Bioengineered Human Corneal Endothelium for Transplantation FREE

Human corneal endothelial cells (HCECs) maintain corneal clarity by means of a barrier function and pump-leak mechanism.¹ Regarded as nonproliferative in vivo,² the number of HCECs decreases with aging and because of other factors such as trauma, inflammation, and the use of contact lenses. Full-thickness corneal transplantation (penetrating keratoplasty) is currently the most common way to treat corneas that are opacified owing to endothelial dysfunction. Because the supply of donor corneas is insufficient and because of the potential for complications in penetrating keratoplasty, use of cultured HCECs for the replacement of the endothelium alone would be a substantial advantage.

Corneal endothelial cell transplantation was attempted to repopulate rabbit cornea with unhealthy endothelium by directly injecting a cell suspension into the anterior chamber.³ However, that trial was limited because only scattered clumps of endothelial cells randomly attached to the targeted cornea and to other normal ocular tissues such as the iris and lens. In recent years, numerous investigators have reported a method to transplant corneal endothelial cells by seeding and cultivating them on different carriers made of natural tissue materials^{4 - 5} or artificial polymeric materials.^{6 - 8} Although a monolayered architecture of cultured cells was maintained, the intraocular grafting of these engineered tissue replacements may cause problems such as unstable attachment of the cell carrier membrane to the host corneal stroma and fibroblastic overgrowth between the membrane and stroma.³ By avoiding the permanent residence of foreign carrier materials in the host, our group has recently established a novel strategy for corneal endothelial reconstruction using cultured HCEC sheets, which were made by the temperature-modulated detachment of cells with uniform polarity from thermoresponsive poly-N-isopropylacrylamide (PNIPAAm)–grafted surfaces.⁹ A biodegradable and cell-adhesive gelatin hydrogel disc was used to provide a temporary support structure during and after in vivo delivery of adult HCEC sheets onto the corneal posterior surface.

Recently, attempts have been made to reconstruct the ocular surface by transplantation of cultured epithelial cell sheets originating from autologous corneas¹⁰ or oral mucosas.¹¹ The results were encouraging because the visual acuity of patients who received tissue-engineered cell sheets was significantly improved.¹² Cell sheets have an intact cellular arrangement and a cellular organization, which are important factors for successful graft-host integration and tissue repair. Our previous report¹³ on transplantation of intact retinal sheets demonstrated that, when positioned with correct polarity, these grafts could grow into ordered and viable laminated retinas. However, the dissociated retinal cell suspensions or microaggregates develop only the rudimentarily differentiated rosettes after being grafted into the subretinal space.¹⁴

Yamada et al¹⁵ reported that cells adhered to and proliferated on the hydrophobic PNIPAAm-grafted surfaces at 37°C and spontaneously detached from the switched hydrophilic surfaces when the culture temperature was reduced below the lower critical solution temperature of the PNIPAAm (ie, 32°C in water). To harvest the cell cultures as whole sheets instead of as isolated suspensions, our group fabricated PNIPAAm-grafted surfaces by means of plasma chemistry, a powerful technique that previously allowed us to develop artificial corneas.^{16 - 18} For the present study, untransformed adult HCECs were cultivated on the nanostructured PNIPAAm-grafted surfaces for 3 weeks at 37°C, and confluent monolayers were obtained at 20°C (Figure 1). The characteristics of bioengineered HCEC sheets were determined in vitro by evaluating their viability and by scanning electron microscopy, immunohistochemistry, and histological studies. Evaluations of native corneal endothelium from human eye bank donors were conducted simultaneously for comparison.

This research followed the tenets of the Declaration of Helsinki involving human subjects and was approved by the institutional review committee of Taipei Veterans General Hospital. The HCECs were isolated and cultured as described previously.¹⁹ All of the cell culture supplements were purchased from Sigma-Aldrich Corp (St Louis, Mo) unless otherwise noted. The 25 human corneas used in this study were obtained from National Disease Research Interchange, Philadelphia, Pa, and were from donors aged 55 to 80 years, and the corneas had been stored in Optisol-GS solution (Bausch and Lomb Surgical, Irvine, Calif) at 4°C. For the isolation of endothelial cells, the Descemet membrane–corneal endothelium complex was aseptically stripped from the stroma and digested using 1.2 U/mL of a grade II neutral protease (Dispase II; Roche Diagnostics, Indianapolis, Ind) in Hanks balanced salt solution (Gibco, Grand Island, NY) for 1 hour at 37°C. The HCECs were pelleted and resuspended in regular growth medium containing modified Eagle medium (OptiMEM; Gibco) as a basal medium, 15% fetal bovine serum (Biological Industries, Kibbutz Beit Haemek, Israel), 40 ng/mL of bovine pituitary fibroblast growth factor, 5 ng/mL of human recombinant epidermal growth factor (Upstate Biotechnology, Lake Placid, NY), 20 ng/mL of nerve growth factor (Biomedical Technologies, Stoughton, Mass), 20 μg/mL of ascorbic acid, 0.005% human lipids, 0.2 mg/mL of calcium chloride, 0.08% chondroitin sulfate, 1% RPMI 1640 vitamins solution, 50 μg/mL of gentamicin (Gibco), and 1% antibiotic-antimycotic solution (Biological Industries). The cultures were incubated in a humidified atmosphere of 5% carbon dioxide at 37°C. The medium was changed every other day. After 1 week in culture, confluent cell monolayers were subcultured by treating with trypsin-EDTA (Gibco) for 2 minutes and seeded at a 1:3 split ratio. Only second-passage HCECs were used during all experiments.

Thermoresponsive supports with different grafting amounts of PNIPAAm on the temperature-modulated cell adhesion and detachment will be described in a separate publication. For this study, we used the culture surfaces grafted with PNIPAAm at an optimal density of 1.6 μg/cm². After surface sterilization with UV light for 2 hours in the laminar flow hood,²⁰ the HCECs were plated on PNIPAAm-grafted culture dishes (35 mm in diameter) at a density of 4 × 10⁴ cells/cm² and cultivated under the same conditions as for cell preparation. Cell morphology was observed by means of inverted phase-contrast microscopy (TMS; Nikon, Melville, NY). To estimate the cell density, a micrometer scale was used to calculate the number of endothelial cells in confluent cultures after 3 weeks of incubation. Six regions on each culture surface were randomly selected, and cell nuclei within each area were counted manually at 40× magnification. For the harvest of cell sheets, the thermoresponsive supports containing confluent cultures were rinsed twice with warmed phosphate-buffered saline (PBS) and replenished with serum-free modified Eagle medium. The HCEC monolayers were detached from the PNIPAAm-grafted surfaces by changing the culture temperature from 37°C to 20°C.

Cell viability of the harvested HCEC monolayers was determined by a membrane integrity assay, using the live/dead viability/cytotoxicity kit (L-3224; Molecular Probes, Eugene, Ore), which contains calcein acetoxymethyl and ethidium homodimer-1. Briefly, after washing them 3 times with PBS, the HCEC sheets were stained with a working reagent, composed of 4 μL of ethidium homodimer-1, 2 mL of PBS, and 1 μL of calcein acetoxymethyl. The samples were incubated for 30 minutes at 37°C and viewed under an inverted fluorescent microscope (Eclipse TS100 equipped with an epifluorescence attachment; Nikon).

Whole corneas from human eye bank donors (as a control group) and detached HCEC sheets were fixed with 2% glutaraldehyde in 0.1M cacodylic acid buffer (pH, 7.4) overnight at 4°C. After rinsing them with 0.1M cacodylic acid buffer 3 times for 10 minutes each time, the specimens were postfixed in 1% osmium tetroxide for 30 minutes and dehydrated in 50%, 70%, 90%, and 100% ethanol solutions for 10 minutes at each concentration, run in 2 consecutive series. The samples were further dried with carbon dioxide in a critical point dryer (HCP-2; Hitachi, Tokyo, Japan) and gold coated by ion sputtering (SPI Module; Structure Probe, West Chester, Pa) before examination under a scanning electron microscope (JSM5600; JEOL, Tokyo, Japan) at an accelerating voltage of 10 kV.

Control samples and HCEC sheets were fixed with 4% paraformaldehyde for 10 minutes at 4°C. After washing with PBS, the fixed specimens were then permeabilized in 0.3% Triton X-100 for 15 minutes and blocked with 4% bovine serum albumin in PBS for 30 minutes. The samples were incubated with primary antibodies overnight at 4°C in a moist chamber. The antibodies, diluted in PBS containing 4% bovine serum albumin, were directed against zonula occludens-1 (ZO-1) (1:100; Zymed Laboratories, South San Francisco, Calif) or Na⁺,K⁺–adenosine triphosphatase (Na⁺,K⁺-ATPase) (1:150; Upstate Biotechnology). The negative controls were incubated without a primary antibody. The specimens were washed in PBS and incubated with fluorescein- or rhodamine-conjugated donkey antimouse IgG secondary antibodies (1:200; Chemicon International, Temecula, Calif) for 2 hours at room temperature in the dark. Unbound excess labels were removed by rinsing the samples in PBS. The samples were viewed under fluorescence microscopy (Axioplan 2 [Carl Zeiss, Oberkochen, Germany] or BX51 [Olympus, Tokyo, Japan]).

The control samples and HCEC sheets were mounted onto precooled chucks in embedding medium (OCT Tissue-Tek; Sakura Finetek, Torrance, Calif) and frozen at −70°C. Frozen specimens were cut with the use of a cryostat into 5-μm sections at −20°C. After fixation with 4% paraformaldehyde for 1 minute, the sections were stained with 4′,6-diamidino-2-phenylindole (Vector, Peterborough, England) for visualization of cell nuclei and examined using a fluorescence microscope (Axioplan 2; Carl Zeiss).

After 4 hours of plating, isolated HCECs attached and spread on the PNIPAAm-grafted surfaces. the HCECs grew readily to reach confluence after 1-week cultivation at 37°C. By incubating the HCECs in medium an additional 2 weeks, a thick layer of extracellular matrix (ECM) was deposited at the basal cell surface to allow the formation of a fully confluent monolayer (not shown). Under a phase-contrast microscope, confluent HCECs on the PNIPAAm-grafted surfaces appeared generally polygonal and had a high cell density (approximately 2500 cells/mm²) that was almost the same as that found in vivo (Figure 2). This indicated that the ex vivo proliferation rate of the HCECs was maintained on the thermoresponsive supports. Subsequently, the culture temperature was lowered to 20°C and cell release from the PNIPAAm-grafted surfaces was observed (Figure 3). At the beginning of the low-temperature treatment, the HCEC monolayers were rolled up at the margin of the culture surfaces and centripetally detached owing to the gradual hydration of the PNIPAAm-grafted chains (Figure 3A). Such a process of cell separation from thermoresponsive supports is a mode of sheetlike movement. After 45 minutes of incubation, a laminated, approximately 0.75-cm² HCEC sheet was harvested from the completely hydrated PNIPAAm-grafted surfaces and was wrinkled because of the contracting force of this cell lamella (Figure 3B). These results suggest that the cell sheet detachment from the thermoresponsive supports correlates closely with temperature-modulated surface hydration ability.

We determined the cell viability of the harvested HCEC monolayers by the live/dead viability/cytotoxicity assay (Figure 4). This assay uses intracellular esterase activity to identify the living cells; the process cleaves the calcein acetoxymethyl to produce a green fluorescence. Ethidium homodimer-1 can easily pass through the damaged cell membranes of dead cells to bind to the nucleic acids, yielding a red fluorescence. Nearly all of the cells were vital throughout the central region of the detached HCEC sheets (Figure 4A). Only a few dead cells were interspersed between the live cells of the endothelial monolayers. However, the peripheral region of the HCEC sheets showed a large number of green-stained cells and red-stained nuclei in the margin of the cell monolayers (Figure 4B and C). This was probably due to the loss of anchorage dependence for marginal cells caused by overgrowth of a fully confluent cell sheet during 3 weeks of culture. These data clearly demonstrated that most of the monolayered HCECs remained viable after detachment at 20°C, suggesting that the low-temperature incubation does not compromise cell viability.

The scanning electron microscopic studies examined the morphological characteristics of native endothelium in the human eye bank corneas and of the monolayered cells in the cultured HCEC sheets (Figure 5). In the control groups, HCECs on the Descemet membrane packed together and formed a single continuous monolayer (Figure 5A). The individual cells were intact, had distinct borders, and possessed a hexagonal shape with minor irregularities. By contrast, after immediate separation from the thermoresponsive supports, the monolayered cells within the HCEC sheets were polygonal and had multiple cellular interconnections (Figure 5B). The absence of clear boundaries between these single cells was probably due to the cell contraction caused by detachment at a low culture temperature. In addition, a thick ECM layer was observed on the basal cell surface of the HCEC sheets (Figure 5B).

Immunohistochemical staining of ZO-1, a tight junction–associated protein, was used to determine whether the cells within the HCEC sheets formed tight junctions. In human donor corneas, the cells of native endothelium retained the tight junctions that are responsible for establishing the passive permeability properties of the endothelial barrier (Figure 6A). Similar to what we found in the control samples, ZO-1 was located at the cell boundaries of the HCEC sheets, suggesting the formation of focal, tight junctional complexes (Figure 6B).²¹ At higher magnification, the discontinuity of ZO-1 localization in the HCEC sheets, which is a normal feature in corneal endothelium, was observed with gaps occurring at the Y-junctions between 3 adjacent cells (not shown). On the other hand, Na⁺,K⁺-ATPase, an integral membrane protein complex responsible for regulating ionic pump functions, was located at the basolateral membrane of the HCECs within the control samples and the detached HCEC sheets (Figure 6C and D).²²

The cross-sections of control samples and HCEC sheets were stained for 4′,6-diamidino-2-phenylindole to examine the histological structure of the endothelium. The endothelial cells from the human donor corneas were organized on the Descemet membrane as a monolayer (Figure 7A). The detached cell sheets also showed a monolayered architecture of cells that mimicked native endothelium (Figure 7B).

Human corneal endothelial cells in vivo demonstrate an age-related decrease in cell density and cannot be compensated because of their limited regenerative capacity.²³ When the cell density is less than 500 cells/mm², a critical level, the endothelium no longer functions, causing corneal edema and loss of visual acuity. In these cases, HCEC transplantation aims to restore vision by reconstituting a structural and functional endothelial monolayer. Central to tissue reconstruction is the cellular arrangement and organization of the grafts (ie, the well-organized cell sheets or the isolated cell suspensions). Therefore, the ability to obtain an intact monolayer of HCECs in vitro should be beneficial for developing an effective therapeutic strategy to rescue damaged endothelium. In the present report, we recount a novel method for harvesting cultured HCEC sheets as a graft source for tissue repair. In contrast to traditional enzymatic digestion (ie, trypsinization), a cell culture system using thermoresponsive supports has been designed to allow the bioengineered endothelial equivalents retain their cellular activity, organization, function, and ECM integrity.

Cultivation of adult HCECs from older donors has proved to be difficult.²⁴ To overcome this, we have developed a growth factor–enriched medium to successfully mass culture untransformed adult HCECs.¹⁹ Our group's previous studies^{19 ,25} also demonstrated that the endothelial cells could be identified as being of endothelial origin on the basis of their morphology and by reverse transcriptase–polymerase chain reaction for keratin 12 and collagen type VIII. In the present study, the HCECs derived from older donor tissue kept their phenotypic characteristics and proliferative capacity when they were cultivated on nanostructured PNIPAAm-grafted surfaces, just as they would if grown on the commercial tissue culture plates (not shown). To strengthen the harvested HCEC monolayers, an additional 2-week incubation enhances cell production of ECM and the confluent cultures are stimulated to be even thicker. As verified by transmission electron microscopy, the detached HCEC sheets had abundant cytoplasmic organelles and deposited ECM (not shown). This unique phenomenon of ECM formation in cultured HCECs possibly indicated the same property of increasing thickness of the Descemet membrane with aging in the human cornea.²⁶ In addition, the cell density of confluent HCEC monolayers is comparable to that in vivo. It suggests that these bioengineered tissue equivalents have sufficient cell numbers to support their barrier and ionic pump functions.

In vitro growth of HCECs is anchorage dependent. During cell sheet detachment from the thermoresponsive supports, each endothelial cell at the leading edge assembles by contracting fan-shaped lamellipodia. After 45 minutes of incubation at 20°C, the area of harvested HCEC sheets decreased from approximately 9.6 cm² to approximately 0.75 cm². These results support the report by Shimizu et al²⁷ demonstrating that spread cells are compacted after the temperature-modulated detachment of cultured cardiomyocyte sheets. One possible explanation for the shrinkage of HCEC sheets is that the reorganization of cytoskeleton is induced by the low-temperature treatment. In autologous transplantation, the HCEC sheet grafts should be designed to fit the size of the recipient's initial biopsy specimen. Despite the contraction observed in these bioengineered tissue grafts, the size of harvested HCEC sheets can be adjusted arbitrarily by controlling the surface size of the thermoresponsive culture supports.

It has been reported that hypothermic preservation of cells at 4°C would result in decreased activity of the sodium pump²⁸ or induce apoptosis.²⁹ Because the cell sheet detachment is performed at 20°C, which is lower than its normal culture temperature (37°C), the effects of low-temperature treatment on the viability of harvested tissue equivalents should be investigated. Our data indicate that the detached HCEC monolayers are tolerant of 45-minute incubation at 20°C. In the present study, we performed viability testing in cell sheets after their immediate separation from thermoresponsive supports. Although the HCEC monolayers were detached and preserved in serum-free modified Eagle medium, the storage time (the time between detachment and the beginning of implantation) is not a concern for our study. By using cell-adhesive and transparent gelatin hydrogel carriers, the detached HCEC sheets with good viability can be immediately transplanted through a 7.5-mm sclerocorneal incision to recipient corneas denuded of endothelium.⁹

In addition to retaining their deposited ECM, the harvested HCEC sheets, when studied morphologically, exhibit cellular interconnections and consist of closely packed, small polygonal cells. These findings are consistent with previous publications regarding morphological observations of cultured corneal endothelial cells on carrier substrates^{5 ,7 - 8} or corneas denuded of endothelium.^{1 ,19 ,30 - 31} Immunohistochemistry studies demonstrated the proper location of ZO-1 and Na⁺,K⁺-ATPase proteins, implying that the HCEC sheets are capable of maintaining intact barrier and ionic pump functions. The monolayered architecture of detached cell sheets is also confirmed by histological examination. These characteristics of cultured HCEC monolayers are similar to those observed in the native endothelium of eye bank donor corneas.

To develop alternative therapy techniques, our group⁹ has evaluated the feasibility of using HCEC sheets for corneal endothelial reconstruction. Results from a short-term study⁹ suggest that the transplanted HCEC sheets could be integrated into rabbit corneas denuded of endothelium. In addition, the corneas have returned to a near normal thickness, indicating the function of bioengineered HCEC sheets. Although these data are encouraging, the long-term efficacy and safety data of HCEC sheets need further investigation.

We have shown that it is possible to fabricate the bioengineered human corneal endothelium in vitro by the temperature-modulated detachment of cultured cell sheets from thermoresponsive supports. Based on the current methods, adult HCEC monolayers with normal morphology and viability can be obtained without the need for cell carriers during cultivation. Despite immunohistochemical data indicating that HCEC sheets maintain intact barrier and ionic pump functions, additional functional assessments are needed. Future studies using a corneal perfusion chamber may determine whether the HCEC sheets can maintain stromal dehydration and corneal transparency. Future studies may also succeed in developing surgical-quality corneas by the incorporation of HCEC sheets into donor corneas denuded of endothelium. The in vivo applications of bioengineered HCEC monolayers for corneal endothelial reconstruction in a rabbit model are currently being investigated.

Correspondence: Ging-Ho Hsiue, PhD, Department of Chemical Engineering, National Tsing Hua University, No. 101, Section 2, Kuang-Fu Road, Hsinchu, Taiwan 30013 (ghhsiue@mx.nthu.edu.tw).

Submitted for Publication: January 5, 2006; final revision received April 13, 2006; accepted May 14, 2006.

Author Contributions: Drs Lai and Chen contributed equally to this study.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grant NSC91-2320-B-007-007 from the National Science Council of the Republic of China (Dr Hsiue), grant VGH-94-365-9 from Taipei Veterans General Hospital (Dr Chen), grant VGHUST94-P1-02 from the Veterans General Hospitals University System of Taiwan Joint Research Program (Drs Hsiue and Chen), and grant NHRI94A1-MEAP01-001 from National Health Research Institutes (Dr Chen).