as: signal transduction, regulation of transcription, cell cycle regulation, angiogenesis, cell adhesion, ECM formation, and cell differentiation.

Genes related to signal transduction pathways were significantly upregulated when hMSCs were grown on rough surfaces; SMAD7 (Mothers Against DPP homolog-Drosophila) and SMAD9 were found to be significantly upregulated in cells grown on the OSSEOTITE surface, SMADH5 and SMADH7 were found significantly upregulated only in the SLA surface and SMADH5 and SMADH6 were found significantly upregulated on the HA surface. SMAD molecules act as signal transducers of both TGF-beta and BMP signals, from the cell membrane to the nucleus and are implicated in bone formation.⁴⁵  Hence, rough surfaces may promote cell signaling that leads to cell differentiation towards a mature osteoblastic phenotype.

Genes involved into the regulation of transcription were found significantly upregulated; NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) was found to be upregulated by cells cultured on all rough surfaces tested as compared to smooth surfaces; NF-κB exists as a protein complex that regulates DNA transcription. It can also act as a “rapid-acting” primary transcription factor, ie, it is present in cells in an inactive state and does not require new protein synthesis to be activated, hence NF-κB has the potential to act as a “first responder” to harmful cellular stimuli. In this way, upregulation of this molecule during the early stages of osseointegration can be important for its fate.

Genes critical for angiogenesis such as the vascular endothelial growth factor C gene were significantly upregulated when cells were grown on the OSSEOTITE surface. This observation is consisted with previous studies^46,47  that showed a close relationship in the gene expression regulation of osteogenesis and angiogenesis coupling.

Both HA and SLA surfaces were found to promote gene expression of ICAM-1 (Intercellular Adhesion Molecule-1, CD54). ICAM-1 is an endothelial- and leukocyte-associated transmembrane protein long known that stabilizes cell-cell interactions and facilitates leukocyte endothelial transmigration.⁴⁸  Upregulation of cell adhesion molecules by implant surfaces highly improves their biocompatibility properties and possibly enhance the clinical success since adhesion molecules control a variety of processes including embryogenesis, homeostasis, tissue repair, and immune response.

Genes related to ECM (extracellular matrix) formation were upregulated when hMSCs were cultured on roughened implant surfaces; gene expression of Collagen, alpha 1, type XXIX (COL19A1) was found to be significantly upregulated compared to cells grown on the control surface. COL19A1 gene provides instructions for making a component of collagen that strengthens and supports many tissues in the body, including bones and the extracellular matrix. In addition CD36 (cluster of differentiation 36 or collagen type I receptor, thrombospondin receptor) was found to be upregulated. CD36 interacts with a number of ligands, such as collagen types I and IV. CD36 forms part of a nonopsonic receptor (the scavenger receptor CD36/alphaV beta3 complex) and is involved in homeostasis of ECM.

Clusters of genes related to osteoblastic differentiation and bone formation were found to be upregulated when hMSCs were seeded on implant surfaces. Such genes were primarily members of the TGF superfamily; BMP 6 (Bone morphogenetic protein 6), bone morphogenetic protein receptor, type IA, and transforming growth factor, beta receptor II (70/80kDa) were upregulated in cells grown on the OSSEOTITE surface compared to control surface. BMP proteins dictate the differentiation of osteoblasts from mesenchymal osteoprogenitor cells and most importantly and stimulate higher bond strength at the level of bone-implant interface resulting in reduced healing times.⁴⁹  Previous studies also noted that BMP molecules are mainly expressed during early stages of bone morphogenesis.⁵⁰ 

Gene expression of growth factors that regulate bone formation and vitality [Insulin-like Growth Factor 1 (IGF1), Insulin-like Growth Factor 2 (IGF2) and Insulin-like Growth Factor 1 receptor (IGF1R)]^51,52  was also upregulated in cultures grown on tested implant surfaces compared to control. A clear cross talk mechanism exists between BMP and IGF molecules during the bone healing process; BMPs have been shown to be major regulatory molecules for IGF gene expression, coincident with increased alkaline phosphatase activity in human osteosarcoma cells and calvarial derived osteoblast cells in fetal rats.^53–55  IGF2 that was found to be upregulated in cell grown on implant surfaces enhances keratinocyte cell migration and proliferation, which are essential for wound healing and skin regeneration. Also loss of genomic imprinting of IGF2 is associated with cellular proliferation in normal hematopoietic cells, as well as with mitogenic effects on Malassez cells in normal periodontal ligament.⁵⁶ 

IGF molecules regulate bone formation in both an autocrine and a paracrine mode. IGF-2 synthesis is regulated by several skeletal factors produced by bone.⁵⁶  Hence both surfaces examined seem to assist bone healing during osseointegration.

A significant upregulation was noticed in gene expression of the fibroblast growth factor (FGF) 2 and 3. FGF molecules are major regulators of the differentiation and development of osteoprogenitor cells to mature osteoblasts with secretory functions. Both acidic and basic FGF hormones affect bone development and maintenance. In addition, the major receptor for FGF (FGFR1) was expressed by cells grown on the OSSEOTITE surface. This indicates that bone induction may be performed via the FGF/FGFR regulatory axis.

In addition, platelet-derived growth factor A polypeptide (PDFGA) gene expression was significantly increased when hMSCs were grown on both SLA and the HA implant surface compared to control. PDGFA (member of the PDGF family) is a potent mitogen for cell growth and differentiation and induces bone formation by osteoblastic cells. Hence, such Ti surfaces might act as potent stimuli for cell growth and differentiation during osseointegration.

The most prominent objective of the present study was to analyze the effects of various forms of surface roughness and modifications on cellular and molecular responses of hMS cell population. In conclusion, initial cell adhesion and proliferation of hMSCs was significantly increased upon culture to rough surfaces and a significant induction of the osteogenic differentiation was observed when cells where cultured on the OSSEOTITE, HA, and the SLA implant surfaces; expression of a series of genes that regulate signal transduction, transcription, cell cycle, angiogenesis, cell adhesion, ECM formation and cell differentiation was found to be significantly upregulated on cells grown on rough surfaces as compared to smooth surfaces. In this way, it was evident that roughened implant surfaces modified by either additive or subtractive methods may enhance osteogenic responses.

Abbreviations

BMP

bone morphogenic protein

ECM

extracellular matrix

FGF

fibroblast growth factor

HA

hydroxyapaptite

hMSC

human bone marrow-derived mesenchymal cell

ICAM-1

intercellular adhesion molecule-1

IGF

insulin-like growth factor

MTT

3-[4,5dimethylthiazol-2-yl]-d,5 disphenyltetrarolium bromide

PDGFA

platelet-derived growth factor A

SLA

sand-blasted/acid-etched

SMO

smooth surface

TGF-β

transforming growth factor-beta

tRNA

total RNA