Comparison of Candidate Materials for a Synthetic Osteo-Odonto Keratoprosthesis Device

Abstract

Purpose.: Osteo-odonto keratoprosthesis is one of the most successful forms of keratoprosthesis surgery for end-stage corneal and ocular surface disease. There is a lack of detailed comparison studies on the biocompatibilities of different materials used in keratoprosthesis. The aim of this investigation was to compare synthetic bioinert materials used for keratoprosthesis surgery with hydroxyapatite (HA) as a reference.

Methods.: Test materials were sintered titanium oxide (TiO₂), aluminum oxide (Al₂O₃), and yttria-stabilized zirconia (YSZ) with density >95%. Bacterial adhesion on the substrates was evaluated using scanning electron microscopy and the spread plate method. Surface properties of the implant discs were scanned using optical microscopy. Human keratocyte attachment and proliferation rates were assessed by cell counting and MTT assay at different time points. Morphologic analysis and immunoblotting were used to evaluate focal adhesion formation, whereas cell adhesion force was measured with a multimode atomic force microscope.

Results.: The authors found that bacterial adhesion on the TiO₂, Al₂O₃, and YSZ surfaces were lower than that on HA substrates. TiO₂ significantly promoted keratocyte proliferation and viability compared with HA, Al₂O_3, and YSZ. Immunofluorescent imaging analyses, immunoblotting, and atomic force microscope measurement revealed that TiO₂ surfaces enhanced cell spreading and cell adhesion compared with HA and Al₂O₃.

Conclusions.: TiO₂ is the most suitable replacement candidate for use as skirt material because it enhanced cell functions and reduced bacterial adhesion. This would, in turn, enhance tissue integration and reduce device failure rates during keratoprosthesis surgery.

Corneal opacification is the second major cause, after cataract, of impaired vision and blindness in developing countries. ¹ The cornea may be damaged by infection, disease, or trauma, and while though transplantation may be suitable for patients with minimal ocular surface destruction, this procedure is greatly limited because of the lack of donor corneas and eye banking capabilities in most developing countries in Asia and Africa. ² An artificial cornea or keratoprosthesis (KPro) procedure thus represents a viable alternative for treatment in many of these patients. The osteo-odonto keratoprosthesis (OOKP) operation is a complex, two-stage surgical procedure that can be used to restore vision in end-stage cases of severe corneal, ocular surface, and dry eye states. The OOKP operation would be useful for patients with Stevens-Johnson syndrome, ocular cicatricial pemphigoid, and chemical or thermal injuries for whom conventional corneal transplantation or the Boston KPro would be contraindicated. ^2,3  

The original OOKP technique, involving two stages, was developed by Strampelli. ⁴ The whole surgery procedure involves the removal of an autologous canine tooth and adjacent bone, which are used to support a polymethyl methacrylate (PMMA) optical cylinder subsequently implanted beneath the oculi muscle and sutured to the cornea, allowing a clear visual axis for vision restoration. The robust nature of full-thickness buccal mucosa coupled with a strong bone-tooth support for the PMMA optic results in an extremely low device extrusion rate spanning decades. In addition, an excellent visual result can be attained through the PMMA optic, generally limited only by previous optic nerve or retinal damage. ^2,5 Although highly successful, OOKP surgery is extremely complex and is limited by the availability of any healthy canine teeth in adults or in children who have yet to develop adult dentition. Furthermore, the use of alternative autologous sources for skirt material (e.g., tibia) has been associated with high rates of bone resorption. ⁶ Dimensional limitations of an autologous lamina also restrict the design and size of the OOKP optical device to either a 3.5- or a 4-mm opening, significantly restricting the field of view. ⁷ These issues have led to an increased search for alternatives, and one promising solution to this is the development of novel biomaterials to be used as skirt material. This would enhance the use of the OOKP procedure and would make it accessible to a wider group of patients. 

Numerous synthetic keratoprosthesis devices have been developed as total replacements of the cornea for the treatment of corneal blindness. ^{8 –10} Most keratoprosthesis devices are designed with a clear central optic and an annular porous surrounding or skirt element. The major challenge facing these keratoprosthesis devices is that a sufficient amount of cellular invasion is needed to anchor the implant firmly in place because stable tissue integration is crucial for the survival of a keratoprosthesis. The potential biomaterial selected should fulfill the following criteria: good biostability, biocompatibility, integration, antibacterial, and immunologic acceptance. Various bioactive materials and bioinert materials have been developed for keratoprosthesis skirt substitution. Bioactive materials, including glass ceramic ¹¹ and hydroxyapatite (HA) ceramic, ¹² have both shown good tissue integration. However, serious complications such as aqueous leakage, retroprosthetic membrane formation, and endophthalmitis are frequently observed because of the degradation of these bioactive materials. Bioinert materials, such as platinum, titanium, and aluminum alloys, are alternative materials that do not react with host tissues when implanted. Titanium or titanium alloy, aluminum, or their modified compounds have been used as keratoprosthesis skirt materials. ^{13 –16} However, there is little published work examining the comparative benefits of each of these materials as skirts for OOKP. Previously, we conducted an in vitro biocompatibility comparison of bioactive materials including glass ceramic, HA, HEMA, and PTFE. We found HA to show superior biocompatibility compared with the other bioactive materials. ¹⁷ These materials were chosen because they represent the core materials used in current keratoprosthesis device skirts (i.e., HA) in OOKP. In this study, we have compared the biocompatibilities of bioinert materials, including titanium, aluminum, and zirconium alloys with bioactive HA. Recent studies have begun to recognize the importance of the tissue-implant interface during keratoprosthesis surgery. Investigations of these tissue-material interactions at the cellular level are vital to design keratoprosthesis devices with the combined properties of an optimized design, biocompatibility, enhanced implant integration, and reduced risk of bacterial infection. 

The main objective of our study was to use an in vitro model to assess a panel of potential keratoprosthesis skirt materials to determine the most suitable material for the development of an artificial keratoprosthesis skirt. Titanium dioxide (TiO₂), aluminum oxide (Al₂O₃), and yttria-stabilized zirconia (YSZ) were assessed with comparison to HA as a control. Antibacterial properties were evaluated using Staphylococcus aureus, which is commonly associated with corneal implant infection. ¹⁸ Comparative studies on cell adhesion, viability and cytokines were conducted with human corneal keratocytes. Immunofluorescence analysis using phalloidin staining was used for morphologic evaluation. Biophysical adhesion force of keratocytes on various material substrates was evaluated with atomic force microscope (AFM). The development of a fully synthetic ceramic replacement would reduce the complexity of the surgical procedure because a large proportion involves tooth harvesting and shaping of the tooth root to accept the optical cylinder. In addition, a fully synthetic replacement would remove the necessity of a donor, hence making keratoprosthesis treatment more widely available. Without the physical constraints of tooth dimension, this replacement would allow for enlarged optical cylinders to provide a wider field of vision. 

Materials and Methods

Synthesis

HA discs were produced by spark plasma sintering HA powder at a nominal pressure of 50 MPa and sintering at 1000°C. TiO₂ discs were produced by pressing the powders at a pressure of 38 MPa and sintering at 1100°C. Al₂O₃ discs were produced by pressing the powders at a pressure of 50 MPa and sintering at 1500°C. YSZ discs were produced by pressing the ZrO₂ − 3 mol% Y₂O₃ powders at a pressure of 50 MPa and sintering at 1550°C (YSZ1550) or 1600°C (YSZ1600). The diameters of all discs were 10 mm, and the thicknesses were 2 mm for HA and TiO₂ specimens and 2.5 mm for Al₂O₃ and YSZ specimens. The individual discs were stored at room temperature and sterilized by pure ethanol washing and ultraviolet irradiation for at least 1 hour before experimentation. 

Antibacterial Properties and Surface Plot

Staphylococcus aureus, a Gram-positive bacterial strain, was used for the antibacterial assays. Yeast-dextrose broth (containing 10 g/L peptone, 8 g/L beef extract, 5 g/L sodium chloride, 5 g/L glucose, and 3 g/L yeast extract) was used as the growth medium. The bacteria were incubated overnight at 37°C with agitation. An aliquot of culture was then added to the yeast-dextrose broth and incubated for another 8 to 10 hours at 37°C. The bacteria containing broth was centrifuged at 2700 rpm for 10 minutes. After the removal of the supernatant, the cells were washed with phosphate-buffered saline (PBS) and were resuspended in PBS at a concentration of 10⁶ cells/mL (calibrated in terms of colony-forming units using the spread plate method ¹⁹ ). One milliliter of the bacterial suspension was then added to each disc in a 24-well plate and was incubated for 6 hours at 37°C. Discs were rinsed with PBS to remove nonadherent bacteria before fixing with 3% glutaraldehyde. After serial-step dehydration with ethanol for 10 minutes, each disc was coated with platinum, and the surfaces of the discs were imaged using a scanning electron microscope (SEM; model 5600LV; JEOL, Tokyo, Japan). 

Quantification of bacterial adhesion was also carried out. After 6 hours of incubation, the discs were removed with sterile forceps and gently washed with PBS. The discs were then placed in broth, and the bacteria retained on substrates were dislodged by mild ultrasonication for 2 minutes in a 100-W ultrasonic bath operating at a nominal frequency of 50 Hz followed by rapid vortex mixing (10 seconds). Serial 10-fold dilutions were performed, and viable counts were estimated using a previously described spread plate method. ¹⁹ The number of bacteria on each disc surface were counted and expressed relative to the surface area of the disc (number of bacteria/cm²). All experiments were performed in triplicate with three discs, and the mean values were calculated. 

Surfaces of the materials were scanned by light microscope (Axioplan 4.7 software; Zeiss, Thornwood, NY). The images were three dimensionally plotted by ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html), and surface roughness was evaluated by a gray dynamic scaling method. ^20,21  

Human Corneal Keratocytes Preparation

Human corneal stroma was collected from donor corneal tissue. The epithelial and endothelial layers of the cornea were surgically removed, and the stroma was then cut into small pieces and digested overnight in 0.2% type I collagenase solution at 37°C. After a brief wash with PBS, a single-cell suspension was collected and cultivated in DMEM containing 10% fetal calf serum. The cells were incubated at 37°C in a humidified atmosphere of 5% CO₂, with the growth medium changed every 2 to 3 days. The cells took approximately 1 to 2 weeks to grow to confluence, and the purity of keratocytes was verified by analysis of cell morphology and immunocytochemistry of vimentin staining. Keratocytes at passage 3 to 5 were used for the experiments. All studies related to human tissues were approved by the Institutional Review Board of Singapore Eye Research Institute and Singapore National Eye Center. 

Cell Adhesion and MTT Analysis

For cell adhesion analysis, discs of each test material were placed in a standard 24-well culture dish. Corneal keratocytes (1 × 10⁴/cm²) were seeded onto the surface of each disc. At 4 hours, 24 hours, or 7 days after seeding, substrate discs were gently washed with PBS to remove the nonadherent cells before fixation with 4% paraformaldehyde and subsequently were stained with phalloidin-rhodamine (Invitrogen, Burlington, ON, Canada) and DAPI (Vector Laboratories, Burlington, ON, Canada). The disc surface was then scanned with a Zeiss fluorescence microscope, and the numbers of adherent cells from five randomly chosen scanning areas were counted. All testing was repeated in triplicate for each time point. For cell proliferation and cytotoxicity assays, keratocytes (1.5 × 10³/cm²) in a 30-μL droplet were seeded on the surface of the substrate discs. At 24 hours or 7 days after seeding, 0.5 mg/mL MTT reagent was added to the medium for formazan developing. Dimethyl sulfoxide was used to dissolve the formazan, and the colored product was transferred into a 96-well plate and placed in a microtiter plate reader. Optical density was read at 560 nm with a reference filter of 620 nm and was normalized by the ratio of implant surface area to culture dish surface area for statistical analysis. Culture medium with cells growing on the culture dish was read as a positive control of MTT reading, and culture medium without cells growing was read as a negative control. 

Cell Morphology

Human corneal keratocytes were fixed with 4% paraformaldehyde after seeding on the disc surface for 24 hours. After permeabilizing with 0.2% Triton in the PBS buffer, keratocytes were stained with phalloidin-rhodamine (Invitrogen) and were counterstained with DAPI. Antibodies used were mouse anti-gelsolin (Sigma, St. Louis, MO), mouse anti-vinculin (Sigma), and mouse anti–β₁-integrin (Abcam, Cambridge, UK), whereas the secondary antibody was FITC-conjugated goat anti–mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA). 

Western Blot Analysis

Lysates from corneal keratocytes were obtained by homogenization in detergent-based RIPA buffer with proteinase inhibitor. Ten milligrams of total cellular protein was run on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane (Hybond-C; Amersham, Buckinghamshire, UK) by electroblotting. After blocking with 5% skimmed milk, the membranes were incubated with anti-gelsolin (1:1000), anti-vinculin (1:1000), anti–β₁-integrin (1:500), and anti-GAPDH (1:2000; Santa Cruz Biotechnology) antibodies, followed by the incubation of secondary antibodies (donkey anti–mouse IgG or donkey anti–rabbit IgG, horseradish peroxidase linked, 1:2000 Amersham). The intensity of the immunoblots was analyzed by image analysis software (LabWorks 4.5; UVP, Upland, CA). 

AFM Measurement

Human keratocytes were seeded onto HA, TiO₂, and Al₂O₃ discs at a density of 3 × 10⁴/cm² for 2 hours, then fixed with 2.5% glutaraldehyde for 2 hours at room temperature. After washing with PBS to remove the fixation solution, the samples were air dried with nitrogen gas for 1 minute immediately before AFM measurement. 

Atomic force microscopy (Bioscope II; Veeco, Plainview, NY) was performed in the contact mode in air (room temperature). The substrate discs carrying cells were mounted onto the XY stage of the microscope, and the integral video camera was used to locate the cells. Movement of the probe was controlled by the microscope. The friction difference caused by the movement of cells on the substrate surface caused the cantilever to bend and twist laterally, and this signal was measured by monitoring the C-D signal from the photodiode array. The data type, called friction versus moving distance, was recorded. In our experimental settings, the real-time friction curve was screen-captured from the monitor, and an AFM probe (model RTESP) was used to detect friction difference (scan range, 50 μm; scan rate, 0.1 Hz; tip velocity, 10 μm/s). Total readings from 50 cells in triplicate on each kind of substrate were recorded. 

Statistical Analysis

At least three samples per time point for each experimental condition were used. The results are reported as mean ± SE and were assessed statistically using one-way analysis of variance (ANOVA) post hoc Tukey test. Statistical significance was accepted at P < 0.05. 

Results

Antibacterial Properties

To determine the intrinsic bacterial resistance properties of different materials, bacterial adhesion assays were performed at 6 hours of exposure. The SEM images of the discs after immersion in a Staphylococcus aureus bacteria suspension are shown in Figure 1. On comparison among all samples, the HA specimens are found to be more amenable to bacterial adhesion (Fig. 1Aa), whereas TiO₂, Al₂O₃, YSZ1550, and YSZ1600 specimens are more resistant to bacterial adhesion (Figs. 1Ab–e). Results of the spread plate method (Fig. 1B) are in agreement with the trend observed with SEM investigations in which HA had a statistically significant higher bacteria adherence rate. 

Figure 1.

 

Bacterial adhesion on test materials. (A) Representative SEM images of bacteria adhesion on HA (a), TiO₂ (b), Al₂O₃ (c), YSZ1550 (d), and YSZ1600 (e). (B) Spread plate counting of attached bacteria. The number of bacteria cells on HA disc was arbitrarily defined as 1. *P < 0.05, one-way ANOVA between HA and TiO₂.

Surface Characterization

To compare the surface properties of the materials and their effects on cells, surface profiles of the five materials were taken. The surface topography of each material was observed using confocal laser scanning microscopy. Images are represented in 2D (Figs. 2a–e) and 3D (Figs. 2a′–e′). Surface topography of TiO₂ and Al₂O₃ discs are shown in Figures 2b, 2b′, 2c, and 2c′. The presence of fine grainlike particles in the TiO₂ and Al₂O₃ substrates provided for increased submicron to nanometer roughness. However, YSZ1550 and YSZ1600 substrates revealed coarse grainlike particle with a greater percentage of interparticulate voids providing for larger scale surface roughness (Figs. 2d, 2d′, 2e, 2e′). Surface roughness was measured by a dynamic scaling graph (Figs. 2a″–e″). The dynamic surface gray values of the HA specimen were distributed from 34 to 38 (Fig. 2a″). Dynamic surface gray values of TiO₂ and Al₂O₃ discs were distributed from 30 to 45 and from 30 to 55, respectively (Figs. 2b″, 2c″). Dynamic surface gray values of YSZ1550 and YSZ1600 were distributed from 40 to 85 and from 40 to 90, respectively (Figs. 2d″, 2e″). Each measurement was repeated three times from three randomly chosen scanning areas. The gray value range indicated that the TiO₂ disc surfaces were rougher than HA discs and smoother than Al₂O₃ and YSZ discs. 

Figure 2.

 

Surface topography of test substrate discs. 2D surface plot (a–e), 3D surface plot (a′–e′), and dynamic gray value distribution (a″–e″) of HA (a, a′, a″), TiO₂ (b, b′, b″), Al₂O₃ (c, c′, c″), YSZ1550 (d, d′, d″), and YSZ1600 (e, e′, e″). Scale bar, 20 μm.

Cell Adhesion and Viability Analysis

To study cell attachment on the materials, human keratocytes were seeded on the substrates. At 4 hours, there were no significant differences (P = 0.25) in cell counts on the five materials. Cell attachment progressed steadily over 24 hours of culture on the discs, as shown in Figure 3A. The number of cells attached on the HA, TiO₂, and Al₂O₃ specimens (223 ± 13, 198 ± 28 and 231 ± 19 per scan area, respectively) were markedly increased, and there were higher cell numbers attached compared with the YSZ1550 and YSZ1600 specimens (165 ± 17 and 160 ± 25 per scan area, respectively; P = 0.008; Fig. 3B). At day 7, the cell numbers were approximately 1.5-fold higher on the HA, TiO₂, and Al₂O₃ specimens (259 ± 7, 265 ± 10, and 281 ± 7 per scan area) compared with the YSZ1550 and YSZ1600 specimens (176 ± 20 and 205 ± 6 per scan area; P = 0.001; Fig. 3B). These findings indicated that HA, TiO₂, and Al₂O₃ promoted greater human corneal keratocyte cell attachment and cell proliferation than YSZ1550 and YSZ1600. 

Figure 3.

 

Cell attachment on test materials at different time points. (A) Fluorescent confocal microscopy images of keratocytes on testing materials at 4 hours (a–e), 24 hours (a′–e′), and 7 days (a″–e″). HA (a, a′, a″), TiO₂ (b, b′, b″), Al₂O₃ (c, c′, c″), YSZ1550 (d, d′, d″), and YSZ1600 (e, e′, e″). Red: phalloidin-rhodamine; blue: DAPI. Scale bar, 10 μm. (B) Cell counts per scan area on testing materials. Data are presented as mean ± SE.

An MTT assay was performed to compare the cell proliferation rate and cytotoxicity of the attached keratocytes at 24 hours (Fig. 4A) and 7 days (Fig. 4B) after seeding. The MTT reading of keratocytes on a normal culture dish was defined as 1. At 24 hours after seeding, relative MTT readings for TiO₂ and HA discs were 1.02 ± 0.03 and 0.78 ± 0.07, respectively (P = 0.03). At 7 days after seeding, relative MTT readings for TiO₂ and HA discs were 0.99 ± 0.06 and 0.75 ± 0.04, respectively (P = 0.03). Hence, at both 24 hours and 7 days after seeding, keratocytes on TiO₂ discs showed a significantly higher proliferation rate and fewer cytotoxicity effects than HA discs. Meanwhile, keratocytes on Al₂O₃ discs showed a significantly lower proliferation rate and higher cytotoxicity effect than that on HA seeded discs (relative MTT readings for Al₂O₃ discs were 0.41 ± 0.02 at 24 hours and 0.47 ± 0.07 at 7 days; P = 0.001 and 0.005, respectively). 

Figure 4.

 

Cell viability measured by MTT assay at 24 hours (A) and 7 days (B). OD absorption readings of cells on normal culture dish are arbitrarily defined as 1. *P < 0.05, **P < 0.05, one-way ANOVA between HA and TiO₂.

Morphologic Studies

Within 4 hours of seeding, most keratocytes adhered on the surfaces of the materials, as shown in Figures 3 Aa–e. At day 1, most of the cells on HA, TiO₂, and Al₂O₃ specimens displayed a spindle-shaped or elongated morphology (Figs. 3Aa′–c′), whereas the cells on YSZ1550 and YSZ1600 exhibited a stunted, less elongated morphology (Figs. 3Aa′–c′). At day 7, fine and long filopodia seen on HA, TiO₂, and Al₂O₃ discs formed a cellular cluster in parallel to the specimen surface, and the cells were interlaced and confluent. 

Focal Adhesion Protein Quantification

Qualitative observation of immunofluorescence-labeled cells suggested expression of gelsolin on HA = TiO₂ > Al₂O₃ (Figs. 5a–c). Focal adhesions were more prominent on TiO₂ than on HA and Al₂O₃ (Figs. 5a′–c′). Expression of β₁-integrin was less prominent on Al₂O₃ than on HA and TiO₂ (Figs. 5a″–c″). To further quantify focal protein expression in keratocytes, immunoblotting was conducted (Fig. 6A), and the blot intensity was analyzed (Fig. 6B). Averaged blot intensities of each protein sample were collected from HA discs and were arbitrarily defined as 1. The expression levels of gelsolin, vinculin, and β₁-integrin on TiO₂ discs were higher or similar to HA discs (relative blot intensities were 1.02 ± 0.01, 1.55 ± 0.09, and 0.98 ± 0.01; P = 0.36, P = 0.01, and P = 0.08, respectively). Whereas the expression level of gelsolin, vinculin, and β₁-integrin on Al₂O₃ discs were significantly lower than on HA discs (relative blot intensities were 0.83 ± 0.02, 0.39 ± 0.05, and 0.87 ± 0.02, respectively; P = 0.002, P = 0.02, and P = 0.001, respectively). These observations were in agreement with our immunostaining studies (Fig. 5). 

Figure 5.

 

Gelsolin, vinculin, β₁-integrin, and F-actin distribution on substrate materials at 24 hours. Confocal images show the representative cells on HA (a, a′, a″), TiO₂ (b, b′, b″) and Al₂O₃ (c, c′, c″). Green: gelsolin (a–c), vinculin (a′–c′), and β₁-integrin (a″–c″); red: phalloidin-rhodamine; blue: DAPI. Scale bar, 20 μm.

Figure 6.

 

Immunoblots of gelsolin, vinculin, β₁-integrin, and GAPDH of corneal keratocytes on substrate materials. (A) Typical bands obtained after immunoblotting. (B) Relative expression level of proteins. Blot intensities were normalized with GAPDH inputs, and the blot intensity of HA was arbitrarily defined as 1.

Measurement of Cell Adhesion Force by AFM

As illustrated in Figure 7A, AFM measurements of the interaction between an individual keratocyte and the substrate involved a series of four steps. First, the cantilever was lowered adjacent to a cell and was positioned above the substrate. The cantilever then engaged the cell with subsequent lateral distortion of the tip. It then applied force to move the engaged cell so it would be pushed away or rolled up, followed by subsequent retraction to the start position. During this process, the AFM continuously monitored the deflection of the cantilever to report on the interaction between the keratocyte and the various materials. 

Figure 7.

 

AFM measurements of the keratocyte /substrate interaction. (A) Schematic representation of a typical cycle of the AFM measurement. (B) Typical friction-distance trace. *Maximum y value of first peak was arbitrarily chosen to compare cell-substrate adhesion. (C) Histograms of statistical results of cell adhesion (friction) force. **P < 0.05, one-way ANOVA between HA and Al₂O₃.

A typical AFM friction-distance record of the individual cell and substrate was presented in Figure 7B. In this figure, the x-axis plots the probe moving distance and the y-axis plots the friction voltage, which is in linear relationship with friction force (Ff = uN; Ff, friction; u, coefficient; N, force). At the start of the force measurement, when the cantilever was several microns away from the cell, there was no strain on the cantilever. During this distance, which was marked by a dashed line in the trace of Figure 7B, the force was zero. After probe-cell contact, the cantilever was bent and marked by an up-slope in the trace image. Further expansion of the cantilever by the cell was marked by a continuous plateau line in the trace image. Subsequently, the cells were detached away from the probe, which was marked by a down-right slope, and the friction force gradually returned to zero. The cell detachment process typically involved a series of rupture events that resulted in one or several rapid jumps in the force curve. ²²  

To compare the difference in adhesion properties of keratocytes on various substrates, the maximum y value of the first upward slope was recorded for statistical analysis. The average friction voltage (y value) of keratocytes on TiO₂ was 9.11 ± 0.19, which was not significantly different from that on HA (9.26 ± 0.13; P = 0.61). However, the average friction voltage of keratocytes on Al₂O₃ (8.03 ± 0.32) was significantly lower than that on HA and TiO₂ discs (P = 0.001 and P = 0.0005, respectively). 

Discussion

It is vital to ascertain and optimize prospective biomaterials to be used as synthetic OOKP tooth substitution before device implantation. HA, TiO₂, Al₂O₃, and YSZ are all well established biomaterials used in biomedical engineering. HA is similar to the principal mineral constituent of bone and tooth and has been shown to promote the integration of orthopedic and dental implants. ²³ Clinically, HA has been used in keratoprosthesis surgery (as the tooth complex) and orbital implants. It is considered as the gold standard with respect to ocular tissue integration. ^12,24 We therefore used HA in our experiments as the standard with which to compare other prospective materials. Although widely used, HA is susceptible to degradation, and long-term implant stability is often compromised. ^7,25 There is a need to develop chemically stable alternatives, especially during instances of infection, when there is a decrease in pH in the peri-implant area. Thus, today titanium is widely used for dental and orthopedic implants and for keratoprosthesis skirts because of its physical, chemical, and mechanical properties, ^13,26,27 making it is more stable for long-term indwelling medical devices. We wanted to assess the in vitro integration of a panel of prospective keratoprosthesis skirt materials (HA, TiO₂, Al₂O₃, and YSZ). To achieve this we studied bacterial and corneal stroma keratocytes, cell adhesion, cell viability, cell focal adhesion protein expression, and surface topography. We have been able to show TiO₂ to be the most suitable candidate for the further development of a fully synthetic keratoprosthesis device. 

The major components of an OOKP lamina are bone and dentine. They have been shown to have good biointegration and to exhibit remodeling because its structure is constantly adjusted in response to external stress. Bioactive HA was developed as a substitute for bone because it is compatible with bone marrow cells and is resorbable by osteoclasts. ²⁸ However, once implanted into the corneal stroma, bone, dentine, and HA can undergo resorption by a multitude of factors, such as low-grade inflammation and cytokine release from activated macrophages. ^{29 –31} Clinically, the resorption of bone, dentine, and HA are potential serious long-term complications for OOKP surgery. Resorption may lead to device tilt, extrusion, infection, and subsequent implantation failure. Therefore, more stable materials are required to replace the current OOKP lamina. TiO₂, Al₂O₃, and YSZ are all well-known non-bioactive metal materials with excellent biocompatibility and high resistance to corrosion. ^{32 –34} Therefore, our studies concentrated on the evaluation of these materials as possible keratoprosthesis skirts. 

The response of bio-implants to bacteria, in terms of bacterial adhesion and colonization, is of crucial importance to materials in biomedical engineering. Initial bacterial adhesion is one of the most important factors that govern the pathogenesis of infections in biomaterials. ^35,36 Especially in the case of percutaneous devices—dental and ophthalmological implants that penetrate soft skin and connective tissues, respectively—this may lead to a breach in the barriers to exogenous pathogens. The ocular surface is enveloped by a commensal layer of bacteria, which may be altered in ocular disease states or in OOKP cases in which the ocular surface is resurfaced by oral mucosa. Hence, there is a potential for organism attachment during device implantation. The first and foremost critical step in the infection process taking place after surgery is bacterial adhesion on biomaterial surfaces. ³⁷ No medical biomaterials are regarded as totally inert to bacteria. ³⁸ For this reason, it is important to select materials that are “inhibitive” to the adhesion of bacteria on the implant surface. Our results showed that TiO₂, Al₂O₃, YSZ1550, and YSZ1600 specimens were relatively more resistant to bacterial adhesion, whereas HA specimens were more amenable to bacterial growth (Fig. 1). We chose to use the 6-hour exposure period because this early postimplantation period is considered the most important and “decisive period” during which prevention of bacterial adhesion is critical to the efficacy and long-term success of an implant. ^39,40  

The quality of tissue integration could be improved by optimization of the implant surface because the texture of the surface (rough/smooth, porous/nonporous) has a direct effect on cell adhesion, viability, and proliferation. ^41,42 Mammalian cells exhibit a decreased proliferation and a more differentiated phenotype on rough surfaces. ^43,44 Moreover, because of their high hydrophilic property and surface energy, nanostructured rough surfaces are able to upregulate angiogenic factors and adhesion molecule expression levels of surface attached cells, which are crucial factors in inflammation control and healing. ⁴⁵ Hence, ideal candidate materials should promote corneal keratocyte cell attachment and proliferation and not stimulate excessive inflammation. Our data indicate that TiO₂ is better than HA, Al₂O₃, and YSZ to improve corneal stroma biointegration. In our studies, the TiO₂ surface showed a smaller dynamic gray value range (30–45) than YSZ1550 (40–85) and YSZ1600 (40–90), which is in accordance with our findings that there was a higher cell adhesion number and cell proliferation rate on TiO₂ than YSZ1550 and YSZ1600 (Figs. 3, 4). Although the surface roughness of TiO₂ is slightly higher than HA (Fig. 2), the cell attachment and proliferation rates to the TiO₂ surface are still higher than HA (Figs. 3, 4), which means that other factors are also involved in regulating keratocyte attachment and proliferation. Chemical composition and elasticity of the substrates have been reported to influence cellular adhesion, migration, and proliferation. ^{46 –48}  

Cellular integration was assessed using primary human corneal keratocytes because these are the cells that are required to integrate with potential skirt materials. The strength of cell adhesion is multifactorial and dependent on focal adhesion formation, cell membrane density of integrin receptors, and cell skeleton organization. ^{49 –51} Therefore, we examined the expression of a variety of cytoskeletal proteins and adhesion complexes on the three best candidate materials: HA, TiO₂, and Al₂O₃. Our results indicated that primary human corneal fibroblast on TiO₂ surfaces showed higher expression levels of gelsolin, vinculin, and β₁-integrin than those on HA and Al₂O₃ specimen surfaces (Fig. 5). Immunoblotting studies (Fig. 6) supported our immunohistochemical analysis with respect to the greater expression of cell adhesive protein on TiO₂ compared with HA and Al₂O₃. The increased expression of β₁-integrin from keratocytes on TiO₂ will allow greater interaction with the extracellular matrix. This interaction was supported by the focal adhesion plaques associated with increased vinculin expression. We can hypothesize that this would be suggestive of greater cellular integration on TiO₂. 

Many elaborate methods, based on shear stress monitoring have been devised to measure the attachment of cells to substrates. Centrifugal force or liquid flow–based methods are designed to measure the force required to remove cells from a substrate. However, cell cohesion, rather than cell substrate adhesion, is often measured during the mechanical disruption of cells on metal surfaces. ⁵² AFM may be used as a sensitive force spectrometer for research in cell adhesion. ^53,54 In our studies, we used AFM in a frictional mode to monitor single-cell detachment from the substrate and quantified the friction force between keratocytes and various substrates. We found that the average adhesion strength of keratocytes on TiO₂ substrate is close to that of HA discs, whereas the average keratocyte adhesion strength on Al₂O₃ discs was lower than that on HA (Fig. 7). These observations are in agreement with the immunostaining and Western blot analysis where more β₁-integrin, vinculin and gelsolin proteins were expressed by keratocytes on HA and TiO₂ discs than on Al₂O₃ discs (Figs. 5, 6). The cell detachment force obtained from AFM corresponds largely to the binding activities of focal adhesions and cell skeleton proteins (e.g., vinculin, β₁-integrin, and gelsolin). This finding is also in agreement with previous reports that interfacial strength is attributed largely to the detachment that occurs between focal adhesion proteins, e.g., α/β integrin, and extracellular matrix proteins. ⁵⁵  

Our study aimed to compare potential biomaterials to be used as skirts for an OOKP keratoprosthesis design, with respect to cellular adhesion/proliferation and bacteria adhesion-binding studies. Using an in vitro model and a corneal keratocyte culture system, TiO₂ was selected as the most suitable candidate for the development of a fully synthetic keratoprosthesis device. We have shown that TiO₂ was superior as a starting material with respect to keratocyte cell integration, cell adhesion, and antibacterial properties compared with the conventional gold standard HA. Further in vivo studies will be needed to corroborate our in vitro findings. 

Footnotes

 Supported by Singapore Eye Research Institute Health Research Endowment Fund Grant R684.

Footnotes

 Disclosure: X.W. Tan, None; A.P.P. Perera, None; A. Tan, None; D. Tan, None; K.A. Khor, None; R.W. Beuerman, None; J.S. Mehta, None