Human epithelial cell cultures from superficial limbal explants

Donor corneal tissue

Human donor corneas discarded before transplantation due to low endothelial cell counts (eye bank corneas) and corneoscleral rims obtained during surgery after 8-mm trephination of the graft (surgical corneas) were used. Corneoscleral rims from the eye bank corneas were obtained using an 8-mm trephine.

Preparation of explants

Full-thickness limbal explants were prepared under a laminar flow. Descemet’s membrane was removed using a scalpel and full-thickness limbal rims were isolated by removing the sclera with scissors. Superficial limbal explants were processed in the operating room under an operating microscope. A stromal dissection between the anterior and the mid stroma was performed using a 15° blade and the sclera was carefully removed with scissors resulting in superficial limbal rims. For the growth assay a 2.5-mm trephine was used to obtain 9 round explants with homogeneous areas of limbal epithelium. For all other experiments 6 explants with homogeneous length (4 mm) were obtained from each limbal rim using scissors. Corneal stromal explants were prepared using the central cornea after 8-mm trephination of eye-bank corneas. The epithelium and Descemet’s membrane were first removed and 2-mm stromal explants were prepared with a scalpel.

Culture media

Limbal explants were cultured in cholera toxin-free Green medium. The medium was composed of a 3:1 mixture of calcium-free Dulbecco’s Modified Eagle’s Medium (Dutscher, Brumath, France) and Ham F12 medium (Invitrogen, Cergy Pontoise, France) with 10% fetal bovine serum (Invitrogen), 1 mM/ml HEPES buffer (Invitrogen), 5 µg/ml human recombinant insulin (Actrapid^®; Novo Nordisk, Paris, France), 0.4 µg/ml hydrocortisone (Pharmacia, Pfizer, Paris, France), 4 µM/ml L-glutamine (Invitrogen), 2 pM/ml tri-iodo thyronine (Sigma, Saint Quentin en Yvelines, France), 200 nM/ml adenine (Sigma), 100 IU/ml penicillin (Invitrogen), 100 µg/ml streptomycin (Invitrogen), 0.25 µg/ml amphotericin B (Invitrogen), and 10 ng/ml human recombinant Epithelial Growth Factor (EGF; Sigma).

Cell culture

In a first series of experiments, the full-thickness limbal explants or the superficial limbal explants either sutured on a plastic lamella (320 mm²; one explant per lamella; Thermanox, Nunc, Illkirch, France) epithelial side up were cultured using 6- (907 mm²) or 24-well (201 mm²) plates (Becton Dickinson, Rungis, France) with 2 ml of medium. To determine whether cultured cells corresponded to epithelial cells, stromal fibroblasts, or mixed cultures, a second series of experiments was performed. Full-thickness and superficial limbal explants (epithelium either side up or down) and stromal explants obtained from two corneas from the same donor were cultured using 6-well plates. Finally, single cell suspensions isolated from limbal tissue were cultured. Briefly, the whole limbal rim was incubated with 1.2 Units/ml dispase II at 37 °C for 1 h. The epithelial sheets were then collected and treated with 0.125% trypsin at 37 °C for 15 min to isolate single cells. No feeders were used to grow cells. The medium was changed three times a week and cells were cultured for three weeks at 37 °C with 5% CO₂.

Morphological analysis

Morphological analysis was performed twice a week with a phase contrast light microscope. Peripheral explant cell sheet areas were calculated after reconstitution of images using the Image J 1.34 software (National Institutes of Health, Bethesda, MD) after 1, 2, and 3 weeks of culture. Confocal microscopy was performed on 3-week cultures using the Heidelberg Retina Tomograph III with the Rostock Cornea Module (HRT III/RCM; Heidelberg Engineering GmbH, Heidelberg, Germany).

Morphometric analysis

After three weeks cells were fixed with 4% paraformaldehyde solution for 10 min at room temperature. In each well, at least 300 cells from 3 fields were analyzed and 36 images (3 per well) were acquired with a color camera for computerized analysis. For each image, The Image J software allowed to study morphology after individual contours were designed manually. The mean values for cell area, Feret diameter, and percentage of small cells (less than 16 µm²) were assessed for each well.

Growth assay

Limbal explants trephined at 2.5 mm were cultivated in 24-well plates. After 7, 14, and 21 days, cells were dissociated enzymatically and counted. The equation giving the number of cells in culture (n_t) at t time is the following: n_t=n₀ × e^(t/T)xln2, where n₀ is the number of cells in the explant at the beginning of culture and T is the time period of the cell cycle. The equation which best-fitted the observed curve of growth was determined using Excel software (Microsoft Excel; Microsoft Corporation, Paris, France).

Immunocytochemistry

Immunocytochemical staining was performed to evaluate the expression of different molecular markers and to identify limbal epithelial cells and progenitors. Six slides were immunostained for each marker for superficial limbal explants and six for full-thickness limbal explants cultured cells. The technique was developed with the objective to get no background staining which allows easy differentiation of stained and unstained cells and easy counting of stained cells. Normal corneas obtained after enucleation for choroidal melanoma were used as control tissues. Tissues were fixed in paraffin and stored at −20 °C. Frozen sections cut with a Microtom (Leica RM2145; Leica, Paris, France) were collected on slides (Superfrost; ThermoScientific, Illkirch, France) and were put in histosol and rehydrated in 100%, 90%, and 70% ethanol and double distilled water. Cultured cells were enzymatically dissociated and concentrated by cytospin. After washing in PBS, the tissue sections and cultivated cells were fixed for 10 min with 4% paraformaldehyde and incubated for 30 min in PBS containing 1% BSA (BSA) and 0.3% Triton X 100 to permeabilize the cells and to block non-specific staining. The endogenous peroxidases were quenched with 0.3% H₂O₂ during 10 min. Cells were incubated for 30 min at room temperature with primary antibodies against cytokeratin 3 (1:200; Clone AE-5; Dako, Trappes, France), cytokeratins 4, 5, 6, 8, 10, 13, and 18 (1:100; MNF116; Dako), cytokeratin 19 (1:50; clone BA17, Dako), vimentin (1:200; clone V9, Dako), Delta N p63α (1:50; clone 4A4, Dako) and ATP-binding cassette, subfamily G, member 2 (ABCG2; 1:20; clone CDw338, BD PharMingen, San Diego, CA) followed by incubation with the biotinylated secondary antibody using a LSAB2 system-HRP Kit (Dako) according to the manufacturer’s instructions. DAB was used as peroxidase substrate and specimens were counterstained with hematoxylin.

Flow cytometry analysis

The flow cytometry analysis was performed on a BD FACSCalibur system (Becton Dickinson). The staining of each population, combined with the forward scatter channel (FSC) and the side scatter channel (SSC) data, identifies cells present in a sample and it allows counting the relative proportions of each. Cell suspensions (10⁵cells/ml) in PBS/BSA buffer were fixed with 4% paraformaldehyde (Fix Buffer I; BD Bioscience) for 10 min at 37 °C. After washing with PBS/BSA, cells were permeabilized with Perm Buffer III (BD Bioscience) and incubated for 30 min at 4 °C. Aliquots were distributed into different test tubes for primary antibody binding (CK3, vimentin, CK 4, 5, 6, 8, 10, 13, and 18, CK19, DeltaN p63α, and ABCG2) and incubated at room temperature for 30 min. An anti-mouse IgG1 FITC or IgG2a FITC secondary antibody was added, and cells were incubated for 30 min at room temperature. Relative size, granularity or internal complexity, and relative fluorescence intensity were registered with the Cell Quest Pro software (BD Bioscience).

Assessment of specific messengers RNAs in cell culture by reverse transcription and Polymerase Chain Reaction (RT–PCR)

Total RNA was isolated using the MagNA Pure Compact RNA isolation Kit (Roche Diagnostics, Mannheim, Germany) after 21 days of culture according to the manufacturer’s instructions. The RNAs were quantified by measuring absorption at 260 nm. RT–PCR was performed using the Qiagen One Step RT–PCR Kit (Qiagen, Courtaboeuf, France) which provides enzymes for both the reverse transcription and the PCR. The first-strand cDNA was synthesized after incubation at 50 °C for 30 min and PCR conditions were 95 °C for 15 min followed by 35 cycles of 94 °C for 30 s, 60 °C for 30 s, 72 °C for 1 min, and finally at 72 °C for 10 min. The sequences of the primers for human Cytokeratins 3 and 19, vimentin, Delta N p63α, ABCG2, and β-actin (ACTB), a housekeeping gene used as an internal control, are presented in Table 1. The amplified products were separated using 2% agarose gel electrophoresis stained with ethidium bromide.

Statistical analysis

Non-parametric tests (χ², Spearman rank correlation coefficient, Kruskal–Wallis ANOVA, and Mann–Whitney test) were used for statistical analysis with the Statistica 6.1 software (StatSoft Inc., Maisons-Alfort, France); a p<0.05 was considered statistically significant.

Donor tissue

Thirty-two human corneas were used in this study. The average donor age was 66±16 years (SD, range 32 to 80 years). Time from death to tissue procurement ranged between 9 and 47 h (mean 23±1 h). All corneas had been organ-cultured as previously described [20,21] for an average of 18±5 days before trephination.

Comparison of explant preparation techniques

Full-thickness limbal explants were first used for growing limbal epithelial cells. Cell growth appeared to be closely linked to adhesion of explants to the bottom of the culture well. Then preparation of superficial explants and fixation of explants was developed to improve adhesion.

In successful cultures, after three weeks, the limbal epithelial cells covered the well surface. Only polygonal cells were observed when superficial limbal explants were cultivated (Figure 1A). Conversely, two types of cells (polygonal cells and fibroblast-like cells) were observed for full-thickness limbal explants (Figure 1B) and one type of cells, spindle-shaped, large and flat, were observed for stromal explant culture (Figure 1C).

In the first series of experiments, significant differences in the percentage of confluent cultures after 3 weeks were found according to explant preparation technique (Table 2). This figure was significantly higher for superficial explants (p=0.0001) and for superficial explants sutured on plastic lamella (p=0.00001) compared to full-thickness explants. However no significant differences (p=0.65) were found between non-sutured superficial explants and superficial explants sutured on plastic lamella.

In the second series of experiments (Figure 2), no cell growth was observed from stromal explants during the first 3 weeks of culture. Stromal fibroblast growth started after 40 days of culture. Dissociated limbal epithelial cells did not attach to the bottom of the well which resulted in no cell growth. The cell sheet area was significantly larger after 14 (p=0.007) and 21 days (p=0.02) of culture and the number of cells per well after 21 days of culture was significantly higher (p=0.002) with superficial than with full-thickness limbal explants (Figure 2, Table 3). The cell sheet area was significantly larger after 7 (p=0.001) and 14 days (p=0.03) of culture, but not after 21 days (p=0.26), with limbal explants epithelial side up as compared with limbal explants epithelial side down (Figure 2). No significant differences in the number of cells per well after 21 days of culture were found between limbal explants epithelial side up and limbal explants epithelial side down (p=0.59; Table 3).

Morphometry results are presented in Table 4. Cells showed higher average cell area, average cell perimeter, and average Feret diameter when full thickness limbal explants were cultured as compared with cultures obtained from superficial limbal explants (p<0.001). Fibroblasts cultured from stromal explants featured dramatically larger cell area, perimeter, and diameter (p<0.001).

Growth assay

From a superficial limbal explant corresponding to 1/6 of the limbal circumference, an average of 5.9×10⁵ cells could be obtained after 21 days of culture (Figure 3). The number of cells obtained after 21 days did not significantly correlate with donor age (p=0.66), death-to-preservation time (p=0.71), and organ culture time (p=0.58). The average cell sheet area was 5 mm² after 7 days, 140 mm² after 14 days, and 227 mm² after 21 days.

Confocal microscopy

Confocal microscopy of 3-week cultures showed a multilayered epithelial cell sheet, with small basal cells and large flattened superficial cells (Figure 4).

Immunocytochemistry

Figure 5 shows that vimentin, cytokeratin 3, cytokeratin 19, and delta N p63α are expressed in normal human corneas (limbus and central cornea). MNF116 antibody (recognizes a group of 7 cytokeratins characteristic of epithelial cells) was expressed in all corneal and all limbal epithelial cells (Figure 5I and Figure 5J). The broad spectrum cytokeratins was not expressed by stromal cells. When limbal epithelial cell from explants were cultured for 3 weeks, the MNF116 antibody stained 82±19% of cells obtained from superficial limbal (Figure 6E) and 37±3% of cells obtained from full thickness limbal explants (Figure 6F; p=0.01). Cytokeratin 3, a corneal specific marker, was strongly expressed by all corneal epithelial cells (Figure 5E) and by the superficial limbal epithelial cells (Figure 5F). For superficial limbal explants 74±19% were CK3 positive (Figure 6A) and only 25±6% for full thickness limbal explants (Figure 6B; p=0.009). Vimentin was detected in 100% of limbal basal epithelial and stromal cells (Figure 5A) but was undetectable in the corneal epithelial layers (Figure 5B). Vimentin was expressed in 56±18% and 53±5% of cells from superficial (Figure 6C) and full thickness (Figure 6D) limbal explants respectively (p=0.48). Delta N p63α protein was detected in the nucleus of limbal epithelial basal cells (Figure 5D). When superficial limbal explants were cultured, delta N p63α was detected in 54±20% (Figure 6I) but undetectable (0%) in cells cultured from full thickness limbal explants (p<0.001). Cytokeratin 19 (a proposed marker for skin hair follicle progenitor stem cell) was highly expressed in the cytoplasm of basal and superficial limbal epithelial cells (Figure 5H) and it was expressed by 46±27% cells when superficial limbal explants were cultured (Figure 6G) and 25±12% when it was full thickness limbal explants (Figure 6H; p=0.19). ABCG2 has been proposed as one of the universal markers for progenitor cells. However, under our experimental conditions the number of cells positive for ABCG2 in cells cultured from full thickness limbal explants (Figure 6L) was negative, while those obtained with superficial limbal explants culture was 5±7% (Figure 6K; p=0.1). In the positive control all the human corneal epithelial cells expressed CK3 (Figure 6N) but not vimentin. Vimentin was expressed by all cells cultured from stromal explants (Figure 6O). CK3 and broad spectrum cytokeratins (MNF116) were not expressed by these cells.

In the second series of experiments, 100% of cells cultured from stromal explants expressed vimentin but not cytokeratins. Differences in the percentage of cells stained by AE5, MNF116, and vimentin antibodies between superficial and full-thickness limbal explants and between limbal explants epithelium side up and limbal explants epithelium side down did not reach statistical significance (p>0.05; Table 5). When superficial limbal explants were cultured, delta N p63α was detected in 50% but undetectable (0%) in cells cultured from full thickness limbal explants.

Flow Cytometry

Figure 7 shows the dot plots corresponding to the analysis of cell suspensions from limbal explants cultured for three weeks by flow cytometry. The proportion of cells with a small size (polygonal cells) was higher in cultures obtained with superficial (83%) than with full thickness limbal explants (68%) and stromal explants (10%). Ninety percent of cells were large when stromal explants were cultured. Thirty-two percents and 17% of cells were large when full thickness limbal explants and superficial limbal explants were cultured, respectively.

High levels of expression of CK3, CK4, 5, 6, 8, 10, 13, and 18 was observed in superficial and full thickness limbal explants cultured cells. Expression of vimentin was higher in full thickness than in superficial limbal explant cultured cells. Lower levels of expression were found for delta N p63α and ABCG2 in superficial limbal explants cultured cells and no expression was found in full thickness limbal explant cultured cells (Table 6).

RT- PCR

For superficial limbal explant cultures, the analysis of the expression profiles for various RNAs (CK3, CK19, vimentin, delta N p63α, and ABCG2) at the end of 21 days of culture showed high levels of expression of the mRNAs encoding CK3, vimentin, and CK19. The expression of mRNA of delta N p63α and ABCG2 was weaker (Figure 8). Cultures obtained from full-thickness limbal explants featured no expression of mRNA of CK19, delta N p63α, and ABCG2, whereas mRNAs encoding CK3 and vimentin were detected. Fibroblasts cultured from corneal stromal explants featured no expression of mRNA of CK19, CK3, delta N p63α, and ABCG2, whereas mRNA encoding vimentin was detected.