A New Human Primary Epithelial Cell Culture Model to Study Conjunctival Inflammation

Abstract

Purpose.: To develop a complete and optimized method to expand in culture human conjunctival epithelial cells from cadaveric donor samples.

Methods.: Epithelial cells were obtained from cadaveric conjunctival tissue (n = 47). Preplating and differential trypsinization were optimized to eliminate stromal contamination. Epithelial cells were grown with five different media: control, epithelial growth factor (EGF)-enriched, H₂O₂-supplemented, fibroblast-conditioned, and human serum media. Adhesion, proliferation, colony forming efficiency (CFE), and percentage of CK19⁺ and Ki67⁺ cells were determined with the five different media. Cells were characterized by immunofluorescence and/or Western blotting techniques for the expression of CK4, CK7, CK19, MUC5AC, vimentin, FSP-1, Ki67, E-cadherin, and zonula occludens (ZO)-1 markers. In addition, cells were treated with TNF-α and levels of secreted IL-6 were measured by enzyme-linked immunosorbent assay.

Results.: Pure epithelial cell cultures were obtained. Human serum medium showed the best properties in proliferation and CFE, while maintaining epithelial phenotype. Cells with this medium were passaged up to five times, although they maintained all epithelial characteristics only through passage 3. Cultured cells expressed epithelial markers, but not stromal ones. The number of MUC5AC⁺ cells increased throughout the passages, whereas Ki67⁺ cell numbers decreased. Cells in culture maintained adherens and tight junctions, and responded to TNF-α treatment by releasing more IL-6, showing that they can be used for inflammation assays.

Conclusions.: We have developed a complete protocol to expand conjunctival epithelial cells from cadaveric tissue. This culture system responded to an inflammatory stimulus, so it could be used to develop a more complex in vitro model of inflammation.

Introduction

The conjunctiva is the mucous membrane that covers the ocular surface from the limbus to the posterior surface of the eyelids. It is composed of stroma and epithelium, which consists of 2 to 7 layers of stratified squamous epithelial cells. In the more external layers, goblet cells, which are specialized epithelial cells, produce mucin. ¹ Epithelial conjunctival cells are essential for maintaining a healthy ocular surface, having themselves an immunomodulatory role ² that makes them an important target of study. To elucidate the effect of cytokines or drugs in the conjunctiva, a good in vitro system resembling the in situ epithelium is needed; however, a good model for this has not yet been developed. 

Most of the reported in vitro studies of conjunctiva are done using established cell lines. To the best of our knowledge, the only human conjunctival cell lines that have been previously described are the Chang conjunctival cell line, ³ the Wong-Kilbourne derivative of Chang conjunctival epithelial cells (American Type Culture Collection 20.2, Manassas, VA), the IOBA-normal human conjunctiva (NHC) cell line, ⁴ and the ConjEp-1/p53DD/cdk4R/TERT (HCjE) cell line. ⁵ These lines have many advantages, such as availability and homogeneous results. However, in many ways they are not a fair representation of the original tissue. ^6,7 This fact makes primary cell cultures a more valuable in vitro working tool. 

Conjunctival primary cell cultures of different species have been established, such as rat, ⁸ rabbit, ⁹ and mouse and monkey, ¹⁰ among others. There are obvious interspecies differences, so it is not always possible to use them because the results cannot be readily extrapolated to the human situation. 

The main source of human tissue for conjunctival cell culture is conjunctival biopsy, ^6,11–13 which can be done for patients undergoing cataract or other scheduled surgeries. ¹⁴ Although the procedure of the biopsy is not very aggressive, some ethical concerns exist when doing biopsies for research alone, even if donors have signed informed consent releases. Additionally, tissue availability may be limited. For those reasons, the use of cadaveric tissues provides a promising alternative to biopsy tissues. 

Independent from the source of tissues, epithelial primary cell cultures are very often established with the aide of feeder layers, usually murine 3T3 fibroblasts ¹⁵ because they improve culture performance. However, this strategy has some risks, especially if the final destination of the culture cells is for transplantation. ¹⁶ Thus, the optimization of the culture procedure and the culture medium is the main concern when culturing epithelial cells without using feeder layers. Several supplements have been described to promote epithelial proliferation. The main one is epithelial growth factor (EGF), ¹⁷ but there are some others. For instance, Pan et al. ¹⁸ reported higher adhesion, migration, and proliferation levels in corneal cells when adding small amounts of H₂O₂ to the medium. Finally, the main source of growth factors is serum, and for that reason its inclusion in the culture system is an issue of great importance. 

Notwithstanding some obvious problems such as elderly donors and the elapse of days between eye extraction and cell culture, the aim of this research was to establish an optimized protocol for expanding human conjunctival cells derived from cadaveric donors. Here, we demonstrate optimized procedures for isolating and culturing conjunctival epithelial cells from cadaveric tissue and efficiently expanding the cells in the absence of feeder layers. Our results provide a good in vitro model to deepen the knowledge of human conjunctiva and may eventually provide abundant cells for transplantation to save vision in diseased and traumatized eyes. 

Materials and Methods

Human Conjunctival Tissues

Healthy human conjunctival tissues were obtained from corneoscleral buttons from cadaveric donors (n = 47, from 40 donors). Fifty-seven and half percent of donors were male and 42.5% were female, with a mean age and SEM of 80.5 ± 1.23 years. The average interval between the death of the donor and harvesting of the conjunctival tissue was 2 to 3 days. Corneoscleral buttons were obtained with informed research consent from Barraquer Eye Bank of Barcelona (Spain). This study was in strict accordance with the tenets of the Declaration of Helsinki and Spanish Regulations concerning the use of human tissues for biomedical research, and had the approval of institutional review board of the University of Valladolid. Human bulbar conjunctiva was carefully isolated from the rest of tissues, and used for research purposes. 

Isolation and Culture of Conjunctival Cells

Human fibroblasts were obtained from conjunctival tissue. The epithelium was gently scraped, and then the underlying stroma was cut in small pieces (approximately 2 mm²/piece), that were plated in 12-well plastic plates (Nunc, Roskilde, Denmark). Fibroblast culture medium was composed by Dulbecco's modified Eagle's Medium (DMEM)/F12 supplemented with 10% fetal bovine serum (FBS), 2.5 μg/mL fungizone, and 5000 units/mL penicillin/streptomycin (all from Invitrogen, Inchinnan, UK). 

Human conjunctival epithelial cells were obtained by disaggregation of the conjunctivas using either an enzymatic method or a combined enzymatic and mechanical method. Briefly, conjunctival tissue was incubated with dispase (1.2 U/mL; Invitrogen) for 2 hours, at 37°C in 5% CO₂. After that, loosened epithelial cells were recovered with the help of a pipette. In the combined method, additional mechanical scraping was performed. The dispase solution containing the cells was then centrifuged for 5 minutes at 900 rpm. The recovered cells were left in 0.25% trypsin/EDTA composed by 2500 mg/L trypsin and 380 mg/L sodium EDTA (Invitrogen) for 5 minutes at 37°C, and then centrifuged and resuspended in five different culture media. The control medium was composed of DMEM/F12 supplemented with 2.5 μg/mL fungizone, 5000 units/mL penicillin/streptomycin, 1 μg/mL insulin, 0.5 μg/mL hydrocortisone (Sigma-Aldrich, St. Louis, MO), 0.1 μg/mL cholera toxin (Gentaur, Brussels, Belgium), 2 ng/mL EGF, and 10% FBS (all from Invitrogen unless otherwise indicated). The four other media were based on the control medium, with the following variations: EGF-enriched medium contained 10 ng/mL EGF; H₂O₂-supplemented medium contained 20 μM hydrogen peroxide solution (Sigma-Aldrich); fibroblast-conditioned medium was prepared as described below (in the “Preparation of Fibroblast-Conditioned Medium” section); and human serum medium had 10% of human serum (Lonza Group Ltd., Basel, Switzerland), instead of FBS. In addition, epithelial cells were also obtained using the explant technique. ¹¹ Conjunctival tissue was carefully cut in small pieces, and plated in 12-well plates, using the same culture media described above. Explants were fed by superficial tension until cell growth was observed. After that, more culture medium was added. 

Cells were maintained in standard conditions (humidified atmosphere of 5% CO₂ at 37°C), and the medium was changed every other day. Cell viability was evaluated with 0.4% trypan blue solution (Sigma-Aldrich). 

Purification Techniques

To assure the purity of the epithelial culture, two combined techniques were used to eliminate contaminating fibroblasts: preplating and differential trypsinization with 0.25% trypsin/EDTA. Different times of preplating (1, 2, and 3 hours) and trypsinization (2, 5, and 10 minutes) were assayed for effectiveness in eliminating all contaminating stromal cells. To determine if cultured cells were epithelial or stromal cells, immunostaining for the epithelial marker cytokeratin (CK) 19 and against the stromal marker vimentin was performed. 

Preparation of Fibroblast-Conditioned Medium

Fibroblasts isolated as described above were grown until confluence in 80-cm² flasks, and they were mitotically inactivated using 10 μg/mL mitomycin C (Sigma-Aldrich). Then, 15 mL control culture medium were added. After 24 hours, the medium was collected, centrifuged at 1000 rpm for 3 minutes, and the supernatant was frozen until used. This procedure was repeated for 7 days. Recovered media from the same flask were mixed and filtered through a 0.20-μm filter (Sartorius Stedim Biotech GmbH, Goettingen, Germany). 

Cell Adhesion Assays With Different Media

To evaluate adhesion, 20,000 cells/cm² were seeded in 96-well plates (Nunc) and allowed to adhere for 24 hours. After that, medium with nonattached cells was removed, and the recovered cells were counted in a hemocytometer (Bright-Line Hemacytometer; Sigma-Aldrich). Plating efficiency was calculated using the following equation:  

 

Additionally, the DNA content of the attached cells after 24 hours was measured. After removing the media, cells were washed with PBS and Hoechst 33342 dye (Sigma-Aldrich) was added. After 3 minutes of incubation, cells were washed with PBS, and fluorescence was measured at 355 nm excitation and 465 nm emission wavelengths, using a fluorescence plate reader (SpectraMax M5; Molecular Devices, Sunnyvale, CA). That measure is proportional to the amount of DNA present in the sample. Three independent experiments were performed in triplicates.  

Cell Proliferation Assays With Different Media

To evaluate proliferation, the Alamar Blue (R) colorimetric indicator assay (AbD Serotec, Oxford, UK) was used. Alamar Blue is a nontoxic fluorescent dye that does not affect viability or proliferation. For proliferation assays, 20,000 cells/cm² were seeded in 96-well plates. The Alamar Blue assay was performed at days 1, 2, 3, 4, 7, 11, and 14. Cells were incubated in medium with 10% vol/vol Alamar Blue for 4 hours. After incubation, that medium was recovered, and its fluorescence was measured at 560 nm excitation and 590 nm emission wavelengths, using the SpectraMax M5 fluorescence plate reader (Molecular Devices). Four independent experiments were performed in triplicates. 

Colony Forming Efficiency

For the colony forming efficiency (CFE) assay, 500 cells were seeded in each well of a 6-well plates, with the five different culture media. After 7 days, the number of colonies with more than eight cells in each well was counted. Additionally, the colony size was measured by counting the number of cells in each colony. 

Characterization of Conjunctival Cells by Immunocytochemistry

Fibroblasts and epithelial cells from the established conjunctival cultures (passages 1–3) were grown in 8-well multichamber Permanox slides (Nunc). When cells reached confluence, they were fixed with ice cold methanol. Immunocytochemistry for different epithelial cells (E-cadherin, CK19, and CK7), goblet cells (MUC5AC), stromal cells (vimentin and FSP-1), and proliferating cells (Ki67) markers (Table 1) was then performed on the fixed cells. For MUC5AC and Ki67 staining, cells were blocked in PBS with 4% donkey serum (Sigma-Aldrich) and 0.3% Triton X-100 (Sigma-Aldrich) for 1 hour at room temperature (RT). For MUC5AC, Ki67, and zonula occludens (ZO)-1, the slides were incubated with primary antibodies (Table 1) for 1 hour at 37°C. For E-cadherin, CK19, CK7, vimentin, and FSP-1, the slides were incubated with primary antibodies (Table 1) overnight at 4°C. Alexa Fluor-conjugated secondary antibodies were applied for 1 hour at RT. Cell nuclei were counterstained with Hoechst dye. The preparations were viewed under an epifluorescence microscope (Leica DMI 6000B; Leica Microsystems, Wetzlar, Germany). Negative controls included the omission of primary antibodies. All antibodies had been previously tested in our laboratory. The number of positive cells for each marker was counted in five different photographs from at least three independent experiments. 

Table 1

 

Antibody Sources and Concentrations

Antibody	Host and Type	Dilution for IMF	Dilution for WB	Code	Source
CK19	Mouse mo.	1:50	1:1000	MO888	Dako, Glostrup, Denmark
CK4	Mouse mo.	-	1:250	C-5176	Sigma-Aldrich
CK7	Mouse mo.	1:100	1:200	MS-1352-P0	ThermoScientific, Rockford, IL
MUC5AC	Mouse mo.	1:500	1:500	MAB-2011	EMD Millipore Corporation, Billerica, MA
Vimentin	Mouse mo.	1:500	1:200	SC-6260	Santa Cruz Biotechnology, Santa Cruz, CA
FSP-1	Mouse mo.	1:100	-	Ab11333	Abcam, Cambridge, UK
Ki67	Mouse mo.	1:50	-	M7240	Dako
E-cadherin	Mouse mo.	1:100	1:1000	610151	BD Bioscience, Franklin Lakes, NJ
ZO-1	Rabbit po.	1:50	1:250	617300	Invitrogen
AlexaFluor 448 anti-mouse IgG	Donkey	1:150	-	A21202	Invitrogen
AlexaFluor 647 anti-mouse IgG	Goat	1:200	-	A21236	Invitrogen
AlexaFluor 448 anti-rabbit IgG	Donkey	1:100	-	A21206	Invitrogen
Texas Red anti-mouse IgG	Donkey	-	1:5000	715-075-150	Jackson Immuno-Research Laboratories, Inc., West Grove, PA
Goat anti-rabbit IgG	Goat	-	1:2000	c-2004	Santa Cruz Biotechnology

 mo., monoclonal; po., polyclonal.

Characterization of Conjunctival Cells by Western Blotting

Cultured conjunctival epithelial cells from passages 1 to 5 and conjunctival fibroblasts were homogenized in ice cold radioimmunoprecipitation assay (RIPA) buffer supplemented with the protease inhibitors 100 μL/ml phenylmethylsulfonyl fluoride, 6 μL/ml aprotinin, and 100 nM sodium orthovanadate (all from Sigma-Aldrich). After incubation on ice for 30 minutes, samples were centrifuged at 14,000 rpm for 30 minutes at 4°C. The supernatants were collected, and the protein concentration was measured using the BCA assay (Pierce, Rockford, IL). 

Proteins in each sample were separated by SDS-PAGE on 10% acrylamide gels (Bio-Rad Laboratories, Hercules, CA) according to the method of Laemmli. ¹⁹ The separated proteins were transferred to nitrocellulose membranes. ²⁰ Membranes were blocked in Tris-buffered saline (TBS; Bio-Rad Laboratories) containing 0.05% Tween-20 BioXtra (Sigma-Aldrich), 5% milk, and 4% FBS for 1 hour at RT. Membranes were incubated with primary antibodies (CK19, CK4, CK7, MUC5AC, vimentin, E-cadherin, and ZO-1; Table 1) at 4°C overnight. Proteins recovered as described above from human colorectal adenocarcinoma HT29-MTX cells (a kind gift by Thécla Lesuffleur, PhD) served as positive controls for CK19 and MUC5AC expression. ²¹ Then the membranes were washed with TBS and incubated with horseradish peroxidase-conjugated secondary antibodies (Jackson Immuno-Research Laboratories, Inc.) for 1 hour at RT. Immunoreactive bands were visualized by a chemiluminescence method using the ChemiDoc gel documentation system (Bio-Rad Laboratories), and images were analyzed with the Quantity One software (Bio-Rad Laboratories). Quantitation of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) assured equal gel loading. Two independent experiments were performed. 

In Vitro Inflammation Model

Cells from passages 0 to 3 were cultured in 24-well plates at a cell density of 25,000 cells/cm². When cells reached confluence they were stimulated with 25 ng/mL TNF-α (PeproTech, London, UK) for 24 hours. After that, supernatants were collected, and levels of IL-6 were measured with an enzyme-linked immunosorbent assay (IL-6 ELISA; Diaclone, Besançon, France), following the manufacturer's instructions. Nonstimulated cells were used as controls for each experiment. Three independent experiments were performed for each cell passage. 

Statistical Analysis

Statistical analyses were done using Statistical Procedures for the Social Sciences software (SPSS 15.0; SPSS Inc., Chicago, IL). Data were expressed as means ± SEM. Levene's test was used to evaluate the equality of variance. When it existed, a one-way ANOVA was performed, followed by pairwise comparisons (Tukey test). When there was no equality of variance, a robust test (Brown-Forsythe test) was performed, followed by pairwise comparisons with Games-Howell test. Differences were considered to be significant when P was less than or equal to 0.05. 

Results

Acquisition and Purification of Conjunctival Epithelial Cells

Epithelial cells were efficiently expanded from single cell suspensions and from explants. There were no significant differences in cell numbers obtained between the two disaggregation methods. However, recovered cell viability was significantly higher with the enzymatic method than with the combined enzymatic and mechanical one (P = 0.0024, Fig. 1A). Explants promoted more rapid expansion of cells. After 11 days, a mean of 98,707 ± 12,967 cells/explant were obtained. In addition, this technique was easier and faster to perform. For those reasons, expansion from explants was used for the rest of the studies, except for adhesion and colonies assays. 

Figure 1

 

Establishment of epithelial cell culture. Two disaggregation methods were studied (A). Briefly, the enzymatic method consisted of a 2 hour incubation with dispase and 5 minutes incubation with trypsin/EDTA. The combined method included a mechanical scraping between both incubations. Viability of recovered cells was significantly higher with the enzymatic method (P = 0.0024) than with the combined one (n = 9). Two purification techniques were assayed (B). After 2 hours of preplating all the fibroblasts were attached. After 5 minutes of contact with trypsin, all the fibroblasts were detached, whereas most of the epithelial cells still remain attached to the culture surface (n = 3). ***P ≤ 0.005.

Purification of the cultures by either preplating or differential trypsinization produced epithelial cells with no contaminating fibroblasts. By discarding cells attached after preplating for 2 hours or by removing cells after 5 minutes of trypsinization, we obtained pure epithelial cell cultures (Fig. 1B). 

Evaluation of Five Different Cell Culture Media

To evaluate the suitability of the five different culture media in maintaining epithelial cell cultures, different parameters were analyzed: (1) cell adhesion to the plastic culture surface, (2) cell growth and proliferation, (3) colony-forming efficiency, and (4) maintenance of the epithelial phenotype (CK19⁺ cells) and proliferative capacity (Ki67⁺ cells). 

Cell Adhesion.

Cell adhesion was measured in passage 0 cells at 24 hours, using two different indicators: seeding efficiency and DNA content of attached cells (Fig. 2). Fibroblast-conditioned medium resulted in better promotion of cell attachment. Seeding efficiency of cells cultured with this medium was significantly higher when compared with that of cells cultured with control (P = 0.023), EGF-enriched (P = 0.035), and human serum (P = 0.005) media. There were no significant differences in DNA content among the cells in the different media. 

Figure 2

 

Cell adhesion. The suitability of five different culture media in promoting cellular attachment was measured in passage 0 cells with two procedures. (A) Seeding efficiency was higher in cells cultured with the fibroblast-conditioned medium than in those of the control (P = 0.023), EGF-enriched (P = 0.035), and human serum (P = 0.005) media. (B) No significant differences were found in the DNA content of attached cells cultured with the different media. *P ≤ 0.05; ***P ≤ 0.005.

Cell Proliferation.

Cell proliferation with the five different media was assayed at different time points from days 1 to 14 (Fig. 3) in passage 1 cells. At day 11, proliferation rates with fibroblast-conditioned and human serum media were significantly greater than that of the control medium (P = 0.015 and P = 0.008, respectively). Cells cultured with H₂O₂-supplemented medium did not survive for 14 days. Cell proliferation rate was significantly higher with human serum medium after 14 days when compared with that of cells cultured with control (P < 0.001), EGF-enriched (P < 0.001), and fibroblast-conditioned (P = 0.002) media. Fibroblast-conditioned medium also allowed higher proliferation rates compared with control and EGF-enriched media (P < 0.001 and P < 0.001, respectively) at day 14. 

Figure 3

 

Cell proliferation. The effectiveness of different media in inducing cell proliferation was evaluated with Alamar Blue assay in passage 1 cultured cells. The highest proliferation rates were reached with the human serum medium, followed by the fibroblast-conditioned medium. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005, compared with control.

Colony Forming Efficiency and Colony Size.

Passage 0 cells grown in human serum medium showed the highest CFE (control, P = 0.049; H₂O₂-supplemented, P = 0.028; Fig. 4A). Furthermore, colonies formed using human serum medium were significantly larger than those formed using the four other culture media (control, P = 0.001; EGF-enriched, P = 0.001; H₂O₂-supplemented, P = 0.001; fibroblast-conditioned, P = 0.006; Fig. 4B). 

Figure 4

 

Colony forming efficiency and colony size. The number and size of colonies was measured in passage 0 cells cultured with the different media. Cells cultured in human serum medium showed the highest CFE (A). These colonies were also the largest ones (B). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005.

Epithelial Phenotype and Proliferation Capacity.

Cells from passage 1 grown in the different media were evaluated for CK19 expression (n = 3). There were no significant differences among the five media (Fig. 5). Evaluation of Ki67 expression (n = 3), a marker for proliferative capacity, was also performed. Fibroblast-conditioned and human serum media maintained the proliferative capacity of cultured cells significantly better than the control medium (P = 0.002 and P = 0.019, respectively; Fig. 5). 

Figure 5

 

Percentage of positive cells for CK19 and Ki67. There were no significant differences in CK19⁺ cell numbers between cells cultured in the different media. Fibroblast-conditioned and human serum media showed a higher percentage of Ki67⁺ cells than the control medium. *P ≤ 0.05; ***P ≤ 0.005.

Characterization of Cells Grown in Human Serum Medium

As the human serum medium showed better results than the others, cells at different passages cultured in this medium were characterized by immunocytochemistry and Western blotting (Fig. 6). Cells from passages 1, 2, and 3 grown in human serum medium were evaluated by immunofluorescence assay for six different markers (Table 2). In addition, epithelial cells from passages 0, 1, 2, 3, and 5, and human conjunctival fibroblasts were also evaluated by Western blotting (Fig. 6). Cells from passages 0 to 3 were positive for epithelial markers CK4, CK19, and CK7, and negative for stromal markers vimentin and FSP-1, whereas the opposite was true for fibroblasts. Epithelial cells from passage 5 showed an intermediate phenotype characterized by some expression of both epithelial and stromal markers and by different morphology. By immunofluorescence, some cells expressing MUC5AC were detected, but it was not detected by Western blotting. There was a significant increase in the amount of cells expressing MUC5AC from passage 1 to 3 (P = 0.05), and a decrease in the percentage of cells expressing Ki67 from passage 1 to 2 (P = 0.021) and passage 3 (P = 0.003). 

Figure 6

 

Characterization of cells cultured in human serum medium. Representative images of the different markers. All cells were positive for CK19 and CK7 ([A, B], respectively), whereas few MUC5AC-expressing cells were detected (C). Cells with proliferative capacity were identified by Ki67 staining (D). Cultured epithelial cells did not show staining for the stromal markers vimentin (E) or for FSP-1 (F). Scale bars: 30 μm (A–F). Also, some markers were evaluated by Western blotting (G). HT-29 cells were used as control for CK19 and MUC5AC. P0-P5: epithelial cells from passages 0 to 5. Fib: human conjunctival fibroblasts.

Table 2

 

Percentage of Positive Cells for Each Marker in Passages 1 Through 3

Marker	Specificity	% P1 + Cells	% P2 + Cells	% P3 + Cells
E-cadherin	Epithelium; adherens junctions	99.9 ± 0.1	99.57 ± 0.43	100 ± 0
CK19	Stratified epithelium	99.6 ± 0.23	100 ± 0	99.75 ± 0.2
CK7	Secretory epithelium	99.73 ± 0.14	99.9 ± 0.06	99.9 ± 0.08
MUC5AC	Mucin, goblet cells	0 ± 0	5.82 ± 2.91	8.82 ± 0.41
Vimentin	Stromal cells	8.72 ± 2.92	7.2 ± 2.75	18.05 ± 3.63
FSP-1	Fibroblasts	0.27 ± 0.27	0.32 ± 0.19	0.46 ± 0.38
Ki67	Proliferating cells	39.12 ± 2.16	27.36 ± 2.08	17.68 ± 1.99

Evaluation of Intercellular Junctions

Cells cultured in human serum medium expressed E-cadherin and ZO-1, which means that they maintained adherens and tight junctions, respectively (Fig. 7). Immunofluorescence assays confirm that both proteins were expressed in the plasma membranes between neighboring cells. In Western blots, immunoreactive bands appeared for each protein at the appropriate molecular weight (120 KDa for E-cadherin and 225 KDa for ZO-1). Both proteins were expressed in epithelial cells from passage 0 to 3. Epithelial cells in passage 5 and fibroblasts did not express either E-cadherin or ZO-1. 

Figure 7

 

Adherens and tight junctions in cultured conjunctival epithelial cells. (A) Representative image of E-cadherin (left) and ZO-1 (right) expression in cultured cells. Scale bars: 30 μm. (B) The same proteins were also detected by Western blotting in epithelial cells from different passages.

In Vitro Inflammation Assay

Cells treated with TNF-α secreted higher amounts of IL-6 than non-stimulated cells (Fig. 8). This increase of IL-6 secretion in cultured cells was statistically significant in passages 0 (P < 0.001), 1 (P = 0.009), and 3 (P < 0.001). Cells in passage 2, also tended to secrete more IL-6, but the increase was not significant (P = 0.089). 

Figure 8

 

IL-6 secretion by epithelial cells treated with TNF-α. Treated cells from passages 0, 1, and 3 showed a significant increase in the secretion of IL-6 compared with untreated cells (P < 0.001, P = 0.009, and P < 0.001, respectively). **P ≤ 0.01; ***P ≤ 0.005.

Discussion

In this study we developed an optimized method for expanding human conjunctival epithelial cells from cadaveric donors for at least 3 passages. Others have used cadaveric samples as a source for conjunctival primary cultures. ²² However, they pooled tissues from 8 to 10 eyes to produce enough cells for culture. In contrast, with our method we obtained enough cells from each eye to establish one culture per eye, obviating the need to pool samples from different eyes. Some other authors also used cadaveric samples, but they only required a small amount of cells for their studies, so they did not optimize cell expansion nor did they establish long-term cultures. 

One of the most common problems that researchers face when working with primary epithelial cell cultures is stromal cell contamination, which has huge implications. ²³ This problem is not usually reflected in the literature, but it produces a high number of failed experiments, which means a waste of resources. We decided to try two combined techniques to eliminate fibroblasts from our epithelial cultures: preplating and differential trypsinization. Usually, researchers remove stromal cells from cultures by scraping away the contaminating cells. However, this procedure is quite subjective, and it is used only when the contamination by the stromal cells is big enough to be detected. The influence of the cocultured stromal cells, until they are removed, can alter some results. ²⁴ Recently, it has been demonstrated that the effects of coculture can persist for weeks after epithelial cell isolation. ²⁵ In addition, in some experiments a coculture situation can exist without knowing. 

With our two simple strategies of preplating and trypsinization, we avoided the risk of fibroblast contamination. Neither procedure is based on morphologic evaluation; rather we used the inherent differential properties of epithelial and stromal cells, which make the procedures more objective. These procedures may result in the loss of some epithelial cells; however, the improved reliability of the outcomes achieved by almost completely eliminating the possibility of stromal contamination is worth it. Something similar happens with feeder layers. While we may have obtained higher efficiency rates using feeder layers, without them we avoided zoonosis and some other risks, such as transmission of prions and animal viruses. ^16,26 Another important advantage of our culture procedure without feeder layers is the simplicity of it. 

As the initial cell numbers are being limited with the purification techniques, and feeder layers are not being used, we consider the use of an optimized culture medium of paramount importance. We started by using the culture medium that maintains an immortalized conjunctival epithelial cell line. ⁴ With this medium as a starting point, the effect of different supplements and growth factors in growth and differentiation was evaluated. With EGF and H₂O₂ we did not obtain the expected results. Contrary to the description by Pan et al. ¹⁸ in corneal epithelial cells, even small amounts of H₂O₂ adversely affected the viability of our conjunctival epithelial cells. 

Several studies have revealed the importance of the signals that cells receive in their native tissues. In mimicking the stem cell niche in culture, epithelial cells better maintain progenitor cell-like in vivo characteristics. ²⁷ Most of the signals received by conjunctival epithelial cells come from conjunctival fibroblasts. ²⁸ For that reason we used fibroblast-conditioned medium. As expected, the results with this medium were better compared with control medium. Proliferation rates and Ki67 levels were significantly higher, and there was an increase in cell attachments at 24 hours. These results are in accordance with a published report. ²⁹ However, the best results came from human serum medium. Although cell adhesion with this medium was worse than with the control one, the promotion of growth as shown by Ki67 proliferation assay and the CFE was good enough to compensate for that inconvenience. These results are in contrast with some others. ^30,31 With human serum, we not only obtained better results, but we also avoided the use of animal serum, such as FBS. Although FBS is widely accepted, when using cultured cells for tissue engineering or transplants it is better to avoid the use of animal products so as to reduce the risk of zoonosis and possible rejection. ^16,30 Great efforts are being made to design a medium without animal products. Recently, Ang et al. ³¹ proposed the use of cord-blood serum as an alternative to adult human serum and FBS. They reached proliferation rates similar to those obtained with FBS, while their results were worse when using adult human serum. Although we think that cord blood is an interesting source of serum, its use in our cultures was not justified. Our results with human serum were very good and cord-blood serum availability is lower, and its price is higher than that of human serum. In addition, with this culture medium we could see that cells began to stratify when maintained in culture after reaching confluence (data not shown). Thus, human serum medium is the better choice for our purposes. Moreover, in the future it may be possible to use autologous serum to culture human cells for transplantation. 

With the selected human serum culture medium, we analyzed different markers to determine the resulting phenotype of the cultured cells. Cells from passages 0 to 3 expressed all of the epithelial markers analyzed, and they did not show stromal characteristics. However, a small percentage of cells positive for cytokeratins were also positive for vimentin, although immunofluorescence staining intensity was weak. The absence of detectable FSP-1 protein suggested that these cells were not fibroblasts. Some authors suggested that cells expressing CK19 and vimentin at the same time could be epithelial stem cells. ^32,33 Further studies with this population of cells could help to test this hypothesis. 

Cells from passage 5 showed an intermediate morphology and phenotype between epithelial and stromal cells. A possible explanation for this observation is that the cells were undergoing an epithelial-mesenchymal transition, ³⁴ since they began to express vimentin at the same time that E-cadherin expression was reduced. In our cells, we observed by immunofluorescence assay a small amount of MUC5AC that increased throughout the passages through passage 3. This could be due to an increase in cell differentiation that occurred concurrently with the decrease in cell proliferation shown by the number of cells expressing Ki67⁺. However, in contrast with immunofluorescence microscopy, MUC5AC was not detected by Western blotting. The small percentage of MUC5AC-producing cells was probably not enough to allow immunoblot detection. Also, the homogenization process to isolate the protein could damage the MUC5AC. These results indicate the importance of characterizing the cultures by both immunohistochemistry and Western blotting. 

In situ, the conjunctiva functions as a barrier for small particles and microorganisms, although the conjunctival barrier is not as tight as corneal barrier. ³⁵ The effectiveness of this barrier depends not only on mucous production, but also on intercellular junctions. ³⁶ We showed that the cultured conjunctival epithelial cells maintained tight junctions (ZO-1) and adherens junctions (E-cadherin). These results indicate that the cultured cells could potentially act in vitro as a biological barrier. 

Finally, we performed an in vitro inflammation assay to determine if the cultured cells were responsive to a well-known inflammatory stimulus. We analyzed the secretion of IL-6 because it is one of the most important molecules in conjunctival inflammation. Tumor necrosis factor-α is reported to induce IL-6 secretion by conjunctival epithelial cells in vitro. ³⁷ The measured increase in IL-6 secretion by the TNF-α–treated cells in our culture system indicates that cultured cells respond to cytokines. We expected these results because TNF-α is one of the major stimuli for the secretion of IL-6. We observed an increased release of IL-6 while the cells passaged in culture. This may have been due to increased cellular differentiation (as suggested by increased numbers of MUC5AC-positive cells and decreased numbers of Ki67 positive cells) that could make them more sensitive to several signals. Although the study of only one cytokine is not enough to represent the complexity of ocular surface inflammation, the ability of our cultured conjunctival epithelial cells to respond to TNF-α and secrete IL-6 opens the possibility of using them in other more complex inflammation assays. 

Optimizing the culture of human conjunctival cells is an issue of great importance, not only to perform in vitro experiments, but also for tissue engineering. We have developed, to the best of our knowledge, the first complete and optimized protocol to expand human conjunctival cells from cadaveric donors. We have shown the feasibility of using this source of tissue for cell culture. Moreover, we have demonstrated that it is possible to subculture these cells up to three times without losing the unique characteristics of the native epithelia. We obtained good performances despite the disadvantages of this underestimated source of tissue. Our results suggest that an even higher efficiency could be reached in the future if small cell samples from living donors are used for further transplantation purposes. 

Acknowledgments

The authors thank Thécla Lesuffleur, from National Institutes of Health and Medical Research INSERM U843, Paris, France, for kindly providing the HT29MTX cell line. 

Supported by European Regional Development Fund - Interministerial Commission on Science and Technology (FEDER-CICYT) Grant MAT2010-20452-C03-01 and Research Staff Training (FPI) Scholarship Program BES-2011-046381 (Ministry of Economy and Competitiveness), and by Regional Government of Castile and Leon Grant VA132A11-2. Additional financial support was provided by Biomedical Research Networking Center on Bioengineering, Biomaterials and Nanomedicine (CIBER-BBN). 

Disclosure: L. García-Posadas, None; I. Arranz-Valsero, None; A. López-García, None; L. Soriano-Romaní, None; Y. Diebold, None