N-Cadherin in the Maintenance of Human Corneal Limbal Epithelial Progenitor Cells In Vitro

Abstract

purpose. To demonstrate the role of N-cadherin (N-cad) in maintaining the progenitor status of primary human limbal epithelial cells in vitro.

methods. Immunohistochemistry and immunoelectron microscopy against N-cad was performed in human limbal tissue. The expression of N-cad, cytokeratin (K) 12, and K14 was also observed in primary cultured limbal epithelial cell colonies in 3T3 feeder cells, which also express N-cad. Laser microdissection of individual colonies was performed to separate central cells from peripheral cells to compare secondary colony formation. Finally, an artificial small interfering RNA (siRNA) expression vector was used to downregulate N-cad in 3T3 cells (N-cad^low 3T3) to observe changes in colony formation and cultivated epithelial sheet phenotype.

results. N-cad was expressed in clusters of basal limbal epithelial cells. Limbal epithelial cell colonies cocultured with N-cad⁺ 3T3 feeder cells showed N-cad expression along the edge of each colony. When individual colonies were divided into peripheral and central sections by laser microdissection, peripheral cells had a significantly higher secondary colony-forming efficiency than did central cells. Furthermore, colonies using N-cad^low 3T3 cells were significantly smaller than mock-transfected cells. Then a duplex-feeder model with two layers of either 3T3 or N-cad^low 3T3 cells was used to produce stratified epithelial sheets. Only N-cad⁺ 3T3 cells produced epithelial sheets with basal K15/ suprabasal K12 expression observed in limbal tissue.

conclusions. N-cad plays a pivotal role in the maintenance of the progenitor phenotype in cultured limbal epithelial cells.

The microenvironment surrounding stem cells is unique to each type of stem cell; however, the requirement of a supporting niche cell is a common observation from the Drosophila ¹ to humans. Stem cells residing in the niche sustain the long-term repopulation of a specific tissue by undergoing self-renewal and by producing progenies of increasingly differentiated cells. Of all adult stem cell lineages, hematopoietic stem cells (HSCs) are the best described; a significant body of knowledge exists concerning the regulation of their self-renewal and differentiation. A subpopulation of osteoblasts plays an important role in the in the adult HSC niche. ^{2 3} In particular, recent studies showed that specialized spindle-shaped N-cadherin (N-cad)–expressing osteoblasts (SNOs) are a key component of the bone marrow stem cell niche, where HSCs interact with SNOs by way of N-cad interaction. ^{3 4 5 6}  

Hayashi et al. ⁷ recently reported that limbal basal epithelial cells in the peripheral cornea express N-cad as a possible marker of putative epithelial stem cells and that melanocytes may be associated with these cells through homotypic adhesion by N-cad in the human limbal epithelial stem cell niche. However, whether N-cad is a functionally important component of the stem cell niche in the limbus or in any other stem cell population outside the bone marrow remains to be shown. We therefore hypothesized that N-cad was functionally vital in the maintenance of the progenitor phenotype in limbal epithelial cells. With the use of colony-forming assays and the duplex feeder system, an in vitro model of the limbal epithelium, ⁸ we found that 3T3 feeder cells lacking N-cad expression significantly lost the ability to support clonal growth and maintain the basal K15⁺ limbal phenotype in epithelial sheets. 

Materials and Methods

Antibodies

Mouse monoclonal antibodies (mAbs) for N-cadherin, K15, and p63 were purchased from Invitrogen (3B9; Carlsbad, CA), Laboratory Vision (LHK15; Fremont, CA) and Calbiochem (4A4; San Diego, CA), respectively. Rabbit polyclonal antibody for K12 and K14 was purchased from TransGenic Inc. (Kumamoto, Japan) and Covance (Berkeley, CA), respectively. Isotype mouse IgG1, mouse IgG2a, and rabbit IgG were purchased from DakoCytomation (Carpinteria, CA) and Jackson ImmunoResearch Laboratories (West Grove, PA), respectively. FITC-, rhodamine-, and Cy3-conjugated secondary antibodies were purchased form Jackson ImmunoResearch Laboratories and Chemicon International Inc. (Temecula, CA), respectively. 

Preparation of Corneal Limbal Tissue

Donor corneas were obtained from the Northwest Eye Bank and preserved for experiments after central corneal buttons were used for transplantation. Limbal segments were embedded in optimum cutting temperature (OCT) compound (Tissue-Tek; Sakura Finetek, Tokyo, Japan), frozen in liquid nitrogen, and stored at −80°C. Frozen sections (5-μm thick) were used for immunohistochemical staining. 

Immunoelectron Microscopy

For postembedding immunogold labeling, limbal tissue specimens were fixed in 4% paraformaldehyde and 0.1% glutaraldehyde in 60 mM HEPES buffer (pH 7.4) for 1 to 2 hours at 4°C. Specimens were dehydrated serially to 70% ethanol at −20°C and embedded in resin (LRWhite; London Resin Co., Basingstoke, UK). 

Ultrathin sections of resin corneal tissue were incubated successively in drops of phosphate-buffered saline (PBS), 0.5% ovalbumin, and 0.5% gelatin in PBS, primary antibody diluted in PBS-ovalbumin overnight at 4°C, and 15 nm gold-conjugated goat anti-mouse IgG (Fc) (GE Healthcare UK Limited, Buckinghamshire, UK) diluted 1:10 in PBS-ovalbumin for 1 hour at room temperature. After rinsing, sections were stained with uranyl acetate and examined under an electron microscope (JEOL-1200 EXII; JEOL, Tokyo, Japan). 

Primary Limbal Epithelial Cell Culture

Limbal rims of donor corneoscleral tissue were prepared by careful removal of excess sclera, iris, and corneal endothelial tissue. Epithelial sheets were then isolated as described previously. ⁹ Dispersed epithelial sheets were treated with trypsin-ethylenediaminetetraacetic acid (EDTA) for 10 minutes to suspend cells, which were seeded onto mitomycin C (MMC)-treated NIH/3T3 feeder layer (3T3 feeder) in the four-well chamber slide or the upper chambers of a six-well culture inserts (Transwell; Costar Corning, Corning, NY) or 35-mm Petri dishes with cell support membrane (Cell Support Kit; Bio-Rad, Carl Zeiss Inc., Thornwood, NY) with supplemental hormonal epithelial medium (SHEM). ¹⁰  

Live-Cell Laser Microdissection

Limbal epithelial colonies prepared on 35-mm Petri dishes with cell support membrane (Carl Zeiss Inc.). Colonies were separated into peripheral cells (PCs) and central cells (CCs) by laser microdissection (Clonis; Bio-Rad, Carl Zeiss Inc.). The dissection area was calculated so that the total areas of PCs and CCs were the same. Separated colonies were treated with trypsin-EDTA for 10 minutes to suspend cells, which were reseeded onto MMC-treated 3T3 feeder layers for comparison by colony-forming efficiency (CFE) or prepared by using autosmear (CF-12D; Sakura Finetek) for cytospin immunocytochemistry. 

Immunostaining

Frozen sections, four-well chamber slides, and cytospin slides were fixed for 10 minutes in 2% paraformaldehyde (Wako, Tokyo, Japan). Chamber slides and cytospin slides were permeabilized cell membranes with 0.1% Triton X-100 (Sigma, St. Louis, MO) for 5 minutes at room temperature. Frozen sections and slides were blocked by incubation with 10% normal donkey serum (Chemicon International Inc., Temecula, CA) and 1% bovine serum albumin (Sigma) for 1 hour at room temperature. Antibodies to N-cad (1:100), K12 (1:100), K14 (1:100), K15 (1:100), and p63 (1:50) were applied and incubated for 90 minutes at room temperature, followed by incubation with FITC-, rhodamine-, or Cy3-conjugated secondary antibody. Isotype antibodies were used as negative controls. After three washes with TBST (0.825 mM Tris, 136.9 mM NaCl, 1.34 mM KCl, 0.1% Tween 20 [Sigma]), the sections were incubated with 1 μg/mL 4′,6-diamidino-2-phenylindole (DAPI; Dojindo Laboratories, Tokyo, Japan) at room temperature for 5 minutes Finally, sections were washed three times in TBST and coverslipped using an aqueous mounting medium (PermaFluor; Beckman Coulter, Marseille, France). 

Colony-Forming Efficiency

To evaluate the proliferative potential of cells in the colonies, 3T3 feeders were used in a CFE, assay as previously described. ^{11 12 13} 3T3 feeders in DMEM containing 10% FCS were plated at a density of 3 × 10⁶ cells in 100-mm culture dishes or 1.5 × 10⁵ cells in four-well chamber slides (Nalgene; Nunc International, Naperville, IL). Single cells were seeded at 1 × 10³ cells/100-mm dish or 1 × 10³ cells/well in SHEM containing 5% FCS. CFE was calculated by the percentage of colonies at day 14 generated by the number of epithelial cells plated in the dish. Colony size (mm²) and CFE (%) was quantified using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). Growth capacity was evaluated on day 14 when cultured cells were stained with rhodamine B (Wako) for 30 minutes. 

siRNA Transfection

We designed an artificial small interfering RNA (siRNA) that interferes with N-cad (5′-TGCAT GTGCTCTCAA GTGAAA-3′ (GenBank accession no. NM_0014711994.1) with the use of online software (BLOCK-iT RNAi Designer; Invitrogen), and ligated it into expression vector (Block-iT Pol II miR RNAi; Invitrogen). PcDNA6.2-GW/miR-neg control plasmid (mock) was used as control. Vectors with N-cad-siRNA plasmid were transfected into 3T3 cells using lipofectamine (Invitrogen) according to the manufacturer’s instructions. Cells expressing low levels of N-cad (N-cad^low) were selected in medium containing 30 μg/mL blasticidin and were used as feeder cells. 

Western Blot Analysis

N-cad^low 3T3 cells or mock-transfected 3T3 cells were dissociated with lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40 [Calbiochem, Darmstadt, Germany]) and were homogenated. Both types of 3T3 cells were incubated for 40 minutes at 4°C and then centrifuged at 15,000 rpm for 30 minutes at 4°C. Protein concentration of the supernatant was determined by the DC protein assay (Bio-Rad Laboratory, Hercules, CA). All samples were then diluted in 2× sample buffer (100 mM Tris-HCl [pH 6.8], 4% SDS (Invitrogen), 20% Glycerol (Wako), and 12% 2-mercaptoethanol (Wako) and were boiled. Ten micrograms of each sample were loaded on a 10% Bis-Tris gel (Novex NuPAGE; Invitrogen) and were transferred onto polyvinylidene difluoride membranes (Millipore, Billerica, MA). Membranes were blocked with 5% skim milk (Difco Laboratories, Detroit, MI)-1.5% normal donkey serum-PBS for 60 minutes at room temperature. Membranes were reacted with antibodies against N-cad (3B9) and β-actin (mabcam8226; Abcam, Cambridge, MA) for 60 minutes at room temperature. After three washes in TBST, donkey biotinylated anti–mouse IgG (Jackson ImmunoResearch Laboratories) was added for 30 minutes at room temperature. Protein bands were visualized with a reagent kit (Vectastain ABC Elite Kit; Vector Laboratories, Burlingame, CA) and substrate (DAB; Vector Laboratories). Semiquantitative densitometry of the bands was performed with ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). 

Epithelial Sheet Cultivation Using the N-cad^low 3T3 Duplex Feeder System

Stratified epithelial sheets were cultivated with the use of a duplex culture system, as previously described. ⁸ N-cad^low 3T3 cells or mock-transfected 3T3 cells were seeded at 2.5 × 10⁵ cells in both cell culture inserts (Transwell; Costar Corning) coated with fibrin (5.0 mg/mL; Bolheal; Fujisawa, Osaka, Japan) ¹⁴ in the bottom of the well. Dissociated limbal epithelial cells were plated at 2.0 × 10⁵ cells in culture inserts with SHEM-supplemented aprotinin (Wako) at 666 KIU/mL. The culture was submerged in medium for 1 week with SHEM, cultured at air-liquid interface for 1 week, and used for immunostaining. 

Results

N-cad⁺ Epithelial Cells Form Clusters in the Basal Limbal Epithelium

In donor limbal tissue, N-cad was sporadically expressed in the basal layer of the limbal epithelium (Figs. 1A 1B 1C 1D) . Double immunostaining revealed that N-cad⁺ cells coexpressed the progenitor markers K14 and K15 but expressed little K12, a differentiation marker for corneal epithelial cells (Figs. 1E 1F 1G) .Immunoelectron microscopy further showed that gold-labeled anti–N-cad antibodies stained cell-cell junctions in the basal limbal epithelium (Figs. 2A 2B) , suggesting homotypic expression of N-cad in these cells. Melanocytes, characterized by abundant melanin granules, also expressed N-cad (Figs. 2C 2D) . However, we were unable to observe direct interaction of epithelial cells with melanocytes in all the samples observed. 

Maintenance of Progenitor Cells in Colonies Requires N-cad

Immunocytochemistry of epithelial cell colonies cocultured with 3T3 feeder cells revealed that N-cad expression was localized at the edges of epithelial colonies in direct contact with feeder cells (Fig. 3) . K14 was uniformly expressed by epithelial cells (Figs. 3C 3D 3E 3F) ; however, K12 was sporadically expressed primarily toward the center of the colony (Fig. 3B) . Longitudinal sections show N-cad was expressed by 3T3 feeder cells and K14⁺ peripheral epithelial cells, suggesting that feeder cells and epithelial cells interacted by homotypic adhesion of N-cad (Figs. 3C 3D 3E) . The proliferative/progenitor marker p63 was expressed primarily by peripheral cells in the colony (Fig. 3F) . 

To determine whether the association with N-cad⁺ feeder cells was required for the preservation of progenitor cells, individual colonies were divided into central cells (CCs) and peripheral cells (PCs) by laser microdissection and were digested to observe secondary colony formation (Fig. 4A) . CCs, PCs, and limbal cells (Fig. 4B)were dispersed and observed by cytospin for N-cad expression. Only PCs and limbal cells contained N-cad⁺ epithelial cells (Fig. 4C) . PCs produced significantly more colonies than N-cad-CCs (Figs. 4D 4E) . To conclusively show the importance of N-cad in colony formation, we used siRNA technology to downregulate N-cad expression in 3T3 feeder cells (N-cad^low 3T3). N-cad expression was significantly lower in N-cad^low 3T3 cells by Western blot analysis (Figs. 5A 5B) . As expected, CFE using N-cad^low 3T3 cells was significantly lower than CFE by mock-transfected 3T3 cells (Figs. 5C 5D) . The average size of individual colonies was also significantly smaller in N-cad^low 3T3 cells (Figs. 5C 5E) . 

Reproduction of the Limbal Phenotype in Cultivated Sheets

We previously reported that using two layers of feeder cells (duplex feeders), one in contact with epithelial cells and one layer to maintain soluble factors, was required to reproduce the limbal phenotype of basal K15 expression in cultivated sheets. ⁸ Using this duplex feeder system, we found that cultivated sheets using N-cad^low 3T3 cells lost the orientation of K15⁺ cells in the basal epithelial layer (Fig. 6A) , whereas mock-transfected 3T3 cells were able to maintain basal K15⁺ epithelial cells (Fig. 6B)resembling the structure of the normal corneal limbus in eye bank eyes (Fig. 6C) . 

Discussion

The structured organization of the stratified corneal epithelium is required to maintain a functional barrier against invading organisms and to support the overlying tear film that supplies nutrients and acts as a lubricant during blinking. Substantial evidence shows that human corneal epithelial stem cells reside in the corneal limbus, a ring of tissue surrounding the peripheral clear cornea. ^{15 16} Although differentiated corneal epithelial cells express E-cadherin, the major junctional protein found in the five- to six-layer epithelium, clusters of cells in the limbus were shown to express N-cad (Fig. 1) . Recent reports have shown that N-cad, along with K15, can be used as a marker for corneal progenitor/stem cells. ^{7 17}  

Despite the recent advances in the characterization of limbal stem cells, little is known of the limbal stem cell niche compared with the well-characterized hematopoietic stem cell (HSC) niche. In the HSC niche, N-cad and β-catenin are asymmetrically localized between SNO cells and long-term HSCs. ³ Indirect evidence supports the importance of N-cadherin in the retention of HSCs in the endosteal niche. ^{18 19 20 21} In the corneal limbus, basal limbal epithelium is associated with melanocytes and receives melanin pigments through a yet to be discovered channel. ²² Because melanocytes also express N-cad, homotypic adhesion may occur between these cells. However, we were not able to find evidence of N-cad–mediated adhesion between melanocytes and basal epithelial cells. Immunoelectron microscopy showed that N-cad was expressed in the intercellular junction between basal limbal epithelial cells, suggesting that N-cad–mediated adhesion plays a role in maintaining basal limbal cells. 

We further determined whether N-cad was indeed functionally required to maintain epithelial progenitor cells. Feeder cells such as the 3T3 cells are often used in culturing epithelial clones from single cells. We demonstrated that N-cad⁺ cells in vitro were clustered at the circumference of each colony, where direct contact with feeder cells was formed (Fig. 3) . Peripheral colony cells were enriched with the progenitor phenotype because secondary colony formation by central colony cells was significantly smaller than by peripheral cells (Fig. 4) . Furthermore, when 3T3 cells transfected with N-cad siRNA were used as feeders, CFE and colony sizes were significantly reduced compared with mock-transfected cells (Fig. 5) . This shows that N-cad is vital in maintaining immature cells, at least in vitro. 

Demonstrating the function of N-cad⁺ epithelial cells in vivo would further enhance our understanding of the limbal stem cell niche. However, not only are N-cad (−/−) mice embryonically lethal, ²³ the localization of corneal epithelial stem cells in mice is not as clear as in humans. Because we performed the above experiments using human cells, we used an ex vivo limbal epithelial model engineered with a “duplex feeder” system recently reported by our group. ⁸ By using two layers of 3T3 feeder cells, one in contact with epithelial cells and another separate layer to provide soluble factors, we could reproduce the K15⁺/ K12⁺ phenotype found only in the basal limbal epithelium in vivo. As expected, when N-cad^low 3T3 feeder cells were used, engineered epithelial sheets did not contain appreciable levels of K15 expression compared with normal 3T3 feeder cells (Fig. 6) . The evidence strongly suggests that N-cad plays a vital role in the maintenance of K15⁺ epithelial stem/progenitor cells in the basal limbal epithelium. 

Our study presents data demonstrating the requirement of N-cad in preserving the limbal epithelial phenotype in vitro. We also confirmed that clusters of K12⁻ basal limbal epithelial cells express N-cad, indicating that N-cad also has a functional role in vivo. However, we were unsuccessful in identifying a putative “niche cell” in the limbus by electron microscopy, which formed homotypic adhesion with epithelial cells through N-cad. The weak point of our study is that we did not take into account other components of the limbal niche, such as blood vessels, Langerhans cells, stromal fibroblasts, and the basement membrane. All these environmental factors may support the phenotype of limbal epithelial cells and may even enhance the expression of N-cad by basal epithelial cells, as was the case in the cultured colonies in our study. There is also the possibility that N-cad functions to maintain clusters of progenitor cells in the local area of the limbal niche. Further investigations are required to elucidate the molecular mechanisms and cellular components involved in the maintenance of the corneal stem cell niche by N-cad. 

 Supported in part by a grant from the Advanced and Innovational Research program in Life Sciences from the Ministry of Education, Culture, Sports, Science and Technology of Japan (KT); a grant from the Advanced and Innovational Research program in Life Sciences from the Ministry of Education; and Grant-in-Aid for Scientific Research (H18-regeneration-young-002) from the Ministry of Health and Welfare, Japan (TK).

 Submitted for publication February 2, 2009; revised March 26, 2009; accepted July 6, 2009.

 Disclosure: K. Higa, None; S. Shimmura, None; H. Miyashita, None; N. Kato, None; Y. Ogawa, None; T. Kawakita, None; J. Shimazaki, None; K. Tsubota; None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

N-cad expression in the human corneal limbus. Eye bank donor corneas were stained with antibodies against the proteins indicated and DAPI nuclear counterstain (blue). (A) N-cad (red)–positive epithelial cells (arrows) were observed in clusters in the basal limbus and conjunctiva but not in the central cornea. (B–D) High magnification of N-cad⁺ cell clusters. (E) K12 (green) was observed in the superficial layers of the limbus but not in the basal N-cad⁺ cells. (F, G) N-cad⁺ cells expressed the epithelial keratin K14 (green) and the basal epithelial keratin K15 (green). Scale bars: (A) 100 μm; (E–G) 20 μm. N-cad, N-cadherin; K, keratin.

Figure 2.

 

Immunoelectron microscopy of limbal tissue. Eye bank donor tissue was examined by immunoelectron microscopy against N-cad. (A) N-cad was expressed along the intercellular junction of basal limbal epithelial cells. (B) High magnification of boxed area in (A). (C) Melanocytes with abundant pigmented granules were observed in the basal limbal area. (D) High magnification of boxed area in (C) reveals N-cad expression by melanocytes. A rrowhead: intercellular junction. Epi, limbal epithelial cell; St, stromal tissue; KC, keratocyte; Me, melanocyte. Original magnifications: (A, C) ×5000; (B, D) ×15,000.

Figure 3.

 

Immunostaining of N-cad and keratins of single human limbal epithelial cell colonies cocultured with 3T3 feeder cells. (A, B) Whole colonies were double stained with N-cad (red) and K14 (A, green) or K12 (B, green) antibodies. Epithelial cells in the periphery of the colony coexpressed N-cad, whereas K12, a marker of differentiated corneal epithelial cells, had a more sporadic staining pattern toward the center of the colony. (C–F) Cross-sections of single colonies stained with N-cad (C, red) and K14 (C, F, green) or p63 (F, red). (C) PCs expressed N-cad and were associated with 3T3 feeder cells that also expressed N-cad. (D, E) Enlarged view of (C). (F) Peripheral cells expressed higher levels of p63 than those in the center of the colony. Nuclei were labeled with DAPI (blue). Scale bars: (A, B) 200 μm; (C, F) 100 μm.

Figure 4.

 

Secondary CFE of laser-microdissected limbal epithelial colonies. (A) Phase-contrast image of a single limbal epithelial colony (dotted lines) divided by live-cell laser microdissection into PCs and CCs. (B) Phase-contrast image of the pigmented human limbal epithelium. (C) Immunocytochemistry of N-cad (red) and K14 (green) of cytospin samples from PCs (left), CCs (center), and freshly dissociated limbal cells (right). N-cad⁺ epithelial cells (arrowheads) were observed only in PCs and limbal cells. Nuclei were labeled with DAPI (blue). Scale bars: (A, B) 200 μm; (C) 20 μm. (D) Secondary CFE of PCs and CCs stained with rhodamine B. (E) Secondary CFE was significantly higher in cells initially from the periphery of colonies (P < 0.001; n = 5).

Figure 5.

 

siRNA downregulation of N-cad in 3T3 feeder cells. (A) Western blot of N-cad. Cell lysates of N-cad downregulated (N-cad^low) 3T3 feeder (right lane) and mock-transfected 3T3 feeder (left lane) cells were loaded. β-Actin was used as an internal control. (B) Relative expression of N-cad (N-cad/β-actin) protein was lower in N-cad^low 3T3 feeder cells than in mock-transfected cells (P < 0.05; n = 3). (C) Colony formation using N-cad downregulated (N-cad^low) 3T3 feeder (right) and mock-transfected 3T3 feeder cells (left). Colonies were stained with rhodamine B. (D, E) Both CFE area (D; P < 0.05; n = 3) and average colony area (E; P < 0.05; n = 3) were statistically smaller in epithelial cells cocultured with N-cad^low feeder cells.

Figure 6.

 

Immunohistochemistry of three-dimensional cultivated limbal epithelial sheets engineered from dissociated limbal epithelial cells and duplex feeder layers. (A) Epithelial sheets cultivated with N-cad^low feeder cells were thin and lost the limbal epithelial phenotype of K15⁺ basal cells. (B) Stratified sheets using mock-transfected 3T3 cells recreated the basal K15⁺ (red), suprabasal K12⁺ (green) pattern observed in normal limbal tissue. (C) Normal limbal tissue stained with K15 (green) and K12 (red) antibodies. Nuclei were labeled with DAPI (blue). Scale bar, 100 μm.

The authors thank Mifuyu Ishiwata and Tomomi Sekiguchi for technical assistance and the staff of the Cornea Center Eye Bank for administrative support.