Cobalt-based materials are widely used for coronary stents, as well as bone and joint implants. However, their use is associated with high corrosion incidence. Titanium alloys, by contrast, are more biocompatible owing to the formation of a relatively inactive titanium oxide (TiO2) layer on their surface. This study was aimed at improving Co28Cr6Mo alloy cytocompatibility via sol–gel TiO2 coating to reduce metal corrosion and metal ion release. Owing to their role in inflammation and tissue remodelling around an implant, endothelial cells present a suitable in vitro model for testing the biological response to metallic materials. Primary human endothelial cells seeded on Co28Cr6Mo showed a stress phenotype with numerous F-actin fibres absent on TiO2-coated material. To investigate this effect at the gene expression level, cDNA microarray analysis of in total 1301 genes was performed. Compared with control cells, 247 genes were expressed differentially in the cells grown on Co28Cr6Mo, among them genes involved in proliferation, oxidative stress response and inflammation. TiO2 coating reduced the effects of Co28Cr6Mo on gene expression in endothelial cells, with only 34 genes being differentially expressed. Quantitative real-time polymerase chain reaction and protein analysis confirmed microarray data for selected genes. The effect of TiO2 coating can be, in part, attributed to the reduced release of Co2+, because addition of CoCl2 resulted in similar cellular responses. TiO2 coating of cobalt-based materials, therefore, could be used in the production of cobalt-based devices for cardiovascular and skeletal applications to reduce the adverse effects of metal corrosion products and to improve the response of endothelial and other cell types.
Metals and their alloys are the materials of choice for bone and joint implants, dental prostheses, coronary stents, etc. In particular, CoCr-based alloys are used as implant material owing to their mechanical strength and stiffness and high wear resistance . However, CoCr-based alloys are often associated with a higher corrosion rate and release of toxic Co and Cr ions. Thus, a higher serum concentration of Co and Cr can be detected in patients with failed total joint implants . Notably, an elevated serum metal concentration was observed even in individuals with well-functioning CoCr-based prosthesis . Moreover, the concentration of metal ions rises locally in the tissues surrounding the implant. Hence, Co concentration reached up to 0.9 mM in the periprosthetic tissue . CoCr corrosion products were shown to induce inflammatory responses both in vitro and in vivo [5,6]. Metal ion release in the course of anodic partial reaction of corrosion is accompanied by formation of reactive oxygen species (ROS) as intermediate products of cathodic partial reaction. ROS have been also shown to modulate inflammatory response  and have been discussed as one of the reasons of implant aseptic loosening .
Titanium-based alloys, on the other hand, are generally considered to be more biocompatible. This is attributed to the formation of the relatively inactive TiO2 layer on the material surface, which reduces corrosion rate. Thus, direct comparison of CoCrMo and Ti6Al4V in vivo showed no adverse reactions to the Ti-based material, whereas the CoCr-based alloy induced a persistent inflammatory response . Nonetheless, titanium is also detected in the serum of patients bearing titanium-based implants (twofold increase compared with control group) . Furthermore, cathodic polarization of Ti6Al4V induced adverse cellular effects in vitro . However, titanium ions are less toxic than cobalt or chromium ions, and a direct comparison of corrosion behaviour of CoCrMo and Ti6Al4V alloy showed a higher corrosion resistance of the latter .
Successful integration of a metal implant depends on the wound-healing processes around an implant early after surgery and on the interaction with the surrounding tissues throughout the lifetime of the prosthesis. Endothelial cells play a crucial role in tissue regeneration around an implanted metal device. These cells are involved in inflammatory reactions during wound-healing and in the formation of new blood vessels (the process of angiogenesis) in the peri-implant tissue. Increased amounts of metal ions and particles in peri-implant tissue can affect this early response of endothelial cells, as well as the integrity of mature vessels around an implant during its lifetime, therefore undermining its stability . Additionally, endothelial cells throughout the entire body are exposed to metal corrosion products in the blood stream that might contribute to a systemic toxicity associated with metal implants. Altogether, this makes endothelial cells an optimal model system to test cytocompatibility.
In vitro as well as in vivo studies have shown that CoCrMo alloys affect the functionality of endothelial cells [9,12]. Similarly, corrosion products of CoCr-based materials, i.e. Co2+ and Co particles, induced pro-inflammatory activation of endothelial cells [6,13,14] and inhibited angiogenic activity of endothelial cells in vitro . A prolonged inflammatory response and alteration of blood vessel formation induced by corrosion products could delay CoCr-implant integration and contribute to systemic toxicity and implant failure. Reducing corrosion and metal release could, therefore, improve the biological response to CoCr-based materials. We have previously shown that TiO2 coating of Co28Cr6Mo alloy significantly reduced metal ion release [16,17].
The aim of this study was to compare the uncoated and TiO2-coated Co28Cr6Mo materials using primary human endothelial cells as a test system for cytocompatibility. Cell phenotype and cDNA microarray analysis showed an improved response of endothelial cells to TiO2-coated Co28Cr6Mo alloy in vitro compared with uncoated material. Thus, the TiO2 coating showed a marked downregulation of important genes involved in oxidative stress and pro-inflammatory responses. CoCl2 induced similar cellular responses as the bulk Co28Cr6Mo alloy, suggesting the shielding effect of TiO2 layer against metal corrosion products.
2. Material and methods
2.1. TiO2 coating of Co28Cr6Mo alloy
Co28Cr6Mo (ESKA Implants, Lübeck) discs with a diameter of 14.5 mm and 1.5 mm thickness were used as substrates. Disc surfaces were polished to the roughness values of Ra = 0.014 ± 0.009 µm. TiO2 coating (140 nm thick) was produced by dip-coating via sol–gel process as previously described [16,17]. Briefly, the sol for dipping consisted of tetrabutoxytitanate and n-butanol with a molar ratio of 1 : 33. The exposure time in the sol was 20 s. The samples were pulled out at a speed of 1.7 mm s−1. After exposing the discs to air for 1 h, they were tempered at 500°C. Afterwards, the samples were autoclaved in water and eventually air-dried under sterile conditions.
2.2. Cell culture
Human dermal microvascular endothelial cells (HDMEC) were isolated from juvenile foreskin and cultured as described previously . The material was collected according to the ethical guidelines of the University Medical Center Mainz and the Ethics Committee of the state of Rheinland-Pfalz. At passage four, HDMEC were seeded on Co28Cr6Mo, on TiO2-coated Co28Cr6Mo or, as physiological (minimally activated) control, on fibronectin-coated glass as a 300 µl drop (45 000 cells cm−2). After cell adhesion (4 h), the drop was aspirated, and fresh medium was added. Alternatively, cell suspension was applied to flexiPerm silicon rings (Greiner) placed on top of the materials. The cells were cultured 30 h for mRNA analysis and 40 h for protein analysis. For CoCl2 treatment, the cells were seeded on fibronectin-coated cell culture plastic.
2.3. Immunofluorescent staining
The cells were fixed in 3.7% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 5 min. CD31 and F-actin were stained with FITC-conjugated anti-CD31 antibody (Chemicon, dilution 1 : 40) and Alexa Fluor 594-labelled phalloidin (Invitrogen, dilution 1 : 40), respectively, in PBS–1% bovine serum albumin (BSA) for 1 h at room temperature. CD31 signal was further intensified with Alexa Fluor 488-labelled secondary antibody (Invitrogen, dilution 1 : 400) in PBS–1% BSA for 45 min. Finally, the nuclei were stained with 1 µg ml−1 Hoechst 33342 for 5 min, and the samples covered with mounting medium.
2.4. cDNA microarrays
RNA was isolated from HDMEC grown on Co28Cr6Mo-based materials or fibronectin-coated glass specimens for 30 h using RNeasy mini kit (Qiagen) with additional DNase I digestion step. RNA concentration and quality were controlled, and corresponding RNA samples from three different donors were pooled together. Microarray analysis was performed by Miltenyi Biotec genomics service. PIQOR skin microarray containing probes for 1301 genes was used. Experimental and control RNAs were amplified and labelled with Cy5 and Cy3, respectively, and hybridized to the same array. Fluorescence intensities were then compared to obtain relative gene expression. Genes that were upregulated 1.7-fold or downregulated 0.58-fold were considered differentially expressed. The data are presented as an average expression ratio (fold of control) and coefficient of variation (in %), calculated from four different spots for the same gene on the same array.
2.5. Quantitative real-time polymerase chain reaction
RNA was isolated 30 h after cell seeding using RNeasy micro kit (Qiagen) and used for reverse transcription with Omniscript kit (Qiagen). Quantitative real-time polymerase chain reaction (RT-PCR) was performed using SYBR green DNA binding fluorescent dye. cDNA obtained from 3.75 ng RNA, and 12.5 pmol of each 3′ and 5′ primers were used per reaction. The volume was adjusted to 12.5 µl with H2O, and 12.5 µl Power SYBR green (Applied Biosystems) was added to each reaction. Following primers were used: Ki67, forward primer: 5′-TCCTTTGGTGGGCACCTAAGACCTG-3′; Ki67, reverse primer: 5′-TGATGGTTGAGGTCGTTCCTTGATG-3′; haem oxygenase-1 (HO-1), forward primer: 5′-ATGACACCAAGGACCAGAGC-3′; HO-1, reverse primer: 5′-GGGCAGAATCTTGCACTTTG-3′; cyclooxygenase II (COX II), forward primer: 5′-TGCATTCTTTGCCCAGCACT-3′; COX II, reverse primer: 5′-AAAGGCGCAGTTTACGCTGT-3′; RPL13A, forward primer: 5′-CCTGGAGGAGAAGAGGAAAGAGA-3′; RPL13A, reverse primer: 5′-TCCGTAGCCTCATGAGCTGTT-3′. PCR amplification was carried out in Applied Biosystems 7300 real-time PCR system (95°C for 10 min; 40 cycles: 95°C for 15 s, 60°C for 1 min). Specificity of the PCR product was controlled with the melting curve. Relative gene expression and 95% confidence intervals (CIs) were calculated with Rest2009 software using RPL13A as endogenous control. Amplification efficiency was calculated with LinRegPCR software.
2.6. Ki67 staining and cell number quantification
Ki67 was stained in fixed and permeabilized cells as described above using anti-Ki67 primary (Dako, dilution 1 : 40) and Alexa Fluor 488-conjugated secondary antibody (Invitrogen, dilution 1 : 400). Nuclei were stained with Hoechst 33342 solution (1 µg ml−1) for 5 min. Nine digital microscopic images were randomly taken for each sample. The number of Ki67-positive or total number of nuclei on an image was quantified with ImageTool software. Ki67 expression was defined as a percentage of Ki67-positive to the total number of nuclei. The number of nuclei on different materials was used to estimate the relative cell number.
2.7. Oxidative stress assays
ROS formation, glutathione (GSH) amount and superoxide dismutase (SOD) activity were measured with 2′,7′-dichlorodihydro-fluoresceine-diacetate, monochlorobimane and Superoxide Dismutase Activity Kit II (Calbiochem), respectively, as previously described . GSH and SOD were normalized to protein concentration.
2.8. Enzyme-linked immunosorbent assay
HO-1 and COX II were assayed in cell lysates with Human HO-1 ELISA Kit (Assay Design) and Human Cyclooxygenase II Enzyme Immunometric Assay Kit (Assay Design), respectively, following the manufacturer's protocol. HO-1 and COX II amount was determined from the standard curves run in parallel with recombinant HO-1 and COX II standards and normalized to the protein concentration of the cell lysates.
2.9. Protein quantification
Protein concentration in cell lysates was measured with NanoOrange Protein Quantification Kit (Invitrogen) according to the manufacturer's protocol.
2.10. Statistical analysis
All experiments were repeated at least three times, and the results are presented as mean ± standard deviation (s.d.) or mean ± 95% CI. Statistical difference between experimental groups was analysed with the paired t-test in Microsoft Excel software. Normal distribution of the data was confirmed with the Shapiro–Wilk test. Statistical analysis of quantitative RT-PCR data was carried out with pairwise fixed reallocation randomization test in Rest2009 software.
To study the effects of TiO2 coating, the F-actin cytoskeleton and distribution of CD31 of HDMEC grown on uncoated and TiO2-coated Co28Cr6Mo was compared with control cells. Under standard conditions, F-actin was concentrated in the so-called peripheral actin rings typical for endothelial cells in monolayer (figure 1a). On the other hand, HDMEC on Co28Cr6Mo alloy displayed an altered appearance of the actin cytoskeleton. Thus, an increase in F-actin stress fibres was observed in the cytoplasm of HDMEC in contact with Co28Cr6Mo alloy (figure 1b). The CD31 pattern within the intercellular contacts was also altered in HDMEC on Co28Cr6Mo (figure 1e) compared with HDMEC grown under standard tissue culture conditions (figure 1d), and showed a more irregular, discontinuous distribution. TiO2 coating of Co28Cr6Mo, on the other hand, resulted in a more normal appearance of HDMEC, with F-actin and CD31 distribution resembling the staining pattern displayed by control endothelial cells (figure 1c,f).
To identify important genes associated with these different effects on endothelial cells between uncoated and TiO2-coated Co28Cr6Mo, cDNA microarray analysis was performed 30 h after seeding HDMEC on the biomaterials. Growth on Co28Cr6Mo alloy induced considerable changes in the gene expression pattern in HDMEC compared with the physiological control conditions (glass). Out of 1301 genes studied, 247 were differentially expressed (125 genes were up- and 122 genes downregulated). TiO2 coating of Co28Cr6Mo, by contrast, markedly reduced the number of differentially expressed genes. Under these conditions, only 34 genes were differentially expressed (13 were up- and 21 downregulated) when compared with HDMEC grown in standard conditions (table 1 and figure 2; electronic supplementary material, S1). Notably, differential expression patterns in HDMEC on uncoated Co28Cr6Mo and TiO2-coated Co28Cr6Mo overlapped almost completely: of the 13 genes upregulated on TiO2-coated Co28Cr6Mo alloy, 12 were also upregulated on uncoated Co28Cr6Mo and of the 21 downregulated genes 20 overlapped with downregulated genes in HDMEC on Co28Cr6Mo.
To understand whether the effects of TiO2 coating are due to the sealing of Co28Cr6Mo or are caused by the TiO2 layer itself, the same coating was performed on Ti6Al4V and glass samples. In contrast to Co28Cr6Mo alloy, only 29 genes were differentially expressed in HDMEC grown on Ti6Al4V (figure 2) compared with the cells grown under standard conditions (on glass). Importantly, a maximum of only twofold regulation in gene expression was observed on Ti6Al4V. TiO2 coating of Ti6Al4V did not prominently change gene expression pattern. Thus, 32 genes were differentially expressed on TiO2-coated Ti6Al4V (figure 2). Furthermore, TiO2 coating did not induce significant changes in gene expression on glass. Only three genes were differentially expressed on TiO2-coated compared with uncoated glass (figure 2). Therefore, the effect of TiO2 coating was the strongest on Co28Cr6Mo alloy and could be attributed to attenuation of Co28Cr6Mo-induced changes in gene expression.
Among the genes differentially expressed in HDMEC on Co28Cr6Mo compared with control cells, few functional groups could be identified. In one group, mRNA from a number of cell-cycle-related genes was regulated (table 1, group 1). Here, the proliferation markers Ki67, proliferating cell nuclear antigen (PCNA), replication factor C (RCF) and replication protein A (RPA), a number of cyclins (A2, B2, D1, E1), cyclin-dependent kinases 1 and 2 (CDK1 and CDK2) were downregulated in HDMEC on Co28Cr6Mo. These mRNAs were in general also downregulated in the cells on TiO2-coated Co28Cr6Mo, but to a lesser extent than on uncoated material. A few CDK inhibitors (p21, p57 and p16) were, in turn, upregulated in the cells in contact with Co28Cr6Mo and were not regulated on TiO2-coated alloy. These data point to the reduced proliferation and possible cell cycle block in HDMEC grown on Co28Cr6Mo. However, some of the cyclins and CDK inhibitors (cyclin G2, p27 and p18) did not match the regulation pattern observed for the majority of the genes.
To verify the effect of Co28Cr6Mo on proliferation, cell number and Ki67 expression were compared in HDMEC on the different materials. Lower cell numbers were detected on both Co28Cr6Mo and Co28Cr6Mo–TiO2 materials compared with the control (figure 3a). There was, however, no significant difference in the cell number between metal surfaces. Quantitative RT-PCR confirmed microarray data and showed lower Ki67 mRNA expression in HDMEC on both alloys compared with the control cells (figure 3b). This also correlated with the lower number of Ki67-positive cells (figure 3c). Ki67 expression in HDMEC on TiO2-coated Co28Cr6Mo was slightly higher than on uncoated Co28Cr6Mo, in accordance with the microarray data.
HDMEC grown on Co28Cr6Mo also showed expression changes in oxidative stress response genes (table 1, group 2). Various genes for antioxidant enzymes (SOD1, SOD2, glutathione peroxidase 3 (GPX3), glutathione-S-transferase 12 (GST12)) were upregulated, whereas some genes (GPX1 and GSTT2) were downregulated in HDMEC on Co28Cr6Mo compared with the control conditions. Importantly, the expression of these genes did not differ in HDMEC on TiO2-coated Co28Cr6Mo compared with the control. Disturbance of regulation of antioxidant enzymes could indicate induction of oxidative stress in HDMEC on Co28Cr6Mo. To study whether the signs of oxidative stress could be detected biochemically, GSH content and SOD activity, both indicators of cellular antioxidant potential, were quantified in HDMEC on different growth substrates. GSH content in HDMEC on both metallic materials was lower than under control conditions (figure 4a). Notably, GSH amount was higher on TiO2-coated Co28Cr6Mo than on uncoated Co28Cr6Mo. Furthermore, HDMEC grown on Co28Cr6Mo had lower SOD activity compared with control conditions (figure 4b). Another sign of oxidative stress, lower SOD activity, however, did not correlate with the microarray data on SOD1 and SOD2 expression. By contrast, SOD activity in HDMEC on TiO2-coated Co28Cr6Mo did not differ from control and was higher than on uncoated material.
A number of mRNAs for heat shock proteins (HSPs) and their accessory proteins were highly upregulated in HDMEC on Co28Cr6Mo alloy (table 1, group 3). Two of these, HSP70B’ and HSP70.1 were increased 131- and 23-fold, respectively. Importantly, most of these genes did not show changes in regulation when Co28Cr6Mo was coated with a TiO2 layer. The expression of mRNA for another stress-related protein, HO-1, was also strongly induced in HDMEC on Co28Cr6Mo and reduced on TiO2-coated material (table 1, group 3 and figure 5a). This correlated with the results observed at the protein level: HO-1 was upregulated 10-fold in HDMEC on Co28Cr6Mo (figure 5b), whereas the induction was only around 3.5-fold on TiO2-coated alloy.
The mRNA expression of the pro-inflammatory molecules, interleukin-8 (IL-8) and COX II, was also differentially expressed on Co28Cr6Mo alloy compared with controls (table 1, group 4). The expression of both mRNAs was strongly increased on Co28Cr6Mo and was similar to the control level in HDMEC on TiO2-coated Co28Cr6Mo. The results for COX II were verified at mRNA (figure 5c) and protein levels (figure 5d). Thus, COX II protein expression was more than twofold higher in HDMEC on Co28Cr6Mo compared with control conditions. Moreover, TiO2 coating significantly decreased protein expression level. Similar effect at mRNA level was observed for another pro-inflammatory marker: monocyte chemotactic protein-1 (MCP-1, data not shown).
Metal ion release from the bulk material could be one of the factors causing stress induction in HDMEC grown on Co28Cr6Mo. To test this, the effects of Co2+ release were simulated by HDMEC treatment with CoCl2. CoCl2 in the concentrations that have been detected in peri-implant tissue (0.7 mM) induced similar effects compared with the bulk Co28Cr6Mo. Thus, CoCl2 treatment downregulated Ki67 expression in HDMEC (figure 6a). Additionally, HDMEC treated with CoCl2 also exhibited the signs of oxidative stress, such as intercellular ROS formation (figure 6b) and decreased GSH amount (figure 6c). Similar to Co28Cr6Mo material CoCl2 upregulated COX II protein level (figure 6d). Therefore, Co2+ affected proliferation, oxidative stress and inflammatory responses in HDMEC in a manner similar to the bulk Co28Cr6Mo.
Corrosion of CoCr-based implants and release of metal ions can cause adverse reactions ranging from acute inflammation and allergy to systemic toxicity. The peri-implant reaction often results in aseptic loosening and eventual need for revision surgery . Material modifications with the aim of minimizing corrosion could improve the biological response to CoCr-based materials, while retaining their beneficial mechanical properties. In our previous reports, sol–gel coating with TiO2 was shown to reduce the release of Co, Cr and Mo ions from Co28Cr6Mo alloy at least twofold in PBS and cell culture medium [16,17]. Therefore, this study was undertaken to examine whether TiO2 coating and the resulting reduced corrosion and metal ion release improve Co28Cr6Mo cytocompatibility. Because Co2+ release from Co28Cr6Mo was the highest, CoCl2 was additionally used to simulate metal ion release during Co28Cr6Mo corrosion. Endothelial cells were used as a model system owing to their role in wound-healing around an implant, and their sensitivity to cytotoxic and pro-inflammatory stimuli in vitro. We have shown previously  that growth on Co28Cr6Mo alloy induced a pro-inflammatory phenotype in endothelial cells, manifested by a redistribution of F-actin and intercellular contact proteins (CD31, VE-cadherin). Importantly, CoCl2 similarly affected F-actin and CD31 distribution in endothelial cells, indicating that Co2+ released from the bulk material may at least, in part, explain the adverse effects of Co28Cr6Mo. By contrast, endothelial cells on commercially pure Ti showed a phenotype similar to control conditions . In line with these observations, the results from the current study demonstrated that TiO2 coating markedly reduced the changes caused by uncoated, bulk Co28Cr6Mo: F-actin and CD31 patterns resembled the control cells. To determine the biological basis for the protective effect of TiO2 coating, an extensive gene expression study was carried out and validated in the case of selected genes by quantitative RT-PCR. In addition, comparisons at protein level were also carried out. Co28Cr6Mo alloy was shown to induce expression changes of several important genes involved in cell cycle regulation, stress response and inflammation. Of practical significance for the clinical use of such metallic implants is the abrogation of this disturbed gene regulation by coating with TiO2.
4.1. Proliferation and cell cycle
A number of genes involved in cell cycle regulation were differentially expressed in endothelial cells on Co28Cr6Mo. These included replication accessory proteins, cyclins, cyclin-dependent kinases and CDK inhibitors. Cell cycle progression is driven by activation and inactivation of CDKs (through phosphorylation and dephosphorylation of CDKs and interaction with cyclin subunits), and expression and degradation of cyclins and inhibitory proteins that collectively regulate the transition to the next phase of the cycle . Downregulation of CDKs and cyclins and simultaneous upregulation of CDK inhibitors indicate a possible cell cycle block in endothelial cells grown on Co28Cr6Mo alloy. The CDK inhibitors induced on Co28Cr6Mo are known to target mostly cyclin D/CDK4/6 complexes, therefore inducing G1 block in the cell cycle . Contrary to other cyclins, cyclin G2 was upregulated in HDMEC on Co28Cr6Mo alloy. However, cyclin G2 is an atypical cyclin that has been shown to induce cell cycle arrest in the G1/S phase , which fits the pattern of cell cycle gene regulation on Co28Cr6Mo. Additionally, in contrast to other CDK inhibitors, p18 was downregulated, whereas p27 was not regulated in the cells grown on Co28Cr6Mo. Although both p18 and p27 are involved in G1 cell cycle arrest and are known to cooperate in tumour suppression , it is possible that not all CDK inhibitors are involved in the growth arrest induced by Co28Cr6Mo alloy.
In agreement with these results, reduced expression of Ki67 at both mRNA and protein level indicated the decreased proliferation rate of endothelial cells on Co28Cr6Mo. Ki67 is widely used as a proliferative marker, because its levels increase during S phase . Ki67 expression correlated with the cell number, which was lower on Co28Cr6Mo compared with the control. Importantly, CoCl2 treatment also induced cell-cycle-related changes in endothelial cells, i.e. reduced cell number, decreased Ki67 expression and p21 upregulation (data not shown), thus supporting the hypothesis that Co2+ release plays a central mechanistic role in the effects that Co28Cr6Mo exhibited on cell proliferation.
4.2. Oxidative stress response
Endothelial cells grown on Co28Cr6Mo clearly showed signs of oxidative stress, with a number of genes involved in antioxidant defence exhibiting changes in regulation patterns. In line with this, the amount of non-enzymatic antioxidant GSH in HDMEC on Co28Cr6Mo alloy was lower than in the controls. Depletion of the GSH level could result from prolonged oxidative stress leading to excessive GSH oxidation . Metal ions can induce ROS formation via Fenton and Haber–Weiss reactions. Hence, in this study, CoCl2 induced intercellular ROS production and similarly decreased cellular GSH amount. Therefore, permanent exposure of the cells on Co28Cr6Mo to Co2+ is the most likely cause of oxidative stress. To adapt to the stress conditions, cells regulate gene expression of antioxidant enzymes [26,27]. mRNA expression of both SOD1 and SOD2 was elevated in endothelial cells on Co28Cr6Mo compared with the controls. Interestingly however, the total SOD activity was reduced in these cells. Differences in the expression of SOD mRNAs and SOD activity could possibly be explained by a delay in SOD upregulation at the protein level or enzyme oxidation and deactivation after prolonged oxidative stress .
mRNAs for a number of heat shock proteins were also upregulated in HDMEC grown on Co28Cr6Mo compared with the control. HSPs are molecular chaperones involved in the restoration of unfolded and misfolded proteins. Metal ions and ROS are known to induce changes in protein conformation, protein misfolding and HSP induction [24,29]. HSP70.1 and HSP105, which are highly induced on Co28Cr6Mo alloy, were also observed to be regulated in keratinocytes and endothelial cells exposed to CoCl2 [30,31]. Importantly, these effects can be ROS-dependent, because ROS have also been shown to induce expression of HSP70.1 .
Expression of HO-1, another oxidative stress-related protein, was increased in endothelial cells exposed to Co28Cr6Mo. Cobalt as well as ROS and H2O2 have been shown to induce HO-1 expression [33,34]. HO-1 catalyses degradation of haem to biliverdin, accompanied by production of the important signalling molecules carbon monoxide and iron . The products of HO-1 catalysis exert a broad range of cellular effects, including cell cycle control (through e.g. p21 upregulation ), as well as regulation of antioxidant and inflammatory responses . Owing to multiple functions of HO-1, it is difficult to determine which of its effects prevails in endothelial cells exposed to Co28Cr6Mo. However, together with oxidative stress and heat shock response, HO-1 upregulation indicates stress induction in the cells on Co28Cr6Mo, possibly mediated by released metal ions and metal-induced ROS.
In addition to induction of an inflammatory phenotype (F-actin and CD31 distribution), the pro-inflammatory markers, IL-8, MCP-1 and COX II, were upregulated in endothelial cells on Co28Cr6Mo. Induction of IL-8 and MCP-1, cytokines involved in the recruitment of leucocytes to sites of inflammation, is in agreement with the reports of CoCl2 inducing IL-8 and MCP-1 release by endothelial cells [6,14]. Another regulated protein, COX II, is an inducible enzyme catalysing conversion of arachidonic acid to prostaglandin H2, a precursor for the synthesis of other prostanoids with inflammatory functions. Importantly, CoCl2 (this study and ), as well as intracellular ROS  and exogenous H2O2  have been shown to induce COX II expression. Interestingly, COX II has been implicated in IL-8 regulation [40,41], thus connecting both inflammatory molecules induced on Co28Cr6Mo. Altogether, the results point to pro-inflammatory activation of endothelial cells by Co28Cr6Mo via Co2+ release and ROS formation.
4.4. TiO2 coating
While Co28Cr6Mo induced expression changes in numerous genes involved in cell cycle, stress response and inflammation, TiO2 coating markedly reduced the number of differentially expressed genes. Importantly, the gene regulation patterns on different materials overlapped, i.e. TiO2 coating reduced the number of genes regulated in HDMEC by contact to Co28Cr6Mo, but did not seem to induce substantial changes in gene expression on its own. Additionally, TiO2 coating did not markedly influence gene expression on Ti6Al4V or glass. Therefore, we assume that the TiO2 layer has a sealing effect, reducing metal corrosion and metal ion release and therefore shielding the cells from metal ion- and ROS-induced changes in gene expression. However, this effect is incomplete, because a number of genes were still expressed differentially on TiO2-coated material compared with control conditions. Although the thickness of the TiO2 layer was 142 ± 18 nm , the coating did not result in a complete sealing. Nevertheless, the differentially expressed genes on TiO2-coated material were regulated to a much lesser extent than on uncoated Co28Cr6Mo. For example, the expression of CDKs and cyclins on TiO2-coated Co28Cr6Mo was lower than in the cells under control conditions, but higher than on uncoated metallic material. This correlated with Ki67 expression, which also showed only partial restoration of cell proliferation. By contrast, CDK inhibitors induced by Co28Cr6Mo were not regulated on TiO2-coated material. Furthermore, the changes in expression of stress- and inflammation-related genes could be reversed to almost control level by TiO2 coating. This protective effect of TiO2 layer was also observed at protein level (HO-1 and COX II). Importantly, no signs of oxidative stress were observed in endothelial cells, further pointing to the shielding effect of TiO2 against metal corrosion products.
Importantly, TiO2 coating of Co28Cr6Mo alloy can protect cells from the products of both anodic and cathodic partial reactions of corrosion. While anodic partial reaction results in metal oxidation and metal ion release, concurrent cathodic partial reaction leads to the reduction of oxygen at physiologic pH. The latter process is a source of ROS formed during intermediate reactions of oxygen reduction . Metal ions can serve as an additional source of ROS, formed in Fenton- and Haber–Weiss-type reactions. ROS, apart from their role in oxidative stress, have been shown to affect cell cycle regulation, heat shock response and pro-inflammatory activation, as discussed above. Therefore, protective effects of TiO2 coating, seen in this study, can be attributed to shielding from both metal ions and ROS. Additionally, TiO2 coating can reduce electronic conductivity at the surface of Co28Cr6Mo alloy. Electronic conductivity at the passive layer of metallic materials can induce conformational changes in adsorbed proteins . Such proteins could mediate the adverse effects of Co28Cr6Mo alloy on attached cells. By reducing electronic conductivity and protein conformation changes, TiO2 coating could further improve cytocompatibility of Co28Cr6Mo.
In our earlier studies, TiO2 coating of Co28Cr6Mo had a protective effect on osteoblasts, increasing metabolic activity and cell number [16,17]. Other reports of positive effects of TiO2 coating can be found in the literature. Sol–gel coating of Ti1.5Al25V effectively shielded fibroblasts from toxic vanadium and significantly enhanced cell viability . TiO2 coating of acid-etched titanium surfaces via slow-rate sputter deposition of molten TiO2 nanoparticles improved attachment, proliferation and extracellular matrix deposition by osteoblast and muscle cells [45,46]. In in vivo experiments, direct attachment of soft tissues after subcutaneous implantation in rats was improved by sol–gel-derived nanoporous TiO2 coating . Furthermore, higher marginal alveolar bone was observed after implantation of TiO2-coated dental implants in dogs .
The new data presented here, together with data from the cited literature, indicate that reactions induced by Co28Cr6Mo in endothelial cells appear to reflect the in vivo response to corrosion and wear products of metal implants. Wear particles and metal ions have been shown to induce pro-inflammatory responses. Thus, CoCr wear debris induced acute inflammatory cell recruitment and expression of pro-inflammatory cytokines in a rodent air-pouch model . Tissues exposed to CoCr particles also showed signs of increased oxidative stress in the same model .
Co28Cr6Mo and its corrosion products could also affect the angiogenic potential of endothelial cells around implanted material. An inhibition of tube formation by CoCl2 was demonstrated in an in vitro angiogenesis assay . However, contradictory effects were shown in vivo. Thus, CoCl2 administration actually induced vessel formation in bladder and kidney of rats [51,52]. The effects of Co2+ on angiogenesis could be concentration- and tissue-dependent and could also differ in the presence of particulate matter, as would be the case for wear and corrosion debris. More importantly, metal corrosion products can affect existing vessels. Hence, in addition to persistent inflammatory response, CoCrMo alloy decreased capillary perfusion in striated muscle in the hamster dorsal skinfold chamber, and simultaneously increased microvascular permeability and leucocyte extravasation .
The impact of Co28Cr6Mo and its corrosion products on peri-implant tissue could determine the overall response to the metal material. Prolonged inflammation and effects on vessel integrity and permeability can undermine implant integration and long-term stability. By reducing metal corrosion and ion release TiO2 coating could reduce adverse effects of cobalt-based materials and increase the lifetime of an implant. However, because TiO2 layer would be exposed to wear processes after implantation, its stability should be tested in further studies. Nonetheless, increased cytocompatibility of TiO2-coated Co28Cr6Mo, as shown by the molecular changes operative in human microvascular endothelial cells in vitro, endorses the possible use of such surface modification to improve the in vivo response to CoCr-based implants.
TiO2 coating attenuated the stress response and disturbed gene expression in endothelial cells induced by Co28Cr6Mo alloy. This abrogation is most likely a protective effect of the TiO2 layer by reducing metal corrosion and ion release. Consequently, this sealing method holds promise in improving local as well as whole body responses to cobalt-based biomaterials. This study presents new molecular data at mRNA and protein levels, which provide a biological basis for the protective effect of TiO2 coating of cobalt alloys and could open up new possibilities for metal device development.
cDNA microarray data are available as electronic supplementary material.
The work was supported by the Deutsche Forschungsgemeinschaft (Priority Programme Biosystem 322 1100, KI 601/1–4) and funds from the State of Rhineland-Palatinate. The authors thank Susanne Barth for her excellent technical support.
- Received May 10, 2013.
- Accepted June 13, 2013.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.