Development and Evaluation of Porcine Atelocollagen Vitrigel Membrane With a Spherical Curve and Transplantable Artificial Corneal Endothelial Grafts

Abstract

Purpose.: To develop a collagen vitrigel (CV) optimized as a corneal endothelial cell (CEC) carrier and create an artificial corneal endothelial graft.

Methods.: We first developed a flat-shaped collagen vitrigel for regenerative medicine (CV-RM) using porcine atelocollagen and ultraviolet (UV) irradiation. The optimal UV amount was determined by measuring the CV-RM transparency under various irradiating conditions. The collagen vitrigel for corneal endothelial regenerative treatment (CV-CERT), a transparent porcine atelocollagen with a curved shape, was made using spherically curved molds and UV irradiation. The membrane permeability of the CV-CERT was tested in vitro. The biocompatibility, transparency, and adhesiveness of the CV-CERT were evaluated in rabbit eyes. We also developed a culture technique for distributing human CECs on the curved CV-CERT.

Results.: The optimal amount of UV irradiation for CV-RM transparency was 2400 mJ/cm². Membrane permeability of CV-CERT at day 5 was higher than that of commercially available CV (P = 0.032). The CV-CERT was transparent and biocompatible in rabbit corneas for up to 4 months. The CV-CERT remained attached to the rabbit corneal posterior surface, whereas the flat-shaped CV-RM, differing only in shape from the CV-CERT, dislocated soon after surgery. Human CECs seeded on the CV-CERT using our technique were evenly distributed with a single layer structure and a mean cell density of 2650 ± 100 cells/mm².

Conclusions.: We developed a transparent and biocompatible porcine-derived atelocollagen vitrigel membrane with a spherical curvature. A transplantable artificial endothelial graft was created by combining cultured human CECs and the CV-CERT

Introduction

Corneal opacity is a major cause of visual impairment worldwide. ¹ Corneal transplantation is the only treatment available for corneal scarring and corneal endothelial dysfunction. Despite new surgical techniques and implementation of layer-specific replacements, corneal transplantation is limited. The cornea tissue supply shortage, coupled with its short shelf life (∼14 days), results in poor accessibility, especially in emergencies, rural areas, and developing countries. In addition, although the cornea is considered to be immune-privileged as it is an avascular tissue ² and corneal transplantations have a much lower rate of graft rejection compared with other organs, rejection remains the major cause of graft failure. ^3,4 Therefore, corneal regenerative medicine has attracted much attention because whole or parts of corneal tissues can be reconstructed to restore normal corneal function. ⁵ Corneal regeneration could solve the donor corneal supply shortage as well as reduce the rejection risk. The number of reports on corneal regenerative medicine, ^6,7 such as corneal epithelial cell sheet reconstruction ⁸ and corneal endothelial cell (CEC) regenerative treatments, ^9–11 is increasing. 

No appropriate material as a CEC carrier is currently available for transplantation, although commercially available collagen vitrigel (CA-CV) has been used as a CEC carrier by some investigators. ¹¹ Collagen vitrigel is a thin collagen membrane first reported by Takezawa et al. ¹² in 2004, and originally developed as a scaffold for three-dimensional cell culture. It is made through the vitrification process, which increases the density of collagen fibrils ∼100 times. Collagen vitrigel has good substance permeability compared with typical collagen gel sheets. There are, however, some concerns regarding the use of CA-CV for corneal regenerative medicine. First, the cornea is curved and the flat shape of conventional CV is a major cause of graft dislocation. In addition, CA-CV is very fragile and easily wrinkles if a nylon frame is not attached, making it very challenging to transplant without damaging the attached CECs. Moreover, CA-CV is made from bovine-derived native collagen and bovine serum is used during the manufacturing process. Care must be taken when clinically adapting ruminant-derived materials for humans, and optimally it should be avoided. Native collagen is also considered more antigenic than atelocollagen because telopeptides, the terminal part of native collagen without a helical structure, act as antigens. ¹³ To address these issues, we optimized CV as a CEC carrier by changing its shape, material, and manufacturing method. Here, we evaluated our newly developed porcine-derived atelocollagen vitrigel, collagen vitrigel for corneal endothelial regenerative treatment (CV-CERT), with regard to its manufacturing process, biocompatibility, and spherical shape. Further, we developed a technique for cultivating human CECs (hCECs) on the CV-CERT and created transplantable artificial corneal endothelial grafts. 

Methods

Preparation of CV Membranes

A conventional CV membrane for basic research, containing 0.5 mg collagen per 1.0 cm² unit area, was originally fabricated from 0.25% native collagen sol uniformly mixed with equal volumes of 0.5% acidic solution of bovine-derived native collagen and culture medium containing fetal bovine serum (FBS), as previously reported. ¹² This CV membrane is commercially available from AGC Techno Glass Co., Ltd. (Shizuoka, Japan) as a collagen vitrigel (CA-CV). Also, a CV membrane for regenerative medicine (CV-RM) and a spherically curved CV membrane for corneal endothelial regenerative treatment (CV-CERT), containing 2.0 ± 0.1 mg collagen per 1.0 cm² unit area, were fabricated from 0.5% atelocollagen sol uniformly mixed with equal volumes of 1.0% acidic solution of porcine-derived atelocollagen for regenerative medicine (Nippon Meat Packers, Inc., Osaka, Japan) and serum-free culture medium. The CV-RM was prepared by gelation, vitrification, and rehydration after pouring 4.0 mL of the porcine-derived atelocollagen sol into a separable container comprising a cylinder with an inner diameter of 35 mm and a plain bottom plate wider than the cylinder's outer diameter, as previously described. ¹⁴ To control the transparency by introducing cross-linkages among the collagen fibers, both surfaces of the revitrified CV-RM were irradiated with UV light (254 nm) under five different conditions as follows: Conditions 1 through 3: CV-RM was revitrified on the smooth surface, irradiated with UV at a dose of 800, 1200, or 1600 mJ/cm², and rehydrated. It was then inverted and the same procedure was repeated once to obtain CV-RM irradiated with UV at a total dose of 1600 (1), 2400 (2), or 3200 (3) mJ/cm². Conditions 4 and 5: CV-RM was revitrified on the smooth surface, irradiated with UV at a dose of 600 or 800 mJ/cm², and rehydrated. It was then inverted and the same procedure was repeated three times to obtain CV-RM irradiated with UV at a total dose of 2400 (4) or 3200 (5) mJ/cm². 

Transparency Evaluation

To optimize the UV-irradiation conditions for enhancing the transparency, CV-RMs without UV irradiation and with UV irradiation at a total dose of 1600 mJ/cm² (800 mJ/cm² × 2), 2400 mJ/cm² (1200 mJ/cm² × 2), 2400 mJ/cm² (600 mJ/cm² × 4), 3200 mJ/cm² (800 mJ/cm² × 4), or 3200 mJ/cm² (1600 mJ/cm² × 2) were subjected to a transparency assay, basically according to a previously reported method. ¹² Briefly, CV-RMs without UV irradiation and with UV irradiation at a total dose of 2400 mJ/cm² (600 mJ/cm² × 4) were attached to a film holder (FLH-267; JASCO International Co., Ltd., Tokyo, Japan) set in a spectrophotometer (V-550; JASCO International Co., Ltd.), and subsequently the absorbance spectrum of each CV-RM was measured at a wavelength from 200 to 700 nm. Three samples for each condition were used in the transparency assay at a 400-nm wavelength. Statistical analysis was performed using Student's t-test. Values of P less than 0.01 were considered significant. 

CV-CERT Manufacture

To create a new material as a CEC carrier that is transparent, thin, and spherically curved with adequate mechanical strength for surgical use, the CV-CERT was produced in two steps using four custom-made molds (Fig. 1); a spherically curved CV membrane with an 8 mm-curvature radius was prepared, and UV irradiation was subsequently performed to enhance its transparency. The spherically curved CV membrane was prepared by gelation, vitrification, and rehydration after pouring 0.7 mL atelocollagen sol into a separable mold comprising a cylinder with a 15-mm inner diameter (Fig. 1A) and a bottom plate with a concave surface having an 8-mm curvature radius and a 2.0-mm depth (Fig. 1B) where an atelocollagen sol-immersed nylon membrane ring with an 11- to 14-mm inner–outer diameter had been preinserted (Figs. 2A–C). A cylindrical piston with a 10-mm outer diameter and a spherical apex with an 8-mm curvature radius (Fig. 1C) was then concentrically set onto the gel during the gelation for 2 hours (Fig. 2D). The piston and cylinder were removed to dry and vitrify the gel (Figs. 2E, 2F). Subsequently, the rehydrated spherically curved CV membrane (Fig. 2G) was detached from the bottom plate with a concave surface, transferred to a bottom plate with a convex surface having an 8 mm-curvature radius and a 2.71-mm height (Fig. 1D), and revitrified (Fig. 2H). The revitrified membrane was irradiated with UV at a dose that was determined by the above-described transparency evaluation to enhance the transparency, resulting in the production of the CV-CERT (Figs. 2I, 2J). 

Figure 1

 

Images of the four custom-made CV-CERT molds. (A) Cylinder with a 15-mm inner diameter. (B) Bottom plate with a concave surface having an 8 mm-curvature radius and a 2.0-mm depth. (C) Cylindrical piston with a 10-mm outer diameter and a spherical apex of 8 mm-curvature radius. (D) A bottom plate with a convex surface with an 8 mm-curvature radius and a 2.71-mm height.

Figure 2

 

Manufacturing procedures for CV-CERT. (A, B) An atelocollagen sol-immersed nylon frame with an 11- to 14-mm inner–outer diameter was preinserted on a bottom plate with a concave surface. (C) A cylinder was set on the bottom plate and atelocollagen sol was poured inside. (D) A cylindrical piston was concentrically set onto the gel during gelation. (E, F) The piston and cylinder were then removed to dry and vitrify the gel. (G) The spherically-curved CV membrane was rehydrated, (H) transferred to the bottom plate with a convex surface, and revitrified. This procedure was followed by UV irradiation. (I, J) Appearance of the CV-CERT.

Membrane Permeability Testing

Membrane permeability of the CV-CERT was examined using a membrane permeability testing device (Permcell; Vidrex Co., Ltd., Kaho-gun, Japan) and compared with that of CA-CV. The permeability testing device (Vidrex Co., Ltd.) comprised two chambers. Either CA-CV or CV-CERT was placed between the two chambers. Four CA-CVs and three CV-CERTs were tested. One chamber contained 30% FBS with PBS and the other chamber contained only PBS. The permeability testing device (Vidrex Co., Ltd.) was placed at room temperature and a microstirrer was used. Samples of 100 μL were obtained from each chamber at several time points: 1 hour, 1, 3, and 5 days. Samples obtained from the PBS chamber were preserved in a freezer at −80°C until all samplings were complete. The samples were used for SDS-PAGE and a bicinchoninic acid (BCA) assay. The FBS chamber was sampled to adjust the liquid volume, but these samples were not used for testing. After thawing, 35 μL of each sample was used for SDS-PAGE. Sample buffer (NuPAGE LDS; Life Technologies, Carlsbad, CA, USA), precast gel (Mini-PROTEAN TGX; Bio-Rad Laboratories, Inc., Hercules, CA, USA), and 10× Tris/glycine/SDS buffer (Bio-Rad Laboratories, Inc.) were used for the electrophoresis. The gel was silver-stained using a commercial stain kit (Silver Stain II Kit; Wako Pure Chemical Industries, Ltd., Osaka, Japan), following the manufacturer's instructions. Ten microliters of the thawed samples were used for the BCA assay to measure the total protein concentration in each sample. A BCA protein assay kit (Thermo Fisher Scientific, Inc., Hanover Park, IL, USA) was used according to the manufacturer's instructions. Absorbance at 560 nm was measured using a multilabel plate reader (ARVO; PerkinElmer, Tokyo, Japan). Based on the absorption spectra, the protein concentration in each sample was calculated. Protein permeability was calculated by dividing the measured protein concentration by the protein concentration in 30% FBS. 

Animal Experiments

All animal experiments were conducted in compliance with ARVO Statement for the Use of Animals in Ophthalmic and Visual Research and the University of Tokyo Animal Care and Use Committee guidelines. Male New Zealand white rabbits weighing 3.5 to 4.0 kg were deeply anesthetized by intramuscular injection of ketamine (50 mg/kg body weight) and xylazine (15 mg/kg body weight). Topical anesthesia was induced by 0.4% oxybuprocaine eyedrops in the right eye of each rabbit before and during surgery. The rabbits were killed by intracardiac injection of an overdose of pentobarbital. 

Biocompatibility Testing

Stromal implantations of the CV-CERT were performed in the right eyes of four rabbits. Through an incision made using a 250-μm knife (Accurate Depth BD Medical, Waltham, MA, USA) at the peripheral cornea, lamellar dissection of the midstroma was performed using Deep lamellar endothelial keratoplasty corneal dissectors (Katena Eye Instruments, Denville, NJ, USA). The CV-CERT punched out by an 8-mm biopsy punch (Kai Industries Co., Ltd., Seki, Japan) was implanted intrastromally through the incision. Anterior chamber (AC) implantation of the CV-CERT was performed in the right eyes of four rabbits. A 3-mm CV-CERT was inserted into the AC through a 3-mm corneal incision and the CV-CERT was attached to the posterior cornea by air tamponade. One suture was placed at 12 o'clock on graft and cornea by 10-0 nylon. The corneal incision was sutured by 10-0 nylon. The rabbits were followed up by slit-lamp examination and photographed every week until postoperative month (POM) 1 and every month until POM 4. Rabbits were killed at POM 4 and the globe was enucleated and fixed in 10% formalin. After embedding the samples in a paraffin block, they were sectioned and stained with hematoxylin and eosin (HE). 

Adhesiveness Evaluation

The adhesiveness of the CV-RM with UV irradiation and the CV-CERT were compared by implanting them onto the corneal endothelial side of the right eyes of three rabbits in each group. A corneal incision was made at the limbus by a slit knife and heparin was injected intracamerally. Each membrane was inserted into the AC and air tamponade was performed to attach the membrane to the corneal posterior surface. Descemet's membrane of the rabbit cornea was not stripped off. No sutures were used for the grafts. The wound was sutured using 10-0 nylon. Postoperative observations and photography were performed by slit-lamp examination every week until POM 1. 

Cultivating hCECs on the CV-CERT

We evaluated the hCEC seeding technique on curved CV-CERT. We used normal hCECs obtained from donor cornea imported from SightLife (Seattle, WA, USA) for research use. The cell cultivating method and growth medium conditions were described previously. ¹⁵ Briefly, after collagenase A (Sigma-Aldrich Corp., St. Louis, MO, USA) digestion of Descemet's membrane obtained from donor corneas and incubation in trypsin/EDTA (Sigma-Aldrich Corp.), the cells were seeded on an atelocollagen-coated dish. The culture medium contained low-glucose Dulbecco's modified Eagle's medium (Wako Pure Chemical Industries, Ltd.), 10% FBS, 20 ng/mL of human basic fibroblast growth factor (Wako Pure Chemical Industries, Ltd.), 0.3 mmol/L ascorbic acid 2-phosphate (Wako Pure Chemical Industries, Ltd.), and antibiotic/antimycotics (Sigma-Aldrich Corp.). Human CECs at passages three through five were used in this experiment. 

A novel hCEC seeding technique was used to evenly distribute the cells on the bowl-like shape of the CV-CERT (Fig. 3). The PBS-rehydrated CV-CERT was attached to the bottom of each well of a 12-well plate and dried again in a safety cabinet to flatten the CV-CERT onto the culture plate (Fig. 3A). The hCECs were seeded at 1.3 × 10⁶ cells/well on the CV-CERT attached well and cultured overnight. The next day, the CV-CERT was gently detached from the well bottom and floated on a polytetrafluoroethylene O-ring with a 9.8-mm inner diameter and a 1.9-mm width, which was placed in another well (Fig. 3B). Cultivation continued in this manner for 1 week. The CV-CERT with cultivated hCECs was then prepared for histologic evaluation. Staining with 4′,6-diamidino-2-phenylindole (DAPI), alizarin Red S, and HE was performed. In the CV-CERT after hCEC seeding, the number of cells in a 0.1 × 0.1 mm square was counted at four different sites in DAPI images using Java-based imaging software (ImageJ; National Institutes of Health [NIH], Bethesda, MD, USA, provided in the public domain at http://imagej.nih.gov/ij/) and the mean cell count per square millimeter was calculated. The CV-CERT thickness was also measured from HE stained images using Java-based imaging software (NIH). 

Figure 3

 

Culturing technique for HCEC on CV-CERT with a spherically curved shape. (A) A PBS-rehydrated CV-CERT was attached to the well bottom of a 12-well plate and dried in a safety cabinet. Then hCECs were seeded at 1.3 × 10⁶ cells per well in wells containing the CV-CERT. (B) After overnight cell cultivation, the CV-CERT was gently detached from the well bottom and floated on a polytetrafluoroethylene O-ring, which was placed in another well. Cell cultivation was then continued for 1 week.

Results

Effect of UV-Irradiation Conditions on CV-RM Transparency

Transparency measurements of CV-RMs without UV irradiation or with various total doses of UV irradiation are shown in Figure 4. The absorbance spectra of CV-RMs without UV irradiation (Fig. 4A, gray line) and with UV-irradiation at a total dose of 2400 mJ/cm² (600 mJ/cm² × 4; Fig. 4B, black line) were remarkably increased in the UV region lower than 300 nm, although they gradually decreased in the visible regions from 700 to 400 nm. The CV-RM with UV irradiation at a total dose of 2400 mJ/cm² (600 mJ/cm² × 4) was the most transparent (Fig. 4B). This dose was then used to manufacture the CV-CERT. 

Figure 4

 

Effect of UV irradiation condition on the transparency of CV-RM. (A) The absorbance spectrum at a 200- to 700-nm wavelength was measured for CV-RMs without UV irradiation (gray lines) and with UV irradiation (black lines) at the total dose of 2400 mJ/cm² (600 mJ/cm² × 4). (B) The transparency of the CV-RM after various levels of UV irradiation was examined using a spectrophotometer and absorbance was measured at a 400-nm wavelength. Three samples for each condition were used. Error bars and asterisks indicate SD and P < 0.01 in Student's t-test, respectively.

CV-CERT Manufacture

The CV-CERTs were manufactured as follows. Based on the above results, the revitrified membrane on the convex surface was irradiated with UV at a dose of 600 mJ/cm² and rehydrated, and then transferred to the concave surface, revitrified, irradiated with UV at a dose of 600 mJ/cm², and rehydrated. The UV-irradiation procedure was repeated once to achieve a total dose of 2400 mJ/cm². 

Membrane Permeability Test

The SDS-PAGE and BCA assay results are shown in Figures 5A and 5B, respectively. In SDS-PAGE, bands the size of albumin (∼66 kD) were observed in both CA-CV and CV-CERT after 1 hour. The protein permeability of the CV-CERT was higher than that of CA-CV at 1 day (Fig. 5B). At 5 days, the protein permeability of CV-CERT was significantly higher than that of CA-CV (P = 0.032, unpaired t-test). Three CA-CV and five CV-CERT membranes were used in this assay. 

Figure 5

 

Results of membrane permeability test. (A) Silver stained membrane after SDS-PAGE. Bands on the left show the results of CA-CV samples and those on right show the results of CV-CERT. Bands the size of albumin (66 kD) were detected in both membranes, suggesting good membrane permeability. (B) Time-course of membrane permeability of CV-CERT and CA-CV measured by BCA. The vertical axis shows the membrane permeability (%) versus 30% FBS protein concentration. Membrane permeability at 5 days was significantly different between the two groups. *P = 0.032, unpaired t-test. Error bars represent standard errors.

Biocompatibility Test in a Rabbit Model

Rabbit corneas in which the CV-CERT was implanted in the stromal interlayer were transparent (Fig. 6A, left) without apparent edema (Fig. 6A, center) during the 4-month study period. We observed neither epithelial thinning nor cell infiltration (Fig. 6A, right). These findings indicate that the CV-CERT has high biocompatibility when implanted intrastromally and had no adverse effects on the corneal stroma and epithelium. Findings of the CV-CERT attached on the surface of the corneal endothelium in the rabbit AC are shown in Figure 6B. The CV-CERT developed no melting or opacification (Fig. 6B, left and center) and no evident cell infiltrations were observed (Fig. 6B, right). These findings indicate that the CV-CERT has high biocompatibility in the AC of rabbit eyes. Four rabbits were used in each group. 

Figure 6

 

Biocompatibility testing of CV-CERT in a rabbit model. (A) Implantation of CV-CERT into the stroma of the rabbit cornea. Slit-lamp images (left, center) show that the CV-CERT was clear and caused no epithelial defects or inflammation up to POM 4. Hematoxylin and eosin staining (right) shows no cell infiltration around the CV-CERT and intact multi-layered epithelial cells. Arrows indicate the CV-CERT in the rabbit corneal stroma. (B) Implantation of the CV-CERT into the anterior chamber of the rabbit eye. Slit-lamp images (left, center) show no inflammation or opacification of either the CV-CERT or cornea. No epithelial defects or corneal precipitates were observed up to POM 4. Hematoxylin and eosin staining (right) showing the intact epithelium and epithelial cell growth over the CV-CERT. Arrows indicate the CV-CERT attached to the rabbit corneal endothelial side.

Adhesiveness Evaluation

Postoperative images of rabbit eyes in which the CV-RM or CV-CERT was implanted are shown in Fig. 7. The top images were obtained just after surgery and those on the bottom were obtained at POM 1. While the flat CV-RM developed large folds just after surgery (Fig. 7A, top), the spherically-curved CV-CERT fit the cornea without any folds (Fig. 7B, top). The flat CV-RM dislocated after the air dissipated (Fig. 7A, bottom), whereas the spherically curved CV-CERT remained attached to the corneal endothelial side (Fig. 7B, bottom). The right eyes of three rabbits were used in each group. 

Figure 7

 

Comparisons of flat CV-RM (A) and a spherically curved CV-CERT (B) implanted onto the rabbit corneal posterior surface. Operative microscopic images just after the air tamponade (top) and slit-lamp images at POM 1 (bottom). The flat CV-RM shows large folds following air tamponade ([A], top) and it became dislocated over time ([A], bottom). On the other hand, the spherically-curved CV-CERT fit to the cornea without folds ([B], top) and remained attached until POM 1 ([B], bottom).

hCECs Cultured on CV-CERT

Images of hCECs cultured on CV-CERT using our technique are shown in Figure 8. A single layer of hCECs was observed from edge to edge on the CV-CERT (Fig. 8, left). The thickness of CV-CERT was 21.2 ± 0.9 μm, when measured in HE images. An even and regular distribution of hCECs was observed and the mean cell density was 2650 ± 100/mm² (Figs. 8, middle, right). 

Figure 8

 

Images of hCECs cultured on CV-CERT using our newly developed technique. A single layer of hCECs was observed from edge to edge on the CV-CERT (left). The CV-CERT thickness was 21.2 ± 0.9 μm. The hCECs were evenly and regularly distributed and the mean cell density was 2650 ± 100/mm² (middle, right).

Discussion

In the present study, we developed a new collagen vitrigel, called the CV-CERT, as a CEC carrier for corneal regenerative medicine. The CV-CERT can be stored at room temperature in dried form just like conventional CV for basic research. It comes with a nylon frame around the membrane for easier manipulation. Surgeons can punch it out to create the desired diameter up to 11 mm using a dermal punch. The CV-CERT has several unique characteristics. First, it is made from porcine-derived collagen using a completely serum-free process, avoiding the risk associated with ruminant-derived material. As porcine heart valve replacements in humans are becoming more common, ¹⁶ porcine-derived material is considered to be safer than ruminant-derived tissue for use in humans. Second, we used atelocollagen, which is considered less antigenic than native collagen. The CV-CERT showed no toxicity and caused no inflammation in rabbit eyes for up to 4 months. Third, it is transparent but rigid. In addition to the vitrification process, UV irradiation was used to increase the mechanical strength of the comparably weak atelocollagen while maintaining the transparency. Fourth, it has a spherically curved shape. The newly developed molds were designed to make the vitrigel curved to fit to the corneal shape. 

The problem with atelocollagen is that it is less rigid than native collagen and thus both transparency and rigidity are difficult to achieve in an atelocollagen membrane. To compensate for the mechanical weakness of atelocollagen, UV irradiation was performed to induce the cross linkage of collagen fibrils. We determined the optimal amount of UV irradiation by applying different amounts of UV light to porcine atelocollagen gels. Based on our results, a total UV irradiation of 2400 mJ/cm² led to better transparency and adequate mechanical strength. The transparency of the CV-CERT was maintained for up to 4 months in rabbit eyes. In addition, when implanting the CV-CERT into rabbit eyes, the surgeon could easily handle the membrane without breaking it and had no difficulty expanding it in the AC to attach it to the posterior corneal surface. As protecting CECs on the membrane is the most crucial part of the surgical procedure, adequate mechanical strength is very important for a successful surgical outcome. 

We measured the CV-RM light absorbance of a wide range (200–700 nm) of wavelengths. Interestingly, the light absorbance of the wavelength of the UV region (>300 nm) was remarkably higher than that of the visible light region. Because the cornea plays a major role of cutting off the UV light that gets into the eye, the fact that the CV acts similarly to the cornea in terms of UV absorbance is very important. 

The CV-CERT has high protein permeability of the CV-CERT. We used CA-CV as a control for the membrane permeability testing and it passed ∼150 kD protein, comparable with that in a previous report. ¹⁷ The protein permeability of the CV-CERT at 5 days was more than twice that of the CA-CV and the difference was statistically significant. As the corneal epithelium is supplied with nutrition by diffusion from the aqueous humor, low membrane permeability can cause epithelial thinning. ¹⁸ Thus, the fact that there was no epithelial thinning in rabbit eyes that underwent stromal CV-CERT implantation suggests that the CV-CERT had adequate membrane permeability. 

The curved shape is important for CEC carriers to adhere to the posterior surface of the cornea. The CV-CERT fit to the corneal posterior surface, whereas the flat UV-irradiated porcine atelocollagen membrane did not. Those two membranes were essentially made of the same materials and differed only in their shapes. Thus, the fact that only the CV-CERT attached to the corneal posterior surface indicates the importance of the spherical shape of the CV-CERT as a CEC carrier. We created a spherical curvature with an 8-mm radius on the CV-CERT. Although it fit well to the rabbit cornea, there might be concerns that the human corneal radius of curvature is different from that of rabbit, ¹⁹ and that the cornea is not completely spherical but actually flatter in the periphery. This will not have a big impact when the CV-CERT is applied to patients because, in the clinical situation, patients and donor corneas often have a different curvature of radii but the donor cornea usually fits the host cornea. Also, the CV-CERT shape could be custom-made, for example, aspheric or with different radii of curvature, by making different molds. 

An even and regular distribution of hCECs was observed on the CV-CERT despite its curved shape. Because the CV-CERT is spherically curved, cells would accumulate at the bottom if they were seeded in the usual manner. Our novel technique of seeding hCECs onto the CV-CERT while flattened on a culture plate and detaching the CV-CERT from the culture plate after overnight culture enabled the uniform distribution of the hCECs. The cell density of 2650/mm² is similar to that of normal human corneas, which is adequate for the pumping function of CECs. In this study, we did not perform actual hCEC sheet transplantation, but this is the next step. 

Some investigators perform intracameral CEC injection therapies for corneal endothelial dysfunction. Actually, our group previously performed a similar study (sphere injection). ²⁰ As the experiment proceeded, however, we faced several problems of AC cell injection, such as ectopic cell deposition, risk of angle block, and systemic dispersion of injected cells. Therefore, we concluded that cell sheet transplantation is a safer way to control cell positioning. On the other hand, another group injected cultured CEC intracamerally and used Rho-associated kinase inhibitor to enhance the attachment of injected cells to posterior surface of the cornea. ²¹ Use of agents to enhance the attachment of injected cells to the corneal rear surface may solve the problems we observed in our previous study. Future clinical studies of cell injection therapies and cell sheet transplantation therapies will elucidate the safest and most effective modality for the treatment of corneal endothelial dysfunction. 

The present study has several limitations. First, although the CV-CERT showed no toxicity and did not induce an immune reaction in rabbit eyes, this does not necessarily mean it is safe when applied to humans. In one report, intracutaneous injection of bovine atelocollagen caused an immune reaction in patients, ²² although as the cornea is immune-privileged and the reaction in the cornea should be lower than that in the skin. Second, the hCEC culture medium contains FBS. Applying hCECs cultured with FBS to humans might not cause any problems because FBS has been used in the manufacture of various medications, such as vaccines, throughout the world. Ideally, however, an FBS-free culturing method would be developed to avoid potential unidentified risks. This issue will be investigated next. 

In conclusion, we developed a transparent and biocompatible porcine-derived atelocollagen vitrigel membrane with a spherical curvature and adequate mechanical strength for surgical use, which was optimized for exclusively corneal endothelial regenerative treatment. Transplantable artificial corneal endothelial grafts were created by combining cultured hCECs and the CV-CERTs. 

Acknowledgments

Supported by the Agri-Health Translational Research Project from the Ministry of Agriculture, Forestry, and Fisheries of Japan, Grant No. 6110 (TT for vitrigel research); and No. 6220 (SA for corneal research). The authors alone are responsible for the content and writing of the paper. 

Disclosure: J. Yoshida, None; A. Oshikata-Miyazaki, None; S. Yokoo, None; S. Yamagami, None; T. Takezawa, None; S. Amano, None 

Footnotes

 JY and AO-M contributed equally to the work presented here and should therefore be regarded as equivalent authors.