The Role of Titanium Surface Microtopography on Adhesion, Proliferation, Transformation, and Matrix Deposition of Corneal Cells

Abstract

Purpose: Titanium (Ti) is an excellent implantable biomaterial that can be further enhanced by surface topography optimization. Despite numerous data from orthopedics and dentistry, the effect of Ti surface topography on ocular cells is still poorly understood. In light of the recent adaptation of Ti in the Boston Keratoprosthesis artificial cornea, we attempted to perform an extended evaluation of the effect of Ti surface topography on corneal cell adhesion, proliferation, cytotoxicity, transformation, and matrix deposition.

Methods: Different surface topographies were generated on medical grade Ti-6Al-4V-ELI (extra-low interstitial), with linearly increased roughness (polished to grit blasted). Biological response was evaluated in vitro using human corneal limbal epithelial (HCLE) cells, stromal fibroblasts (HCF), and endothelial cells (HCEnC).

Results: None of the Ti surface topographies caused cytotoxicity to any of the three corneal cell types. However, rough Ti surface inhibited HCLE and HCF cell adhesion and proliferation, while HCEnC proliferation was unaffected. Long-term experiments with HCF revealed that rough Ti surface with R_a (the arithmetic average of the profile height from the mean line) ≥ 1.15 μm suppressed HCF focal adhesion kinase phosphorylation, changed fibroblast morphology, and caused less aligned and reduced deposition of collagen matrix as compared to smooth Ti (R_a ≤ 0.08 μm). In the presence of transforming growth factor β1 (TGFβ1) stimulation, rough Ti inhibited alpha-smooth muscle actin (α-SMA) expression and collagen deposition, leading to decreased myofibroblast transformation and disorganization of the collagen fibrils as compared to smooth Ti.

Conclusions: This study suggests that Ti surface topography regulates corneal cell behavior in a tissue-dependent manner that varies across the corneal strata. Contrary to the accepted paradigm, smooth surface topography can enhance cell adhesion and proliferation and increase matrix deposition by corneal cells.

Medical grade titanium (Ti) is the most widely used biomaterial in medicine due to its inertness and its excellent biocompatibility and physical/mechanical properties.¹ It is estimated that 1 million Ti dental implantations are performed per year,² and its use continues to grow in orthopedics.³ Titanium is also gaining more attention from ophthalmology, as this material is now widely used to fabricate glaucoma devices, such as iStent and the new Boston Keratoprosthesis (B-KPro) artificial cornea. 

The B-KPro is a meso-implant for the treatment of corneal blindness in patients not amenable to standard penetrating keratoplasty.^4,5 The design is composed of a “collar-button” front plate with a stem, a Ti back plate, and a donor carrier cornea clipped between the front and back plates and sutured to the host cornea (Fig. 1). This design bridges all layers of the cornea, each uniquely populated by different cell populations, with potentially different responses to prosthetic materials. The Ti surface is roughened by grit blasting primarily to remove machining residues and reduce vision-impairing light reflectivity. But it is unclear how Ti surface roughness affects the behaviors of different corneal cells and potential biointegration between host cornea and the implant. 

Figure 1

 

(A) Configuration of the Boston Keratoprosthesis with the optical stem, donor corneal graft, titanium back plate, and titanium locking ring. (B) A patient implanted with B-KPro.

Despite the excellent visual outcomes with the B-KPro device, several complications can impede its performance. One of the most important complications is the formation of retroprosthetic membrane (RPM) on the posterior surface of the back plate.^6,7 Retroprosthetic membrane formation increases the risk of implant extrusion and can be caused by migrating corneal fibroblasts from the corneal host–donor junction to the posterior surface of the Ti back plate where they differentiate into myofibroblasts.^8,9 The incorporation of Ti as a new material for the B-KPro back plate provides the opportunity to modify surface topography and modulate cellular responses. Generating a surface topography that does not promote matrix deposition and cell proliferation on the posterior surface of the back plate, while allowing cellular functions on the anterior surface, might help to inhibit RPM formation and improve the “biological interaction” between the device with the donor cornea. 

Studies using osteoblasts, gingival fibroblasts, and keratinocytes have shown that modifying Ti surface topography can be employed to modulate cell adhesion and matrix deposition, avoid unwanted biological responses, and reduce implant complications.¹⁰ A limited degree of surface roughness and specific textures in implants were shown to be more favorable to osteoblasts.^11–13 Despite these findings in other cell types, there are no published studies addressing the effect of Ti surface topography on corneal cells in the context of ocular implants. Therefore, there is a compelling need for studying the potentially beneficial and adverse effects of Ti surface topography in corneal tissue. Our goal is to develop effective surface modification strategies to reduce postoperative complications such as RPMs and infection^8,14 and improve potential biointegration between B-KPro and corneal tissue. 

To our knowledge, this is the first study to assess the role of macro- and microscale Ti surface topography modulation on the biological response of human corneal epithelial, stromal, and endothelial cells. We use four different grades of topographic surface roughness, from polished to coarse, with linear increments, to study the influence of Ti surface topography in different corneal cells in vitro in order to reveal new potential improvements in the design of Ti ophthalmic implants. Furthermore, we expand our study on the primary human corneal fibroblast (HCF) response to Ti surface topography, since this cell type may interact directly with B-KPro Ti back plates in vivo.⁸ Transforming growth factor β1 has been shown to play a key role in transformation of corneal fibroblasts to myofibroblasts¹⁵ and in enhancing matrix adhesion¹⁶ and collagen production and deposition by myofibroblasts,¹⁷ and is commonly linked to corneal fibrosis.¹⁸ Thus, we also included TGFβ1 in our experiments to assess whether this growth factor can trigger different responses of HCF to Ti surface topography. 

Methods

Titanium Surface Modification and Characterization

Sandblasting of Titanium Disks.

Mechanical treatment with grit blasting is an effective technique to generate uniform surface roughness. It also provides the added benefit of removing residue and debris, which are generated during the milling/machining process of Ti. Shot peening utilizes different types and sizes of particles (alumina, glass, ceramic), which are blasted at high velocity at the Ti surface using a high-pressure air gun. The choice of particle type, size, shot velocity, and particle density defines the subsequent surface topography. 

Titanium disks were prepared using Ti-6Al-4V-ELI medical grade Ti (grade 23) 580 μm in thickness. Circular disks were machine cut to a 15.6-mm diameter, cleaned, and surface modified using grit blasting as follows: grade 0 (polished surface), grade 1 (grit blasted with ballotta XXL glass impact beads), grade 2 (grit blasted with 220 glass beads), or grade 3 (grit blasted with 40/60DM aluminium oxide beads). Blasting was performed at 27 kg of force, at 5-cm distance. This resulted in linear increments in surface roughness. All disks were cleaned and sterilized using the same protocol as for the commercially distributed B-KPro, a device approved for marketing by the U.S. Food and Drug Administration. 

Laser Scanning Confocal Microscope.

Noncontact surface topography analysis of the Ti disks was performed using the VK-X X200 laser scanning confocal microscope (Keyence, Itasca, IL, USA)^19,20 with a ×50 objective lens. Small beam spot 408-nm laser combined with a high z-axis resolution (0.5 nm) scale module and a 16-bit photomultiplier were used to obtain fully focused, high-precision surface profile and roughness data from the reflection light. Multiple wide-field regions (220 × 287 μm) of the Ti disks were scanned in XYZ directions, and profile roughness was determined by the arithmetic average of the profile height from the mean line (R_a), the sum of the largest peak height value and the largest valley depth value within the defined area (R_t), and the mean spacing of profile irregularities at the mean line of a profile (RS_m). Two-dimensional topographic maps were rendered using VK-X software. 

Contact Angle Measurement.

In order to assess surface energy of the different Ti grades, H₂O contact angle measurements were performed using the sessile drop technique.^21,22 Briefly, a drop of ultrapure H₂O (10 μL) was placed on the clean Ti surface, and the static contact angles of the air-water-substrate interface were captured by a computer-controlled charge-coupled device camera (XCD-V50; Sony, Park Ridge, NJ, USA) with green light-emitting diode background illumination. Static contact angles were calculated using drop shape Laplacian analysis (plug-in for ImageJ; National Institutes of Health, Bethesda, MD, USA). Water contact angles below 90° were regarded as hydrophilic and above 90° as hydrophobic. 

Cell Culture

In order to assess the biological responses of corneal cells to Ti surface topography, Ti disks were cultured separately with immortalized human corneal limbal epithelial cells (HCLE cells), HCF, and immortalized human corneal endothelial cells (HCEnC). All research adhered to the tenets of the Declaration of Helsinki. The development of the immortalized HCLE cell line and its culture technique were described previously.^23,24 Briefly, HCLE cells were originally derived from corneal donor rims and were subsequently immortalized by transfection with human telomerase reverse transcriptase (hTERT). The HCLE cells were maintained in keratinocyte serum-free medium (K-SFM; Invitrogen, Carlsbad, CA, USA) supplemented with 25 μg/mL bovine pituitary extract (BPE) and 0.2 ng/mL epidermal growth factor (EGF) and grown at 37°C in a 5% carbon dioxide atmosphere. Cells were passed to new flasks at 80% confluence. 

Human corneal fibroblasts were harvested from human cornea tissues as previously described.²⁵ Following scraping off of corneal epithelial and endothelial layers, corneal stroma explants were allowed to adhere to the bottom of culture wells and cultured ex vivo in Eagle's minimum essential medium (EMEM; American Type Culture Collection [ATCC], Manassas, VA, USA) supplemented with 10% fetal bovine serum (FBS; ATCC), 1% antibiotic/antimycotic (Sigma-Aldrich Corp., St. Louis, MO, USA). After 1 to 2 weeks of culture, fibroblasts derived from the explants were passaged. The cells were maintained in the same culture media and grown to 80% confluence before each subculture and were used in experiments at or before the third passage. 

The HCEnC cell line was established by Schmedt and colleagues²⁶ as previously described. Briefly, a uniform subpopulation of endothelial cells was derived from primary cells harvested from a 21-year-old male donor. Following transduction with hTERT, this subpopulation yielded highly hTERT-expressing cells (HCEnC). This cell line preserves proliferative activity, as well as corneal endothelial morphology, marker characteristics, and functionality.²⁶ The HCEnC were maintained in media containing OptiMEM-I (Invitrogen), 8% FBS (ATCC), EGF 5 ng/mL (Millipore, Billerica, MA, USA), pituitary extract 100 μg/mL (Alfa Aesar, Ward Hill, MA, USA), calcium chloride 200 mg/L, 0.08% chondroitin sulfate (Sigma-Aldrich Corp.), gentamicin 50 μg/mL (Gibco, Grand Island, NY, USA), and antibiotic/antimycotic (Sigma-Aldrich Corp.) solution diluted 1:100. Feeding media were changed every 2 days for all regular cell cultures, unless specified elsewhere. 

Characterization of Cell Growth on Ti.

Cell proliferation and cytotoxicity assays were performed as previously described.^27,28 Human corneal limbal epithelial cells, HCF, and HCEnC were assayed, and all experiments were performed in triplicate. Titanium disks (grades 0, 1, 2, 3) were placed in a 24-well plate (CoStar, Corning, NY, USA), completely covering the bottom surface of each well. Tissue culture plastic was used as a positive control surface. Corneal cells (2 × 10⁴) were seeded in each well and cultured in 1 mL feeding media until the prespecified time points were met (day 2, 3, or 5). Culture supernatants were collected every day and stored at −80°C for the cytotoxicity assay. Fresh medium was added to the cultures daily. 

Proliferation Assay.

Cell proliferation was evaluated using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA). Cells were gently washed three times using unsupplemented media in order to remove nonadherent cells. The number of viable adherent cells per well was evaluated by incubating the cells in 100 μL Cell Titer 96 AQueous One Solution Proliferation Assay with 0.5 mL unsupplemented media at 37°C for 60 minutes. A microplate reader (SpectraMax Plus 384; Molecular Devices, Sunnyvale, CA, USA) was employed to measure the absorbance of the developed chromogenic assay at 490 nm. Standard curves for each cell line were generated and culture medium was used to measure the background noise, which was then subtracted from the sample signal. 

Cytotoxicity Assay.

The amount of dying or injured cells was quantified by measuring the levels of lactate dehydrogenase (LDH) released by cells in the culture media at day 2, 3, or 5 of culture using the Cytotox 96 nonradioactive cytotoxicity assay (Promega). Relative amounts of LDH were evaluated by reading the absorbance of reacted solution at 490 nm with a microplate reader. Culture media background (negative) and LDH (positive) controls were also measured and used for sample signal correction. 

Corneal Fibroblast Response Assessment.

Primary HCF were cultured on Ti disks using EMEM. L-Ascorbic acid 2-phosphate trisodium salt (1 mM) (P-Asc; Wako Chemicals, Richmond, VA, USA) was added to media to promote collagen production.²⁵ In some experiments, 10 ng/mL human recombinant TGFβ1 (R&D Systems, Minneapolis, MN, USA) was also added to imitate the biological response of corneal wound healing following B-KPro implantation. Media were changed every 2 days. 

Short-term cell response was evaluated at 4 days using culture samples seeded at ∼10⁴/disk initial cell density. Long-term cell response was evaluated at 1.5 months using culture samples seeded at ∼10⁵/disk initial cell density. At the end of experiments, cells were fixed with 4% paraformaldehyde (PFA) and incubated with mouse monoclonal antibody against α-SMA (1:150; ab7817; Abcam, Cambridge, MA, USA), rabbit polyclonal antibody against collagen type I (1:150; ab34710; Abcam), or rabbit polyclonal antibody against collagen type V (1:150; NB120-7046; Novus, Littleton, CO, USA), and then treated with Alexa Fluor 647 conjugated secondary antibody, Alexa Fluor 488 conjugated phalloidin (Molecular Probes, Eugene, OR, USA) for F-actin expression, and 4′,6-diamidino-2-phenylindole (DAPI) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) to stain cell nuclei. 

Immunocytochemistry and Imaging.

Cells were fixed in 4% PFA, permeabilized with 0.2% Triton X-100 for 5 minutes, and incubated with 1% bovine serum albumin (BSA) for 1 hour at room temperature. Primary antibodies were diluted in 1% BSA and incubated with the cells overnight at 4°C. Secondary antibodies were incubated with the cells for 2 hours at room temperature. Phalloidin was diluted in PBS and incubated with the cells for 2 hours at room temperature. 

Titanium disks with attached cells were carefully lifted off from the wells and transferred onto glass slides. Mounting medium with DAPI was used together with glass cover slips for imaging using the Zeiss Axio Imager M2 (Zeiss, Oberkochen, Germany) fluorescence microscope with ×20 dry and ×63 oil immersion objective lenses. The thickness of cell-assembled matrix on the Ti disks was measured using the computer-controlled stage of the microscope by focusing the ×20 objective lens at the top and bottom surfaces of the matrix at four different quadrants of each disk. The z-axis resolution with the ×20 objective lens is 780 nm. The measured matrix thickness was in the order of 10 μm. Based on Nyquist theorem, this resolution is adequate to measure the matrix thickness. In addition, the step size of the microscope stage in the z-direction is 25 nm, well below the optical resolution. 

Total Collagen Amount Quantification.

At 45 days of HCF culture, the total amount of collagen deposition per Ti disk was quantified using a total collagen assay kit (MAK008; Sigma-Aldrich Corp.), which measured the amount of hydroxyproline in hydrolyzed lysates.^29,30 The assay was performed per manufacturer's protocol. Briefly, the cell-assembled matrix was released from the Ti disk and homogenized. Each cell lysate was mixed with an equal volume of HCl (6 N) and hydrolyzed in a 95°C dry bath overnight. Hydrolyzed supernatant (75 μL) was added to each 96-well and the plates were baked at 60°C until dry. Residues were incubated with the chromogenic assay reagents at 60°C for 90 minutes. Total collagen quantification was performed by measuring the absorbance of reacted solution at 560 nm using a microplate reader. Serial-diluted hydroxyproline was used as standard. All samples and standards were performed in triplicate. 

Image Analysis

Images of HCF attached on Ti disks were analyzed for collective cell directionality and cell morphology using the ImageJ software (version 1.49i; National Institutes of Health). Tiled images (×20) of HCF F-actin were analyzed for overall cell directionality with the Directionality plug-in in ImageJ. Morphologic measurement of individual cells was performed with the Analyze Particles module in ImageJ. At least 20 single cells were used for the morphologic analysis of each Ti grade. 

Protein Analysis by Western Blot

Cells were rinsed twice with ice-cold PBS (pH 7.4) and lysed for 30 minutes on ice with tissue protein extraction reagent (Thermo Scientific, Rockford, IL, USA). Protein (20 μg) was resolved by SDS 4% to 20% PAGE and transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). Blocking of membranes was performed in 5% BSA in Tris-buffered saline with Tween 20 (BSA-TBST) for 1 hour at room temperature followed by overnight incubation at 4°C with primary antibodies: anti α-SMA, 1:250, ab7817 (Abcam); anti-focal adhesion kinase phosphorylated tyrosine 397, 1:1000, ab81298 (Abcam); anti-total focal adhesion kinase, 1:1000, ab40794 (Abcam); anti-beta actin, 1:1000 (Bio-Rad); anti-Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH), 1:3000 (Santa Cruz Biotechnology). After washing three times with TBST, transfer membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1/1000), and signal intensity was measured using enhanced chemiluminescence (SuperSignal West-Dura extended duration substrate; Thermo Scientific) and scanned by Syngene G:BOX Digital ECL Room (Syngene; Frederick, MD, USA). 

Statistics

Quantitative results were reported as mean ± standard deviation. The normality of data was assessed by Shapiro-Wilk test. One-way and 2-way ANOVA were performed in data sets containing multiple variables using the R Studio (Boston, MA, USA), followed by Holm-Sidak pairwise multiple comparison test. Statistical significance was set at P < 0.05 (2-tail). 

Results

Surface Characterization of Titanium

Contact Angle.

In order to assess the effect of surface roughness in Ti hydrophilicity, water contact angles were measured using static sessile drop method as previously described.^21,22 All surface grades exhibited comparable water contact angles (grade 0: 58° ± 0.93°, grade 1: 64° ± 0.73°, grade 2: 63° ± 5.77°, grade 3: 59° ± 2.78°) and therefore were assumed as equally hydrophilic. 

Ti Surface Topography.

Microscale surface topography of Ti disks was evaluated by laser scanning confocal microscopy. Grade 0 (polished surface) had low surface roughness (Fig. 2A, R_a = 0.081 ± 0.008 μm; R_t = 0.744 ± 0.093 μm), with presence of linear, parallel grooves whose depths ranged from 0.4 to 0.6 μm (as maximum valley depth = 0.568 ± 0.082 μm) and distances between grooves in the range of 25 to 53 μm (as RS_m = 33.64 ± 6.41 μm). As a result of grit-blasting treatment, surface roughness increased linearly, with grade 1 the least rough (Fig. 2B, R_a = 0.182 ± 0.019 μm; R_t = 1.328 ± 0.15 μm; RS_m = 28.54 ± 3.52 μm), grade 2 surface with increased roughness (Fig. 2C, R_a = 0.618 ± 0.047 μm, R_t = 5.043 ± 0.901 μm, and RS_m = 26.96 ± 3.70 μm), and grade 3 surface with the roughest topography, largest average grit size, and broad spacing between surface peaks and valleys (Fig. 2D, R_a = 1.148 ± 0.185 μm; R_t = 8.149 ± 1.268 μm; RS_m = 36.34 ± 8.10 μm). Titanium grades 0 to 3 had linear increase in R_a and R_t surface roughness, with an average step of 0.48 μm (Fig. 2E). Grade 3 RS_m roughness was significantly higher compared to the other grades (P < 0.0001; ANOVA) (Fig. 2F). 

Figure 2

 

Titanium surface topography. (A–D) ×50 laser confocal microscopic images of the Ti surfaces, presented as two-dimensional scanning electron microscope (SEM)-like topography maps. (E) Linearity of surface roughness of the test Ti surfaces. Numbers shown are arithmetic average of the profile height from the mean line (R_a) and the mean spacing of profile irregularities at the mean line of a profile (RS_m) (n = 15).

Effect of Ti Surface Topography on Corneal Cell Proliferation and Cytotoxicity

HCLE Cells.

Polished Ti (grade 0) and culture plastic significantly promoted HCLE cell proliferation as compared to rough Ti (grades 1, 2, 3), which exhibited equal reduction in cell proliferation (P < 0.05), (Slope_{Grade 0} / Slope_{Grade 3} = 4.24) (Fig. 3A). None of the Ti grades had a cytotoxic effect on the cells, suggesting that the measured difference in cell proliferation was mainly attributed to change in cell metabolic activity rather than to accelerated cell death (Fig. 3B). The HCLE proliferation rates and Ti surface roughness (R_a) were negatively correlated (fitted with four-parameter logistic regression model, R = 0.9946, hill slope = −37.80), since grade 0 had much higher cell proliferation rate than grades 1 to 3 (P < 0.05, by 1-way ANOVA). Based on these results, we conclude that HCLE cell proliferation can be promoted by a smooth Ti surface. 

Figure 3

 

Results of corneal cell proliferation and cytotoxicity on Ti surfaces. (A) HCLE proliferation rate was significantly higher on polished Ti surface than on any tested grit-blasted surface. Proliferations on grades 1, 2, and 3 exhibited similar growth rates. (C) HCF proliferation was significantly delayed on grades 2 and 3 rough Ti surfaces (P < 0.05). The large gap in growth rates between plastic surface/grade 0/grade 1 surfaces and grade 2/grade 3 surfaces suggested that HCF proliferation is susceptible to Ti roughness ≥ 0.68 μm R_a (grade 2). (E) Proliferation rates of HCEnC were similar between the tested surfaces. (G) Examples of Rose Bengal staining of HCF in culture at day 5. Numbers indicate coverage of Rose Bengal stain over the Ti surface. (B, D, F) Ti surface topography had no significant effect on HCLE, HCF, and HCEnC cell cytotoxicity within the studied time. *Group is significantly different from all the other groups at the specified time point, P < 0.05, 1-way ANOVA.

Human Corneal Fibroblasts.

Human corneal fibroblasts exhibited a high proliferative activity in serum culture. All Ti grades and the plastic surface supported HCF proliferation within 5 days (Fig. 3C). However, HCF proliferation was significantly retarded on rough Ti (grades 2 and 3) as compared to smoother Ti (grades 0 and 1) at all the specified time points (all comparisons, P < 0.05, by 1-way ANOVA) (Fig. 3C). This difference was visualized by Rose Bengal staining, which displayed the HCF coverage on different Ti surfaces at day 5 (Fig. 3G). The difference in proliferation presented between grade 0 and grade 1 was not statistically significant at day 2 and day 5, but reached statistical significance at day 3 (P = 1.06 × 10⁻⁵, 1-way ANOVA). These results indicate that HCF proliferation is also affected by Ti surface roughness and becomes biologically important at R_a values > 0.681 μm. 

Human Corneal Endothelial Cells.

The rate of HCEnC proliferation was similar on all Ti grades (Fig. 3E) and was slower compared to HCLE or HCF. 

Cytotoxicity

Titanium surface topography had no cytotoxic effect on HCLE, HCF, and HCEnC within the studied time frame (Figs. 3B, 3D, 3F). The LDH levels of wells with Ti disks were statistically equal to those of the control plastic wells. 

HCF Morphologic and Biochemical Changes in Response to Different Ti Surfaces

Cell Adhesion and Morphology.

Cell cultures with HCF on Ti grades 0 to 3 were performed for 4 days, at which time cell morphology was assessed and compared by immunocytochemistry. There were significant morphologic differences between cells grown on grade 0 and those on grade 3 Ti. Human corneal fibroblast cells on grade 0 Ti had larger cell area and diameter, longer perimeter, increased solidity, and larger ellipsoid boundaries as compared to HCF cultured on grade 3 (Fig. 4A; Table [without TGFβ1]). Phalloidin staining of cells plated on grade 0 (polished) Ti showed increased assembly of stress fibers as compared to grade 3 (rough) (Fig. 4A). This increase may be a result of the upregulation of focal adhesions on grade 0 Ti, as stress fibers frequently anchor on focal adhesion complexes.³¹ 

Figure 4

 

Cell morphology and alignment at 4 days and 1.5 months of culture. (A) HCF grown on grade 0 and grade 3 Ti surfaces at 4 days of culture without TGFβ1. Cells were fixed with 4% PFA and stained with Alexa Fluor 488 conjugated phalloidin and DAPI. Scale bar: 50 μm. (B) HCF organization on grade 0 and grade 3 Ti surfaces at 1.5 months of culture. Upper: cell morphology (green, F-actin; blue, nuclei). Lower: directionality histograms of above images. Grade 3 showed greater dispersion of angles than grade 0. Scale bar: 100 μm. (C) Cell densities on different Ti surfaces after 1.5 months of culture in the presence or absence of 10 ng/mL TGFβ1. In each culture group, cells achieved equal cell densities on all Ti grades by 1.5 months (×20 objective; green, F-actin; blue, nuclei).

Table

 

Morphologic Comparison of Human Cornea Fibroblasts on Titanium Surfaces in the Presence or Absence of TGFβ1 Stimulation

 

In summary, HCF at 4 days of culture showed better cellular attachment, increased cell spreading, and increased expression of stress fibers on grade 0 Ti as compared to grade 3. This difference in cell adhesion may, in part, contribute to the variation observed in HCF cell proliferation between the Ti grades. 

The addition of TGFβ1 to the HCF cell feeding media (Table [with TGFβ1]) equilibrated the morphologic discrepancies observed between grade 0 (polished) and grade 3 (rough) Ti, as difference in the values of the measurements became statistically insignificant (Table [with TGFβ1: NS]). 

Long-Term Culture.

At 1.5 months of culture in serum media with ascorbic acid (pAsc), HCF were 100% confluent with equal cell density between the different Ti grades (Fig. 4C). Phalloidin staining demonstrated HCF stratification into multiple cell layers. Grade 0 Ti promoted uniform alignment of cells within the stratum, and each stratum showed a different orthogonal orientation of cells with respect to other strata (Fig. 4B). The enhanced directionality of cell alignment was incrementally diminished with increased Ti surface roughness, with grade 3 inducing random cell positioning (Fig. 4B). 

Addition of TGFβ1 to the feeding media caused HCF to transform to myofibroblasts, as indicated by the increased expression of α-SMA (Supplementary Fig. S1). However, all roughness grades achieved equal cell densities by 1.5 months (Fig. 4C). Unlike corneal fibroblasts, myofibroblasts did not show significant differences in cell morphology or alignment between the different Ti grades after 1.5 months of culture (Supplementary Fig. S1). 

Collagen Deposition on Ti.

In order to assess whether or not Ti surface topography had an effect on matrix deposition by HCF, cells were stimulated to secret elevated levels of collagen in vitro by adding pAsc to the culture media. Human corneal fibroblast synthesis of collagen type I and type V at 4 days of culture was minor, primarily found within the boundaries of cells and between the cell bodies and the Ti substrate (Fig. 5). 

Figure 5

 

Collagen type I and type V expression by HCF on grade 0 and grade 3 Ti surfaces at 4 days of culture. HCF synthesized small quantities of collagen type I and type V at 4 days of culture, largely contained within cellular areas, lying between cell bodies and the Ti substrate. (A) Extracellular collagen clusters (arrowhead) and thin collagen fibrils between cells (white arrow) were regularly observed. (B) HCF cultured on grade 3 Ti at day 4 secreted fewer extracellular collagen clusters or fibrils as compared to those cultured on grade 0 Ti. (C, D) HCF secreted less collagen type V than collagen type I, especially when on grade 3 Ti surface. Scale bar: 100 μm (green, F-actin; red, collagen; blue, nuclei).

Extracellular collagen clusters and thin collagen fibrils interconnecting the cells were frequently observed (Fig. 5A). Human corneal cell fibroblasts cultured on rough (grade 3) Ti at day 4 secreted fewer extracellular collagen clusters or fibrils (Fig. 5B) as compared to those cultured on polished (grade 0) Ti. Collagen type V secretion was significantly reduced on rough (grade 3) Ti surface (Figs. 5C, 5D). Addition of TGFβ1 had a minimal effect on collagen type I and V expression at 4 days of culture as compared to the culture without TGFβ1 (Supplementary Fig. S2). 

After 1.5 months of culture in serum media plus P-Asc, confluent HCF had deposited visible membranous extracellular matrix on the entire surface area of the Ti disks. The total amount of collagen in the cell matrix composition on grade 3 Ti was significantly lower than on the smoother Ti surfaces (grades 0–2) (Fig. 6A). Likewise, polished (grade 0) Ti had a thicker deposition of cell matrix compared to rough Ti grades 1 to 3 (Fig. 6B). Immunofluorescence staining with antibody against collagen types I and V in the matrix showed distinct differences in collagen organization between the Ti grades. Collagen staining in grade 0 showed a parallel alignment of the fibrillar collagen meshwork (Fig. 6C, a), while staining in grade 3 showed randomly scattered collagen deposits (Fig. 6C, d). 

Figure 6

 

Matrix organization and collagen type I and V secretion by HCF on different Ti surfaces in the presence or absence of TGFβ1 at 1.5 months of culture. (A) Total collagen amount/Ti disk, measured by hydroxyproline assay. (B) Cell-assembled matrix thickness. (C) Immunofluorescence images showing collagen type I and type V organization on different Ti surface grades. Scale bar: 100 μm. *Group that is significantly different from all the other groups, as indicated by 1-way ANOVA, n = 3.

The addition of TGFβ1 to HCF culture caused a marked increase in collagen deposition in all Ti grades except grade 3, in which the total collagen amount remained the same as with no TGFβ1, and was significantly lower than those for grades 0 to 2 (gray column in Fig. 6A). Likewise, the thickness of cell matrix composite on grade 3 Ti surface was significantly lower than on smoother grades 0 to 2 (gray column in Fig. 6B). Consistent with the increase in collagen quantity, immunofluorescence staining with antibody against collagen type V also demonstrated increased expression of this collagen type in all Ti grades (Fig. 6C, m–p). Treatment with TGFβ1 partially equilibrated the effect of Ti surface roughness on matrix thickness; in the absence of TGFβ1, Ti grades 1 to 3 significantly reduced the matrix thickness, while in the presence of TGFβ1, thickness was significantly reduced only at grade 3. 

Focal Adhesion Kinase

Human corneal fibroblasts cultured on different Ti grades in the presence or absence of TGFβ1 stimulation were lysed and analyzed for total and phosphorylated focal adhesion kinase (FAK) (tyrosine 397; pFAK) by Western blot. The levels of pFAK were similar among the different Ti grades at 4 days of culture without TGFβ1 (data not shown). At 1.5 months of culture without TGFβ1, the level of pFAK was significantly lower on the rough Ti (grade 3) as compared to smoother Ti grades 0 to 2 (Fig. 7; P < 0.05, 1-way ANOVA). This coincides with the findings that grade 3 results in less cell spreading, less stress fiber assembly, slower HCF proliferation, and reduced collagen deposition. 

Figure 7

 

FAK phosphorylation (Y397) in HCF at 1.5 months of culture without TGFβ1. pFAK at 1.5 months of culture was significantly reduced on the roughest Ti surface (grade 3), suggesting that HCF had less mature focal adhesion sites on high-roughness Ti surfaces. The results are normalized to total FAK. Attached image: a representative Western blot. *Statistical significance from the other groups, n = 3.

However, no significant difference in pFAK levels between different Ti grades was found in TGFβ1-supplemented culture at 1.5 months (data not shown), suggesting that addition of TGFβ1 in HCF culture overrides the effect of Ti surface topography on FAK phosphorylation, consistent with the findings that cell morphology and collagen deposition are less impacted by Ti surface topography. 

Alpha-Smooth Muscle Actin Expression

In corneal wound healing, myofibroblasts are instrumental to wound closure, collagen synthesis, and extracellular matrix (ECM) remodeling and contribute to corneal fibrosis.^17,18 In the cornea, α-SMA is expressed only by myofibroblasts³² and is a marker for assessing transformation of fibroblast to myofibroblast in wound healing.^33,34 

Consistent with previous studies, α-SMA expression by primary HCF was negligible at 4 days and 1.5 months in FBS + P-Asc media. Furthermore, Ti surface topography had no significant effect on α-SMA expression. However, addition of TGFβ1 in HCF cell culture led to a significant increase in α-SMA expression at 4 days (Fig. 8A), and at 1.5 months the majority of cells were α-SMA positive (Supplementary Fig. S1). After 1.5 months of culture, HCF cells on grade 3 Ti had the lowest α-SMA expression level compared to grades 0 to 2, suggesting that rougher Ti surface downregulates the transformation of HCF to myofibroblasts (Fig. 8B). 

Figure 8

 

α-SMA expression by HCF cultured with TGFβ1 on the test Ti surfaces at 4 days and 1.5 months of culture. (A) Immunofluorescence images of cells on grade 0 and grade 3 Ti surfaces at 4 days. (B) α-SMA expression at 1.5 months on tested Ti surfaces. Western blot shows that rough Ti surfaces have a long-term effect on α-SMA expression induced by TGFβ1 signaling. The results are normalized to GAPDH. Scale bar: 50 μm.

Discussion

This study investigated the effect of macro- and microscale Ti surface topography in corneal cell behavior. Our results suggest that corneal cells are affected by topographic cues in a cell type-specific manner. Contrary to accepted paradigms in orthopedics and dentistry, rough Ti was not found to promote corneal cell proliferation, adhesion, or collagen deposition in vitro. 

Our proliferation study demonstrated diverse biological responses among corneal cells of different strata. Our results showed that HCLE cells and HCF were the most susceptible to topographic cues. A previous study of corneal tissue constructs using membranous polymers found that HCLE cells were affected by nanoscale surface topography, but not HCF.³⁵ However, our results suggest that macro- and microscale topography strongly regulates cell cycle and biological behavior for both cell types. In our study, HCF proliferation was inhibited by topographic parameters equal to or above R_a = 0.618 ± 0.047 μm, RS_m = 26.96 ± 3.70 μm of grade 2, while HCEnC remained largely unaffected. Insusceptibility of HCEnC to topographic cues may be attributed to unique cellular characteristics. For example, HCEnC are nonproliferative under normal circumstances in vivo. They are arrested in the G1 phase of the cell cycle,^36,37 which may be partially attributed to leaky barrier properties because of incomplete zonula occludens and an ATPase-dependent pump that regulates hydration of the corneal stroma while allowing nutrients to pass.³⁸ None of the corneal cell types exhibited cytotoxicity in response to Ti topography but the topographic characteristics of Ti as a pivotal factor in determining cell proliferation and signaling. The significant inhibitory effect of rough Ti on HCLE cell and HCF proliferation is in striking contrast to the response of osteoblasts to Ti roughness,^39,40 and suggests significant cell type-specific differences in response to surfaces of prosthetic materials. 

In corneal implants such as the B-KPro, RPM formation is a common postoperative complication. Retroprosthetic membrane impedes vision and increases the risk of implant extrusion.¹⁴ Retroprosthetic membrane can be caused by the migration of corneal fibroblasts from the corneal host–donor junction to the posterior surface of the Ti back plate where they differentiate into myofibroblasts.⁴¹ Loosely woven collagen fibers and myofibroblasts were found to be the major components of RPM.⁴¹ In vitro, HCF cell transformation into myofibroblasts and the subsequent deposition of collagen matrix can be used to simulate RPM formation. This is of particular relevance to understanding the molecular/topographic cues that can prevent this complication. The presence of unidirectional multicellular organization of collagen fibrils on grade 0 Ti suggests that polished Ti may promote RPM formation and that roughening of the posterior surface of the Ti back plate may prevent HCF migration and proliferation and circumvent RPM formation. However, an in vivo model is required to test this hypothesis. 

Our study also showed increased expression of collagen type V following TGFβ1 stimulation on all Ti surface grades. However, the fibrils in grade 3 were significantly less organized and aligned as compared to those in grade 0. Increased collagen type V expression leads to reduced collagen fibril diameter and increased collagen cross-linking.^42,43 Although we did not directly measure the diameter or cross-linking of collagen matrix, we found significant increases in total collagen amount and insignificant change in collagen matrix thickness, which indirectly suggests increased collagen compactness following TGF-β1 stimulation. During corneal wound healing, TGFβ1 can increase fibrosis via increased collagen expression by HCF and increase the compactness of collagen fibrils. These changes may contribute to the process of RPM formation in the early reparative phase following B-KPro implantation, an effect that may be inhibited by increasing the macro- and microscale surface roughness of Ti on the posterior surface of the B-KPro back plate. 

Focal adhesion kinase is an important cellular tyrosine kinase in fibroblasts, and is recruited to the site of integrin clusters during adhesion.⁴⁴ The cytoplasmic tail of β-integrins facilitates FAK activation through FAK clustering, autophosphorylation at Y397,⁴⁴ and the mechanical linkage of integrins to the actin cytoskeleton. Phosphorylation of FAK triggers further activation of intracellular signaling molecules and enables cell adhesion, spreading, proliferation, and migration.^45–47 Therefore, FAK phosphorylation may serve as a biomarker for cell–material interactions.^48–50 In our study, rough Ti (grade 3) inhibited FAK phosphorylation of HCF at 1.5 months in the absence of TGFβ1. According to proposed FAK signaling pathways,⁵¹ FAK mediates TGFβ1 signaling, which in turn activates α-SMA expression and promotes collagen deposition by myofibroblasts. In presence of TGFβ1 stimulation in culture media, FAK phosphorylation was not affected by Ti surface topography. However, the expression of α-SMA in grade 3 Ti was significantly reduced, suggesting that α-SMA expression via FAK phosphorylation is inhibited in the presence of rough topographic cues. Since cells can form various adhesion structures even without FAK,⁵⁰ alternative signaling routes, such as the integrin-Src linkage,⁴⁵ may be more critical in the signaling evoked by Ti surface topography. Therefore, the role of TGFβ1 signaling in corneal fibroblast transformation is yet to be determined in the context of surface topographic stimulation. 

In conclusion, to our knowledge, ours is the first study to evaluate the effect of macro- and microscale Ti surface topography on different corneal cell types. Our study demonstrates that surface topography affects corneal cells in a cell type-dependent manner, varying across the different corneal strata, each populated by distinct cell populations. Macro- and microsurface topography modifications can be manipulated to alter the biological behavior of corneal cells and to suppress the impact of TGFβ signaling on α-SMA and collagen expression by HCF. Tissue-targeted optimization of Ti surface topography may prove beneficial in promoting collagen deposition and biointegration of corneal implants and in reducing the occurrence of RPM, thereby improving implant survival and retention in patients with artificial corneas. 

Acknowledgments

The authors thank Ilene Gipson, PhD, for providing HCLE cells, James Zieske, PhD, for providing primary human corneal fibroblasts, and Ula Jurkunas, MD, for providing HCEnC. 

Supported by the Boston Keratoprosthesis Research Fund, Massachusetts Eye and Ear, and the Eleanor and Miles Shore Fund. All authors are full-time employees of Massachusetts Eye and Ear Infirmary, a nonprofit organization that is the manufacturer of the Boston Keratoprosthesis. The authors alone are responsible for the content and writing of the paper. 

Disclosure: C. Zhou, None; F. Lei, None; J. Chodosh, None; E.I. Paschalis, None