Functional Limbal Epithelial Cells Can Be Successfully Isolated From Organ Culture Rims Following Long-Term Storage

Abstract

Purpose.: Because of a shortage of fresh corneal tissue for research, it was of interest to investigate the potential of successfully isolating human limbal epithelial cells (hLECs) from organ culture corneal-scleral (OCCS) rims.

Methods.: Superficial segments of corneal limbus were dissected and digested using collagenase (0.5 mg/mL, 16 hours at 37°C). Cell suspensions were separated into four different growth conditions: corneal epithelial cell medium (CM); CM + 3T3-Swiss albino cells; stromal stem cell medium (SM); and SM + 3T3 cells. Colony number, hLEC count, cell density, and colony forming efficiency (CFE) were quantified to assess different growth conditions. The expression profile associated with basal hLECs was assessed by immunofluorescence, and epithelial integrity was measured using our real architecture for 3D tissue (RAFT) corneal tissue equivalent.

Results.: Human limbal epithelial cells can be successfully isolated from OCCS rims following 4 weeks in storage with an 80.55% success rate with 36 corneal rims. Stromal stem cell medium + 3T3s provided optimal growth conditions. Colony number, total cell number, and cell density were significantly higher at day 7 in cultures with SM than in CM. There were no significant differences between SM and CM when assessing CFE and the expression profile associated with basal hLECs. Cells maintained in SM were found to produce a higher quality epithelium than that cultured in CM.

Conclusions.: Organ culture corneal-scleral rims can be a valuable source for hLEC. Using a combination of collagenase-based isolation and medium designed for stromal stem cell isolation, a high number of good quality hLECs can be cultured from tissue that would have otherwise been ignored.

Corneal maintenance and transparency are essential for normal vision and are facilitated by the continuous renewal of corneal epithelium. The superficial cornea is susceptible to a number of insults and injuries that can damage the limbal epithelial stem cells (LESCs) responsible for this physiological renewal process. The importance of continually isolating LESCs for research for stem cell therapies begs for more detailed studies of optimal isolation techniques. However, a worldwide shortage of donor corneas for transplantation reflects a shortage of corneal-scleral rims available to the research community. This problem is exacerbated by different methods used for storing corneas for transplantation,^1,2 with the preferred short-term storage method for cell isolation being less prevalent in most European eye banks.³ 

The two main methods of corneal storage adopted by European eye banks are organ culture and hypothermic storage. Organ culture involves long-term (3–4 week) storage in culture medium supplemented with fetal calf serum, antibiotics, and antimycotics at room temperature to 37°C, whereas hypothermic storage involves short-term maintenance (up to 7 days) in commercially available medium such as Optisol-GS (Bausch and Lomb, Rochester, NY, USA) at between 2°C and 8°C.² Although hypothermic storage is the most popular storage method worldwide, most European Eye Banks opt for organ culture storage due to extended storage time.³ Since 2009, the number of corneas stored in organ culture and processed at Moorfields Lions Eye Bank (London, UK) was 2.5-fold higher than that of hypothermically stored corneas (K Shah, 2014, personal communication with the manager of Moorfields Lions Eye Bank, London, UK). 

A major clinical requirement of corneal transplantation is healthy endothelium. Organ culture preservation of corneas for transplantation was therefore introduced as a way to monitor the stability of the endothelium during storage, which provides knowledge to eliminate any corneas that may not be successful for transplantation.^4–6 There are also a number of advantages to long-term donor tissue storage prior to surgery, such as offering more possibilities in terms of operation schedules, tissue type matching, and minimizing the waste of donor tissue, all of which are more restricted with hypothermic storage due to time constraints.² 

The research community prefers short-term stored or fresh tissue for human limbal epithelial cell (hLEC) and LESC isolation, both for research and for stem cell therapy purposes. However, a few studies have indicated the potential for using organ culture corneal-scleral (OCCS) rims to isolate limbal epithelial cells, suggesting a possible use for OCCS rims in hLEC isolation^7,8 and even limbal allograft transplantation.^9,10 All of these studies have focused on the limbal explant as a method for hLEC isolation. However emerging evidence also suggests a collagenase-based method as promising tool for hLEC isolation and expansion.^11,12 Chen et al.¹¹ first demonstrated the advantages of using collagenase to release basal progenitor cells from the limbus of fresh tissue in a culture method that does not rely on 3T3 cells.¹¹ In this study we investigated the use collagenase digestion as a method to isolate hLECs from OCCS rims. We also assessed the effects of different medium and the presence of a 3T3 feeder layer on the success of these cultures, with the aim of introducing a robust and reproducible method of hLEC isolation from organ culture rims. 

Methods

Materials

Cadaver donor corneal-scleral rims were obtained with appropriate research consent from Moorfields Lions Eye Bank. Ethical permission for this study was obtained from the Research Ethics Committee (UK, ref. no. 10/H0106/57-11ETR10), and all tissue was handled in accordance with the Declaration of Helsinki. Corneas were stored by organ culture method at ambient temperature after enucleation for 4 to 6 weeks before hLEC isolation. 

Human limbal epithelial cells were isolated in either corneal epithelial cell medium (CM) containing Dulbecco's modified Eagle's medium-F12 (DMEM:F12) basal medium (3:1 dilution), 10% fetal bovine serum, 1% antibiotic–antimycotic, epidermal growth factor (EGF; 10 ng/mL; Life Technologies Ltd., Paisley, UK), hydrocortisone (0.4 μg/mL), insulin (5 μg/mL), adenine (0.18 mM), transferrin (5 μg/mL), T3 (2 nM), cholera toxin (0.1 nM; Sigma-Aldrich, Dorset, UK), or in stromal stem cell medium (SM) containing DMEM-MCDB-201 (3:2 dilution; Sigma-Aldrich), 2% fetal bovine serum, penicillin (100 IU/mL), streptomycin (100 μg/mL), gentamycin (50 μg/mL), insulin (10 mg/mL), transferrin (5.5 mg/mL), selenous acid (6.7 ng/mL; 1× ITS; Life Technologies), albuMAX-I (1 mg/mL; Life Technologies Ltd.), dexamethasone (10 nM), EGF (10 ng/mL), l-ascorbic acid 2-phosphate (120 μM), cholera toxin (100 ng/mL), platelet-derived growth factor (10 ng/mL; Sigma-Aldrich). Isolation of hLECs was carried out with or without a feeder (F) layer of 3T3 cells (3T3-Swiss albino cells, catalog no. CCL-92; American Type Culture Collection, Manassas, VA, USA) whose growth was arrested with 4 μg/mL mitomycin C (Sigma-Aldrich) for 2 hours at 37°C. 

Isolation and Culture of hLECs

Isolation of hLECs was carried out using the same isolation method used for corneal stromal stem cell (CSSC) isolation.^13,14 For this method, a thin layer of superficial limbus containing the limbal crypts was trimmed away from the remaining stroma of OCCS rims (Fig. 1A), using fine sprung scissors. The superficial limbus was cut into 2-mm segments with a scalpel (Fig. 1B) before incubating with 0.5 mg/mL collagenase type-L (Sigma-Aldrich) for 16 hours at 37°C. Cells and tissue were dissociated by pipetting up and down with a 1-mL pipette, and the cell pellet was collected by centrifugation before being resuspended in the desired medium (CM or SM). Mixed populations of epithelial cells/clusters and stromal cells were cultured in six-well plates unless otherwise stated, with or without a feeder layer. All cultures were incubated at 37°C with 5% CO₂ in air, and medium was changed three times per week. 

Figure 1

 

Isolation of hLECs and CSSCs from OCCS rims. (A) The superficial limbus was cut away from the remaining stroma and (B) dissected into 2-mm pieces before overnight collagenase digestion. Scale bar: 400 μm. (C) Evidence of limbal crypts could be seen in superficial segments under the dissecting microscope. Scale bar: 100 μm. (D, E) Limbal epithelial clusters were visible after overnight collagenase digestion. (F) Colonies of hLECs started to grow after 24 hours in culture. (G, H) After 7 days in culture, epithelial colonies (black arrow) grew well with neighboring stromal cells (white arrow). Scale bars: 150 μm. (I) Corneal stromal stem cells (passage 1) could also be isolated from the same cultures in which hLECs were isolated. Scale bar: 400 μm.

For direct comparison of the dispase and collagenase isolation methods, OCCS rims were bisected into temporal and nasal segments in which half was treated with collagenase (as described above) and the other half with dispase. Dispase segments were incubated overnight in 1.2 U/mL dispase (Roche Pharmaceuticals, Welwyn Garden City, UK) at 4°C before hLECs were mechanically scraped into fresh CM by using forceps to generate a cell suspension. 

Rhodamine Staining

For rhodamine staining, cultures were fixed at day 7 or 14 in 4% paraformaldehyde (PFA; VWR, Soulbury, UK) for 10 minutes at room temperature before being stained with 1% rhodamine B (Sigma-Aldrich) for 10 minutes at room temperature. Cells were then gently rinsed in water five times and left inverted overnight to air dry. Images were captured using a digital camera and light box. 

Cell Density, Colony Number, and Total hLEC Number

Organ culture corneal-scleral rims were digested using collagenase, and each cell suspension was divided among the four different growth conditions. At day 7, 3 sets of photomicrographs were captured at random for each cell culture condition. The total number of cells in each field of view was counted using ImageJ software (National Institutes of Health [NIH], Bethesda, MD, USA), and data were expressed as cell number/mm². Cultures were also analyzed for total colony number by counting the number of colonies at day 7 using a microscope. Aborted colonies were also counted and identified as small and highly irregular and terminal colonies as originally described by Barrandon and Green.¹⁵ Total hLEC number was also counted and analyzed. Cells were trypsinized using 0.5% trypsin-EDTA (Life Technologies), after initial removal of 3T3 cells with 0.05% trypsin-EDTA, and counted using a hemocytometer. 

Colony-Forming Efficiency

OCCS rims were digested using collagenase, and each cell suspension was divided among the four different growth conditions. At day seven, hLECs were trypsinized and counted as above. Seven hundred and fifty cells from each growth condition were seeded onto new growth-arrested 3T3 feeder layers in 6 well plates and cultured for 7 to 10 days. Cells were fixed and stained with rhodamine B, images were captured using a light box, and colonies were counted using ImageJ software. Colony-forming efficiency (CFE) was expressed as the percentage of the number of colonies per cell plated using the following equation [CFE (%) = number of colonies/number of cells seeded × 100]. 

Statistical Analysis

Data are means ± SEM for a minimum of five OCCS rims. Statistical differences were calculated using Excel (Microsoft, Redmond, WA, USA) using Student's unpaired t-test to compare the CM-CM(F) to the SM-SM(F) group and the CM-SM group to the CM(F)-SM(F) group. A P value of <0.05 was considered statistically significant. 

Real Architecture for 3D Tissue (RAFT) Tissue Equivalent

RAFT tissue equivalents (TEs) were made by preparing CSSC-populated collagen matrices by using 80% v/v type I rat tail collagen solution (2 mg/mL; First Link, Birmingham, UK) and 10% vol/vol 10× minimum essential medium (Life Technologies). Corneal stromal stem cells (10% v/v) in SM were added after neutralization at a density of 1 million cells/mL. A volume of 2.2 mL of gel solution was cast into the wells of a 12-well plate and allowed to set for 30 minutes at 37°C. Confined gel compression was carried out as previously described,¹⁶ and the resulting tissue equivalents were submerged in 2 mL of CM or SM containing 1 million hLECs (pre-expanded with 3T3 feeder cells and SM for 14 days). At day 7, RAFT TEs were transferred epithelial side up onto transwell inserts (Millipore, Watford, UK), and placed in 6-well plates. The bottom chamber was filled with 1 mL of medium (CM or SM), and medium was changed every other day for a week. 

Immunofluorescence

Cells were fixed using 4% PFA for 10 minutes (cells on plastic) or 30 minutes (cells on RAFT-epithelial clusters) at room temperature. Samples were washed (3× for 5 minutes each) with Dulbecco's phosphate-buffered saline (DPBS), followed by a 10-minute (cells on plastic) or 30-minute (cells on RAFT/epithelial clusters) incubation in 0.5% Triton X-100 in DPBS (DPBS-T). Nonspecific binding sites were blocked with 5% goat serum in DBPS-T for 1 hour at room temperature. Samples were incubated overnight at 4°C in primary antibodies (p63-α, 1:100 dilution, Cell Signaling Technology, Danver, MA, USA; cytokeratin 3, mouse anti-keratin K3/K67 monoclonal antibody, 1:200 dilution, Millipore; Pax6, Pax-6 polyclonal antibody, 1:100 dilution; vimentin, antivimentin antibody [SP20], 1:100 dilution; pancytokeratin [PCK], anti-pancytokeratin [AE1/AE3], 1:50 dilution, Abcam, Cambridge, UK) diluted in 1% goat serum DPBS-T. Samples were washed and incubated in secondary antibody Alexa fluor 594 (1:1000; Life Technologies) with Phalloidin-Tetramethylrhodamine B isothiocyanate (TRITC, 50 μg/mL; Sigma-Aldrich) for 1 hour at room temperature. Cells were washed and mounted on slides with Vectorshield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Labs, Peterborough, UK) and sealed with nail varnish for imaging on an LSM 710 model confocal microscope (Zeiss, Oberkochen, Germany) using Zen software (ZenSoftware.com). 

Results

CSSC Isolation Technique Is a Useful Tool for hLEC Isolation

A technique used to isolate CSSCs¹⁴ was also found to be a valuable method for hLEC isolation. By digesting the superficial layer of the limbus (Figs. 1A, 1B) in which limbal crypts are evident (Fig. 1C), cell suspensions containing a mixed population of limbal epithelial cells or clusters (Figs. 1D, 1E) and limbal stromal cells (among other undefined cell types) were generated. Colonies of hLECs were evident after 24 hours in culture (Fig. 1F) and continued to expand with stomal cells lining the edges of colonies (Fig. 1G). After 7 to 14 days in culture, hLEC colonies maintained a corneal epithelial phenotype (Fig. 1H). Before stromal cells became confluent, CSSCs could be selectively trypsinized to produce pure CSSC cultures from the same OCCS rim (Fig. 1I). 

Limbal epithelial clusters isolated using collagenase digestion expressed both vimentin and PCK, suggesting the presence of both mesenchymal cells and epithelial cells (Fig. 2). It is clear to see the advantage of using collagenase over dispase to isolate hLECs from OCCS rims when directly comparing these isolation methods (Fig. 3A). This agrees with previously reported data comparing dispase and collagenase isolation methods.¹¹ After 14 days in culture using the collagenase isolation method, hLECs covered more than 50% of a 55-mm² culture dish (Fig. 3B, top). This was reproducible in most samples, with 80.55% of 36 organ culture corneal-scleral rims producing epithelial colonies and the remaining 19.45% of samples failing to produce epithelial colonies but instead producing a confluent layer of stromal cells (Fig. 3B, bottom). No correlation was found between successful cultures and donor characteristics. 

Figure 2

 

Expression of pancytokeratins and vimentin in limbal epithelial clusters. (A) Low resolution micrograph showing overall expression of pan-cytokeratin (PCK, green) and vimentin (Vim, red). Vimentin expression is evident throughout the cluster. Scale bar: 200 μm. (B) High-resolution micrographs show Vim-positive and PCK-negative cells among PCK-positive and Vim-negative cells (arrows). Scale bar: 50 μm. (C) Line scan z-stack showing a cross section of a limbal epithelial cluster. Vim-positive and PCK-negative cells can be seen lying in a basal position to Vim-positive and PCK-negative limbal epithelial cells. Scale bar: 20 μm.

Figure 3

 

Effects of different isolation techniques on hLEC culture were assessed. (A) Organ culture corneal-scleral rims were bisected into nasal and temporal segments and subjected to either dispase or collagenase digestion. Cell suspensions were plated into 6-well plates and cultured for 7 days. The collagenase method of isolation was superior in terms of hLEC coverage, colony number, and hLEC morphology. Scale bar: 160 μm. (B) Organ culture corneal-scleral rims were subjected to collagenase digestion, plated in 55-mm² culture dishes, and cultured for 14 days. Successful isolation (top) of hLECs using the collagenase method occurred in 80.55% of cultures, and failure of hLEC isolation (bottom) occurred in 19.45% of cultures (36 OCCS rims).

Effects of 3T3 Feeders and SM on hLEC Isolation and Culture

Due to the observations of successful hLEC isolation from long-term stored OCCS rims, it was important to investigate the differences between SM (designed for isolation CSSCs)¹³ and CM (the medium we routinely use for hLEC isolation). Following collagenase digestion of OCCS rims, each cell suspension was separated into four different growth conditions: (1) CM only, (2) CM + 3T3 cell feeder layers; (3) SM only; and (4) SM + 3T3 feeders. Figure 4 shows that, although initial hLEC colonies were slightly larger in CM and SM without 3T3 feeders at day 1 (Fig. 4, top), after 7 days in culture, these differences were eliminated (Fig. 4, middle). Colonies cultured in SM, either with or without 3T3 feeders, were more successful and more organized than those cultured in CM with or without 3T3 feeders. Furthermore, rhodamine staining at day 14 (Fig. 4, bottom) showed a high number of aborted colonies in CM compared to SM growth conditions, suggesting that SM is preferable for hLEC isolation. 

Figure 4

 

Effects of different cell medium types and 3T3 feeders on hLEC colony formation were investigated. Collagenase digestion of OCCS rims was carried out, and cell suspensions were plated on 55 mm² dishes. Representative photomicrographs of colonies at days 1 and 7 and 14 are shown. Initial colonies formed at day 1 were similar in size and morphology, with slightly smaller colonies present in cultures with 3T3 feeders. Scale bar: 325 μm. By day 7, there was a noticeable difference in colony size. Cultures maintained in SM appeared to thrive more than those maintained in CM. Scale bar: 1755 μm. Similarly, on day 14, epithelial colony size and coverage were noticeably higher in cultures maintained in SM than in those maintained in CM.

Analysis of cell growth was also carried out for these four conditions. Figure 5A shows the total number of colonies counted for each growth condition at day 7 of culture. A significantly higher number (Student's t-test; P < 0.05) of total colonies were counted in CM[F]) compared to CM alone, and in SM[F]) compared to SM alone. A significantly higher number (Student's t-test; P < 0.05) of total colonies was also counted in CM(F) cultures than SM(F) cultures. Although there were no significant differences in total colony counts between CM(F) and SM(F) cultures, when aborted colony number was counted, there were significantly more aborted colonies in CM(F) cultures than in SM(F) cultures (Fig. 5B). This was reflected in total hLEC count in that there were significantly more hLECs counted in SM(F) cultures than in CM(F) cultures (Fig. 5C). There was also a significantly higher number of hLECs in SM cultures than in CM cultures and in CM(F) cultures than in CM cultures. Overall these data suggest a more successful hLEC isolation in SM than in CM and that hLEC isolation favors the presence of 3T3 feeder cells. A summary of the percentage of successful hLEC isolations from OCCS rims in this particular experiment is represented in Figure 5D. 

Figure 5

 

Isolation of hLECs from OCCS rims favored SM over CM and 3T3 feeders (F) over no feeders. (A) Total hLEC colonies were counted across 8 different donors at day 7 of culture. Total colony count was significantly higher in cultures with 3T3 feeders than in those without. The number of hLEC colonies was also significantly higher in SM cultures than in CM cultures. (B) Although there were no statistical differences between total colony number of CM(F) and that of SM(F), there was a significantly higher number of aborted colonies in CM(F) cultures than in SM(F) cultures. (C) In agreement with graph B, there was also a significantly lower number of cells counted in CM(F) cultures than in SM(F) cultures. Cell number was also significantly higher in culture maintained in SM than in CM and in CM(F) compared to that in CM. (D) Percentage of successful isolations are compared to failed isolations in eight OCSCS rims. Isolations were considered a success if the colony number was greater than zero and a failure if colony number was zero. Data are means ± SEM for eight OCCS rims (*P < 0.05; **P < 0.01; Student's unpaired t-test).

hLECs Isolated From OCCS Rims Are a Potential Source of LESCs

Cell size was analyzed by counting confluent patches of cells to give an overall cell density. Cells on three separate photomicrographs per condition per n, were counted using ImageJ software. Figure 6 shows the different cell densities produced using different growth conditions. Stromal stem cell medium yields higher cell densities per unit area than CM indicating the maintenance of small, poorly differentiated hLECs (Figs. 6A, 6C). However, no significant differences were found among these four conditions when assessing the colony forming efficiency of the cultures (Figs. 6B, 6D). 

Figure 6

 

Stromal stem cell medium produced epithelial cells with a higher cell density than that with CM. (A) Cell density was calculated based on the premise that a higher cell density would result in smaller cells. Following 7 days in culture, three photomicrographs of confluent patches of epithelial cells were produced per growth condition for each donor, and all cells in each field of view were counted. Stromal stem cell medium cultures produced a significantly higher cells density than CM cultures. Data are means ± SEM for eight OCSCS rims (**P < 0.01; Student's unpaired t-test). (B) Representative photomicrographs show confluent patches of hLECs. Scale bar: 200 μm. (C) Colony-forming efficiency was compared among different conditions, and no significant differences were found among each. Data are means ± SEM for five OCCS rims. (D) Images show representative colonies for each condition.

Protein expression associated with limbal basal epithelial cells (p63α) and terminally differentiating suprabasal cells (CK3) were assessed in cells isolated with 3T3 feeders in the presence of either CM or SM (Fig. 7). Positive expression for p63α and negative expression of cytokeratin 3 (CK3) corresponded to poorly differentiated hLEC profile, with the exception of a small amount of CK3 staining indicating the presence of larger more differentiated epithelial cells. Pax6 was used as a corneal epithelial cell marker and was positively expressed in hLECs maintained in CM and SM. 

Figure 7

 

Protein expression levels of hLECs isolated in SM and CM were investigated. Immunofluorescence confirmed universal expression of p63a in nuclei of hLECs, and positive expression of the differentiation marker was detected only in a few large, more differentiated, suprabasal cells (arrows). Universal expression of the corneal epithelial marker pax6 was also detected in the nuclei of hLECs. No differences were observed between protein expression of hLECs isolated in the presence of CM or SM. Scale bar: 200 μm.

hLECs Grown in SM Support a Corneal Phenotype

Pre-expanded hLECs isolated using SM and 3T3 feeders were seeded onto RAFT TEs to assess their potential to form a multilayered epithelium. Interestingly, even though hLECs had been pre-expanded under ideal conditions (SM + 3T3s), shape and epithelial multilayering were found to differ between RAFT TEs maintained in SM and those in CM (Fig. 8). Although RAFT TEs maintained in SM displayed a typical corneal phenotype, with columnar basal epithelial cells and overlying squamous epithelial cells. RAFT TEs maintained in CM displayed a multilayering phenotype but lacked structure in terms of basal columnar cells and superficial squamous cells. Differences also became apparent when observing the basal epithelial layer of each RAFT TE (Fig. 8). RAFT TEs maintained in SM presented a tight layer of basal cells with highly organized actin borders compared to the irregular and differentiated basal cells observed in RAFT TEs maintained in CM. Epithelial cells on RAFT TEs maintained in SM also display a poorly differentiated hLEC profile when observing p63α, CK3, and Pax6 expression patterns (Fig. 8B). 

Figure 8

 

Limbal epithelial cells displayed a more cornea-like phenotype in 3D when maintained in SM. Epithelial cells were expanded in the presence of SM for 10 days before being seeded onto RAFT TEs. RAFT TEs were maintained either in CM or SM for 7 days and then airlifted for 7 days before fixing and staining. (A) Representative confocal z-stack projections and cross sections (CS) show f-actin (red, phalloidin) and nuclear (blue, DAPI) staining. Epithelial cells maintained in the presence of SM showed a more uniform corneal phenotype and were more likely to be multilayered then cells grown in the presence of CM, which displayed larger cells with less uniform morphology and more differentiated phenotype. Scale bar: 45 μm. (B) Immunofluorescence confirmed universal expression of p63α and pax6 in nuclei of hLECs, and expression of the differentiation marker CK3 was undetected.

Discussion

Availability of donor corneal-scleral rims is essential for the progression of basic research in the cornea and research in stem cell therapies for the treatment of corneal disease. Naturally, there is a preference for fresh corneal tissue in the research community as long-term tissue storage has been shown to have a negative impact, for example, on the success and speed of hLEC isolation.^17,18 However, now that organ culture has become the corneal storage method of choice in most European eye banks, not only have supplies of fresher tissue for research become limited, but also surplus OCCS rims that could be used for research are being overlooked. Hence, it was advantageous to attempt to optimize hLEC isolation from OCCS rims. In this study, we demonstrated that functional hLECs can be isolated successfully and reproducibly cultured from OCCS rims by using collagenase to digest the superficial limbus. 

Several studies have shown that hLECs can be cultured from OCCS rims using tissue explants^{7–10,17–21} or by isolation of a single-cell suspension, using a combination of dispase and trypsin.^8,18–20 A direct comparison between cell suspension and explant culture of limbal epithelial cells showed that the former generated a significantly higher number of stem cells than the explant technique.^19,20 In these studies, dispase was used to generate the epithelial cell suspension. However, recent studies by Chen et al.¹¹ have shown that using collagenase to generate cell suspensions is a valuable technique for isolating LESCs. This study shows that using dispase and trypsin for LESC isolation separates the basal progenitor cells from their supporting niche cells leading to less successful isolations. Therefore, collagenase isolation may have potential to supersede both the explant method and the dispase/trypsin method of isolation in terms of producing a reliable source of LESCs. Our findings demonstrate that by adapting the collagenase based method used for CSSC isolation^13,14 and shifting the focus to hLEC isolation; functional hLECs can be successfully and reproducibly isolated from OCCS rims that have been stored for 4 weeks or more. 

Interestingly, when we isolated hLECs in medium that was initially designed for mesenchymal stem cells²² and adapted for CSSC isolation,¹³ hLEC colonies thrived, and cultures contained a higher number of colonies than those maintained in CM. Studies have shown that different types of medium can have a different effect on LESC growth, cell phenotype, and expression of putative LESC markers.²³ These differences can be due to a number of different factors as there are many components to cell culture medium. One striking difference between SM and CM is the concentrations of fetal bovine serum (FBS), 2% and 10% respectively. Although studies suggest that FBS stimulates limbal epithelial cell proliferation,²⁴ other studies have shown that it is possible to cultivate limbal epithelial stem cells without the use of serum in the medium.^25,26 Our findings suggest that a lower concentration of FBS may be beneficial for hLEC growth as colonies are more likely to become aborted and that cells also become more differentiated when maintained in CM. However, this may not be due to serum content alone as there are other factors involved. Further analysis into the components of these medium will need to be carried out to elucidate the contributing factors for successful isolation. 

All components involved in hLEC isolation, if used for transplantation, require consideration in terms of safety and good manufacturing practice. Although there are banked 3T3s that comply with good manufacturing practice standards at the National Institute of Biological Standards and Controls, it would be ideal to eliminate the use of animal derived products for use in clinic due to the potential risk of transmitting xenotic adventitious agents. Studies have therefore reported the use of human equivalent feeder layers^27,28 and indeed culture methods without feeder layers.¹² Similar to Chen et al.,¹¹ we have found that by using a collagenase-based isolation method, it is possible to successfully isolate primary cultures of hLECs without the use of 3T3 cells as a feeder layer. Chen et al.¹¹ isolated clusters of limbal epithelium that contained supporting niche cells and separated these clusters from stromal cells before culturing. In our studies, we digested only the superficial limbus and cultured the superficial stromal cells together with the limbal epithelial clusters and found that even though cultures were significantly improved in the presence of a 3T3 feeder layer, cultures of hLEC were also successful in SM without a 3T3 feeder layer. This provides further evidence that, with the right isolation technique and optimized cell culture medium, it may be possible to isolate LECS for transplantation without the use of a 3T3 feeder layer. 

In summary, to the best of our knowledge, this is the first study that has examined collagenase digestion of the superficial corneal limbus to isolate hLECs from OCCS rims. Due to the findings we present in this manuscript, we propose that surplus OCCS rims, which might have previously been disregarded, can now be used as a valuable source of research tissue for the isolation of hLECs and possibly LESCs. We also suggest a system whereby hLECs can be isolated without a 3T3 feeder layer by using a specific culture medium and have highlighted a potential culture system in which CSSCs and hLECs can be isolated simultaneously from the same donor rim. 

Acknowledgments

Supported by Shine the Light on Aniridia, the Special Trustees of Moorfields Eye Hospital, and the National Institute for Health Research Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital, and University College London Institute of Ophthalmology. 

Disclosure: V.E. Tovell, None; I. Massie, None; A.K. Kureshi, None; J.T. Daniels, None