Centre for Stem Cell Biology and Developmental Genetics, University of Newcastle, Newcastle upon Tyne, United Kingdom
Institute of Human Genetics, University of Newcastle, Newcastle upon Tyne, United Kingdom
Centre for Stem Cell Biology and Developmental Genetics, Institute of Human Genetics, University of Newcastle upon Tyne, International Centre for Life, Central Parkways, Newcastle upon Tyne NE1 3BZ, Telephone: +44 191-241-8688; Fax: +44 191-241-8666
Human embryonic stem cells (hESCs) are pluripotent cells capable of differentiating into any cell type of the body. It has long been known that the adult stem cell niche is vital for the maintenance of adult stem cells. The cornea at the front of the eye is covered by a stratified epithelium that is renewed by stem cells located at its periphery in a region known as the limbus. These so-called limbal stem cells are maintained by factors within the limbal microenvironment, including collagen IV in basement membrane and limbal fibroblasts in the stroma. Because this niche is very specific to the stem cells (rather than to the more differentiated cells) of the corneal epithelium, it was hypothesized that replication of these factors in vitro would result in hESC differentiation into corneal epithelial-like cells. Indeed, here we show that culturing of hESC on collagen IV using medium conditioned by the limbal fibroblasts results in the loss of pluripotency and differentiation into epithelial-like cells. Further differentiation results in the formation of terminally differentiated epithelial-like cells not only of the cornea but also of skin. Scanning electron microscopy shows that some differences exist between hESC-derived and adult limbal epithelial-like cells, necessitating further investigation using in vivo animal models of limbal stem cell deficiency. Such a model of hESC differentiation is useful for understanding the early events of epithelial lineage specification and to the eventual potential application of epithelium differentiated from hESC for clinical conditions of epithelial stem cell loss.
Disclosure of potential conflicts of interest is found at the end of this article.
The inner cell mass (ICM) of the developing embryo gives rise to all three germ layers of the embryo itself (ectoderm, mesoderm, and endoderm). Isolation of the ICM from the rest of the embryo (trophectoderm in particular) and its subsequent culture results in the formation of embryonic stem cells (ESC) [1, –3]. These cells are pluripotent and capable of forming cells from all three germ layers both in vitro and in vivo (in the form of teratomas [1, –3]). Mouse ESCs (mESCs) were first derived from hatched blastocysts in 1981 , and human ESC (hESC) were subsequently derived in 1998 from discarded in vitro fertilization embryos . For the purposes of the studies outlined in this article, the two hESC lines used were the H1 (WiCell Research Institute Inc., Madison, WI, http://www.wicell.org) and the hES-NCL1 (University of Newcastle, Newcastle upon Tyne, U.K.)  cell lines.
The cornea is the clear front of the eye, and it is composed of three main layers—the outer stratified epithelium, the stroma, and the inner single-cell layered endothelium. The corneal epithelium is maintained by stem cells (SCs) located at the periphery of the cornea, in a region known as the limbus . Radiolabeling studies have shown that these so-called limbal stem cells (LSCs) are located in the basal layer of the limbal epithelium . The corneal epithelium itself is devoid of its own stem cells. Various factors within the LSC niche are thought to contribute to their SC state. These include close proximity to vasculature, which brings nourishment and important blood-borne factors ; the basement membrane (BM) composition of the limbal epithelium, including specific isoforms of collagen IV, laminin, and fibronectin ; and close proximity to limbal fibroblasts in the underlying stroma, which produce various cytokines that aid corneal epithelial wound healing .
It is now well established that the niche plays an important role in the maintenance of stem cell properties in several tissues, and this is also true in the case of the LSC niche [8, , –11]. The best interaction between the stem cells and the niche itself is exemplified by hematopoietic stem cell (HSC) contact with osteoblasts in the bone marrow that are thought to provide Notch ligand and cadherin interactions to maintain quiescent HSCs in vivo [12, 13]. The inductive properties of the niche have been used to direct differentiation of ESC to desired lineages. Successful examples include differentiation of hESC to dopaminergic neurons using the stromal cell line PA6  as well as hematopoietic differentiation of both murine and human ESCs to hematopoietic lineages using stromal cells derived from the bone marrow and fetal liver [15, –17]. In this study, we have endeavored to replicate the LSC niche in vitro using extracellular matrix (ECM) components and limbal fibroblasts and have investigated the effects of the LSC niche into the differentiation capacity of hESCs. Collagen IV and laminin are major basement membrane components of the limbal and corneal epithelia [6, 18, 19]. Fibronectin is an ECM glycoprotein that is deposited in the ECM in response to corneal epithelial wound healing and promotes epithelial adhesion and migration [20, 21]. For these reasons, collagen IV, laminin, and fibronectin were investigated for their suitability at directing the differentiation of hESC into corneal epithelial-like cells and/or maintenance of newly generated corneal epithelial progenitors in culture. Indeed, previous studies have already shown that collagen IV can be used to differentiate mESC into corneal epithelial cells . Limbal fibroblasts form the major cellular component of the limbal stroma upon which the LSCs reside [7, 8], and they produce specific cytokines that promote corneal epithelial wound healing by the LSCs . For these reasons, limbal fibroblasts were used to condition epithelial medium for subsequent use in the hESC differentiation studies outlined in this article.
The potential ability of ESC to differentiate along epithelial lineages has been the subject of several recent studies. In particular, skin-like epithelial cells have been differentiated from ESC using various methods [22, , , , –27]. The first differentiation methods have relied heavily on coculture with human fibroblasts or 3T3 mouse embryonic fibroblast-derived ECM to drive the differentiation of mESC into skin epithelial cells [23, 24]. Another method of differentiating ESC into skin-like epithelium is by the injection of hESC into severe combined immunodeficiency mice, which results in the formation of teratomas after approximately 2 months [25, 26]. Isolation and disaggregation of these teratomas and subsequent culture on a 3T3 fibroblast feeder layer has been shown to promote skin epithelial cell differentiation from H9 hESC line . Recently, it has been shown that cells expressing cytokeratin (CK) 12, a marker specific to terminally differentiated cells (TDCs) of the corneal epithelium, can be obtained by plating mESC onto collagen IV . Moreover, transplantation of such 8-day-old differentiation cultures onto the surface of a cornea denuded of its epithelium results in the formation of an epithelium with morphological similarities to corneal epithelium 24 hours post-transplant, suggesting that functional corneal cells can be obtained from murine ESCs .
Materials and Methods
Preparation of Media
Fibroblast medium without fetal calf serum (FCS) consisted of low-glucose Dulbecco's modified Eagle's medium (DMEM) without pyruvate, 1% nonessential amino acids, 1% penicillin-streptomycin, and 1% l-glutamine (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Fibroblast medium was also made up using the same recipe and supplementing with 10% FCS (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Epithelial medium was made up containing three parts of low-glucose DMEM with pyruvate (Gibco, Grand Island, NY, http://www.invitrogen.com) and one part of Ham's F12 medium (Gibco), 10% FCS (Gibco), 1% penicillin-streptomycin (Gibco), hydrocortisone (Sigma-Aldrich), insulin (Sigma-Aldrich), tri-iodothyronine (Sigma-Aldrich), adenine (Sigma-Aldrich), cholera toxin (Sigma-Aldrich) and epidermal growth factor (Sigma-Aldrich). All media were filter sterilized using a 0.22-μm filter (Millipore, Billerica, MA, http://www.millipore.com) and stored at 4°C.
Coating of Tissue Culture Plates with Extracellular Matrix Components
Lyophilized collagen IV from human placenta (Sigma-Aldrich) was reconstituted with 0.25% acetic acid (VWR International, http://uk.vwr.com) to a concentration of 0.5 mg/ml. Upon adding the acetic acid, the mixture was placed at 4°C for 3 hours with intermittent swirling. Two-square-centimeter tissue culture wells were coated with collagen IV by adding 200 μl of this collagen solution and then placing the culture plates at 4°C overnight. The following morning, the collagen IV solution was removed, and the wells were briefly washed with phosphate-buffered saline (PBS) before plating of cells. Laminin solution from human placenta (Sigma-Aldrich) was thawed at 4°C. A 1:25 dilution of this solution in PBS was made and applied for 2 hours in a tissue culture incubator (at 37°C with a humidified atmosphere containing 5% carbon dioxide) to 2-cm2 culture wells. The excess laminin solution was then removed, and the wells were briefly irrigated with PBS before plating of cells. Lyophilized fibronectin from human foreskin fibroblasts (Sigma-Aldrich) was reconstituted with sterile water to a concentration of 0.5 mg/ml. A 1:10 dilution of this fibronectin solution in PBS was made. Two cm2 tissue culture wells were coated with fibronectin by adding 200 μl of this diluted fibronectin solution and incubating for 1 hour at room temperature. The excess fibronectin solution was then removed, and the wells were briefly irrigated with PBS before plating of cells.
Culture of Human Limbal Epithelium Using the Various Extracellular Matrix Components
Cadaveric limbal tissue composed of peripheral cornea and limbus were obtained from UK Transplant (Bristol, U.K., http://www.uktransplant.org.uk; consent for research had been obtained). The deeper layers of the limbal rings were dissected away and the remaining limbal tissue containing limbal epithelium was cut into 1-mm2 pieces. These limbal pieces were incubated with 0.05% trypsin solution (Sigma-Aldrich) for 20 minutes in a tissue culture incubator. The resulting cell suspension was removed from the limbal pieces, and epithelial medium was added to this suspension. The cell suspension was then centrifuged for 3 minutes at 1,000 rpm, and the supernatant was then removed. The remaining cell pellet was resuspended in epithelial medium. This process of trypsinization of the limbal pieces and centrifugation of the resulting cell suspension was repeated for a further three times using the same limbal tissue. The resulting limbal cell suspensions were pooled together. After we performed a count of the viable cells, 30,000 viable limbal epithelial cells in epithelial medium were added to the 2-cm2 tissue culture wells coated with each of the ECM components. In addition to the cultures established on the ECM components, cocultures of limbal epithelial cells and mitotically inactivated 3T3 mouse fibroblasts (plated at a density of 24,000 per cm2) were also established for comparison purposes. The 3T3 fibroblasts were mitotically inactivated by incubation with 10 μg/ml mitomycin C (Sigma-Aldrich) in FCS-containing fibroblast medium for 2 hours. All cultures were maintained in a tissue culture incubator (at 37°C with a humidified atmosphere containing 5% carbon dioxide) and fed with epithelial medium on the third day and every other day thereafter.
Colony-Forming Efficiency Assays
To determine the efficiency of the limbal epithelial cells from the various ECM component cultures, colony-forming efficiency (CFE) assays were performed. Mitotically inactivated 3T3 fibroblasts were plated in a 9.6-cm2 tissue culture well at a density of 24,000 per cm2 and incubated in a tissue culture incubator overnight. The following day, after performing a count of the viable limbal epithelial cells, 300–1,000 viable cells in epithelial medium were plated on the 3T3 fibroblasts. This CFE assay was then placed in a tissue culture incubator, and the epithelial medium was changed on the third day and then every other day thereafter with fresh epithelial medium. On the 12th day of culture, after removal of the epithelial medium, the assay culture was briefly irrigated with PBS and then fixed with 3.7% formaldehyde (BDH) in PBS for 10 minutes at room temperature. The formaldehyde solution was then removed, and the culture was briefly irrigated with PBS and then incubated with 1% rhodamine B (Sigma-Aldrich) in methanol (BDH) for 10 minutes at room temperature. The number of colonies formed after the 12 days was then counted. The CFE (%) was calculated using the formula number of colonies formed/number of cells plated ×100. To determine the extent of epithelial differentiation of the hESCs to epithelial-like lineages, undifferentiated and differentiated hESCs were also plated on 3T3 fibroblasts to assess their CFEs.
To determine the extent of hESC differentiation within the cultures plated on various ECM components, flow cytometry was performed. A cell suspension of epithelial cells was obtained by trypsinization and then centrifuged (all centrifugation steps were performed for 3 minutes at 1,000 rpm). The supernatant was removed and the cell pellet was resuspended in 100 μl of 1× FACS Permeabilizing Solution 2 (BD Biosciences, San Diego, http://www.bdbiosciences.com) in distilled water. The resultant suspension was incubated for 10 minutes at room temperature. After centrifugation, the supernatant was removed and the remaining cell pellet was resuspended in 1 ml of 5% FCS in PBS. After repeat centrifugation and removal of the supernatant, the cell suspension was resuspended in 100 μl of primary antibody diluted in PBS (supplemental online Table 1). The resulting suspension was incubated for 30 minutes at 4°C. After adding 1 ml of PBS supplemented with 5% FCS to the cell suspension, centrifugation was performed. After removal of the supernatant, the cell pellet was resuspended in 100 μl of fluorescein isothiocyanate (FITC)-conjugated sheep anti-mouse immunoglobulins secondary antibody diluted in PBS. This suspension was then incubated in the dark for 30 minutes at 4°C. After the addition of 1 ml of PBS supplemented with 5% FCS to the cell suspension, centrifugation was performed. After removal of the supernatant, the cell pellet was resuspended in 500 μl of PBS supplemented with 5% FCS. This final suspension was analyzed using a FACSCalibur flow cytometer (BD Biosciences) and the results analyzed using CellQuest Pro (BD Biosciences). A similar staining procedure was applied when undifferentiated and differentiated hESCs were analyzed by flow cytometry for the expression of various cell surface antigens with the exception that the permeabilization step was removed.
Isolation and Culture of Limbal Fibroblasts from Human Limbal Tissue
Cadaveric human limbal tissue, donated and consented for research, was obtained from UK Transplant. The limbal tissue was cut into 1-mm2 pieces. A 3-mg/ml solution of collagenase IV (Gibco) in fibroblast medium without FCS was added to these limbal pieces. The resulting mixture was incubated for 1 hour in a tissue culture incubator. After 1 hour, the collagenase IV solution was removed from the limbal pieces and discarded. A further 3 mg/ml fresh collagenase IV solution was added to the limbal pieces and the mixture was incubated for 8 hours in the tissue culture incubator. After 8 hours of incubation, the collagenase IV solution containing cells released from the limbal pieces was removed and centrifuged for 3 minutes at 1,000 rpm. The supernatant was removed, and the resulting pellet was suspended in fibroblast medium containing 10% FCS. The cell suspension was placed in a 2-cm2 tissue culture wells and incubated overnight in a tissue culture incubator. The following day, the limbal fibroblast culture was fed with fibroblast medium, and then every 2–3 days thereafter. The human limbal fibroblasts were expanded by subculturing thereafter up to a maximum of 10–15 passages.
Conditioning of Epithelial Medium by the Limbal Fibroblasts
The limbal fibroblasts were mitotically inactivated by adding 10 μg/ml mitomycin C to the culture medium and incubating these cells for 2 hours at 37°C. These fibroblasts were washed carefully three times with PBS and replated on tissue culture flasks at a density of 56,000 viable cells per cm2. The tissue culture flasks were then placed in a tissue culture incubator overnight. The following morning, the fibroblast medium was removed, the flasks were briefly irrigated with PBS, and 400 μl/cm2 epithelial medium was added. The flasks were maintained in a tissue culture incubator. The limbal fibroblast-conditioned epithelial medium was collected daily and replaced with 400 μl/cm2 fresh epithelial medium for a total of 7 days. The conditioned epithelial medium was stored at −20°C. After 7 days of repeated collection, all stored conditioned epithelial medium was centrifuged for 3 minutes at 1,000 rpm. The conditioned epithelial medium was removed from any cell pellets and filter sterilized using a 0.22-μm filter. The filtered conditioned epithelial medium was stored at 4°C and used within a month or stored at −20°C for a maximum of 3 months.
Culture of Human Embryonic Stem Cells
Both of the hESC lines were maintained on a feeder layer of mouse embryonic fibroblasts as previously described . The hESC colonies from the two hESC lines (hES-NCL1 and H1) were released from the culture using manual passaging. Approximately 5–10 colonies in limbal fibroblast conditioned epithelial medium were then plated in each 2-cm2 tissue culture well coated with the most efficient ECM component determined from the limbal epithelial cultures The culture medium was replaced every other day with limbal fibroblast-conditioned epithelial medium. The wells were viewed regularly under an inverted microscope (Zeiss, Jena, Germany, http://www.zeiss.com) and photographs taken using Axiovert software (Zeiss).
Immunocytochemistry of Cell Cultures
The medium in the tissue culture well was removed and the well briefly irrigated with PBS. The well was then incubated with 3.7% formaldehyde (Sigma-Aldrich) in PBS for 30 minutes at room temperature, PBS three times for 5 minutes at room temperature, 0.5% Triton X-100 (Sigma-Aldrich), 2% sheep serum in PBS for 1 hour at room temperature, and diluted primary antibody (supplemental online Table 1) in PBS overnight at 4°C. The following day, the primary antibody was removed from the culture well and the well incubated with PBS three times for 5 minutes at room temperature. The well was incubated with 10 μg/ml FITC-conjugated sheep anti-mouse Igs (Sigma-Aldrich) diluted in PBS for 30 minutes at room temperature in the dark. After removing the secondary antibody, the well was incubated three times with PBS for 5 minutes in the dark. The cells in the well were then incubated with a solution of 10 μg/ml Hoechst 33342 in sterile water for 10 minutes at room temperature in the dark. After removal of the Hoechst 33342 solution, each well was incubated three times with PBS in the dark, and then left filled with PBS. The wells were then viewed under an inverted microscope (Zeiss) immediately and photographs taken using the Axiovert software. For double staining, cells were fixed, permeabilized, and blocked as described, before incubation with the first antibody (p63, CK12, or CK3/12) for 1 hour. The cells were washed with 5% FCS and PBS and then incubated with the second antibody (CK3/12, CK12, or CK10) for an additional hour. The cells were again washed with 5% FCS and PBS before addition of secondary antibodies (tetramethylrhodamine isothiocyanate-conjugated anti-mouse IgG1 [RDI, Concord, MA, http://www.researchd.com], 1:100 dilution; FITC-conjugated anti-mouse IgG2, 1:100 dilution [RDI]; Rhodamine/FITC-conjugated anti-rabbit IgG [Chemicon, Temecula, CA, http://www.chemicon.com], 1:100 dilution; or Rhodamine-conjugated anti-goat IgG [Chemicon], 1:100 dilution for 30 minutes). The cells were washed before fluorescence microscopy.
RNA Extraction and Isolation from Cultures
The tissue culture well was incubated with 1 ml of TRIzol reagent (Invitrogen) for 5 minutes at room temperature. After collecting the resulting solution from the well, 0.2 ml of chloroform (BDH) was added to this solution. The tube containing the reaction mixture was shaken vigorously for 15 seconds, and then centrifuged at 12,000g for 15 minutes at 4°C. The colorless phase of the centrifuged reaction mixture was removed, 0.5 ml of isopropyl alcohol (BDH) was added, and the reaction was incubated for 10 minutes at room temperature, and then centrifuged at 12,000g for 10 minutes at 4°C. The supernatant was removed from the resulting RNA pellet, and the pellet was allowed to dry for 10 minutes at room temperature. The dried pellet was then dissolved in 11 μl of sterile water, and this reaction incubated for 10 minutes at 60°C. Either this RNA mixture was stored at −80°C or reverse transcription (RT) was performed.
Before reverse transcription (RT), the concentration of RNA was assessed by analyzing 1 μl of the RNA using a NanoDrop (LabTech International, East Sussex, U.K., http://www.labtech.co.uk). A 10-μl final solution containing 2 μg of RNA, 1× DNase reaction buffer (40 mM Tris hydrochloride at pH 8, 10 mM magnesium chloride, and 1 mM calcium chloride), 1 unit of DNase/μg of RNA used, and sterile water (all reagents from Promega, Madison, WI, http://www.promega.com). This solution was incubated at 37°C for 30 minutes. After the addition of 1 μl of DNase stop solution (20 mM egtazic acid at pH 8; Promega), the mixture was incubated at 65°C for 10 minutes. One microgram of random primers (Promega) and 1.5 μl of sterile water were added to this mixture, and the resulting mixture was incubated at 70°C for 5 minutes. The mixture was then placed on ice for 5 minutes. The cooled mixture was made up to a 25-μl final solution containing 1× reverse transcriptase reaction buffer (25 mM Tris hydrochloride at pH 8.3, 37.5 mM magnesium chloride, and 5 mM dl-dithiothreitol), 0.5 mM deoxynucleotide phosphate mix (0.5 mM each of deoxyadenosine triphosphate, deoxycytidine triphosphate, deoxythymidine triphosphate, and deoxyguanosine triphosphate), 25 units of rRNasin ribonuclease inhibitor, and 200 units of reverse transcriptase (all reagents from Promega). The resulting mixture was incubated at 37°C for 60 minutes and then 99°C for 5 minutes. This final mixture was either stored at −20°C or used for real-time RT polymerase chain reactions (RT-PCRs).
LightCycler capillary tubes (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) were placed in cooled centrifugation tubes (Roche), and each capillary tube was filled with 1 μl of cDNA, 10 μl of QuantiTect SYBR Green PCR Master Mix, 1 μl each of specific 10 μM forward and reverse primers (MWG Biotech, Ebersberg, Germany, http://www.mwg-biotech.com; supplemental online Table 2), and 7 μl of sterile water (all reagents from Qiagen, Hilden, Germany, http://www1.qiagen.com). The filled LightCycler tubes were briefly centrifuged at the lowest setting of a microcentrifuge (Eppendorf, Hamburg, Germany, http://www.eppendorf.com). The capillary tubes were then removed from the centrifugation tubes and then placed in a LightCycler (Roche). Real-time RT-PCR was performed using the LightCycler at 95°C for 15 minutes, followed by 50 cycles at 94°C for 15 seconds, primer-specific annealing temperature for 30 seconds, and 72°C for 20 seconds, with a single data-acquisition step. The crossing point for each transcript was determined using the LightCycler software (Roche), and the LightCycler Relative Quantification software (Roche) was used to analyze the data. The gene-to-glyceraldehydes-3-phosphate dehydrogenase (GAPDH) ratio was calculated using hESC samples and limbal or skin epithelial cell samples as the reference points for each gene investigated.
Scanning Electron Microscopy
Thermanox round plastic coverslips (Agar Scientific, Essex, U.K., http://www.agarscientific.com) were placed in the culture wells, and these wells used for the culture of differentiating hESCs. Both differentiating hESC cultures (as described herein) and human limbal epithelial 3T3 fibroblast cocultures were established on the coverslips in the culture wells. The coverslips were removed from the culture wells on the seventh day of culture and fixed in 2% gluteraldehyde in Sorenson's phosphate buffer (SPB; TAAB Laboratory Equipment, Berkshire, U.K., http://www.taab.co.uk) overnight at 4°C. The coverslips with the fixed cultures were then washed twice in SPB for 15 minutes at room temperature, and then dehydrated in 25% ethanol (BDH) in sterile water for 30 minutes at room temperature, 50% ethanol in sterile water for 30 minutes at room temperature, and 75% ethanol in sterile water for 30 minutes, and finally stored in 100% ethanol at 4°C before processing. The cultures on the coverslips were then dehydrated further with carbon dioxide in a Samdri 780 Critical Point Dryer (Tousimis, Rockville, MD, http://www.tousimis.com). The coverslips were mounted on an aluminum stub using Achesons Silver ElectroDag (Agar Scientific) and the cultures on the coverslips coated with 15 nm of gold using a Polaron scanning electron microscopy Coating Unit (Empdirect, Houston, TX, http://www.empdirect.com). The specimens were examined using a Stereoscan 240 SE microscope and photographs taken (Leica).
All statistical analysis was carried out using Student's t test. Results were considered significant at p < .05.
Effects of Extracellular Matrix on the Differentiation of hES-NCL1
Human limbal epithelial cultures were successfully established on all the three ECM components as well as traditional 3T3 fibroblast cocultures. CFEs on the cultured limbal epithelial cells from all four cultures (using collagen IV, laminin, and fibronectin and the 3T3 fibroblast coculture) showed that the colony-forming ability was maintained equally well in ECM-coated plates when compared to the 3T3 coculture as the gold standard (Fig. 1A). There were, however, statistically significant differences in the CFE among the three ECM components with the collagen IV and fibronectin providing the best substrate (paired t test; p < .05). The expression of p63, a transcription factor used to identify epithelial SCs from various tissues including the limbus, was maintained in 3T3 coculture as well as ECM coated plates (Fig. 1B). On the basis of these results and those of previous publications that have used collagen IV to differentiate corneal epithelial cells from mESCs , it was decided to use collagen IV-coated plates to differentiate the hESC cells in the subsequent studies.
Isolation and Culture of Human Limbal Fibroblasts
Human limbal fibroblasts were successfully isolated from cadaveric human limbal tissue and cultured on tissue culture plastic using FCS-containing fibroblast medium. The cultured cells had fibroblast morphology and were slow to grow initially, but reached confluence within the first 7 days of culture (Fig. 2A, 2B). Fibroblasts have been isolated from corneal and limbal tissue previously using the techniques outlined here . Flow cytometry analysis of confluent fibroblast cultures indicated that they expressed CD44 (hyaluronan receptor) and CD90 (THY-1) cell surface markers (Fig. 2C) while lacking the expression of mesenchymal cell specific markers CD106 (V-CAM1), CD71 (transferrin receptor) and expression of endothelial-specific cell marker CD31 (PECAM-1; data not shown). A similar expression profile has been obtained by our group for hESC-derived and human foreskin fibroblasts , thus confirming the fibroblastic nature of these limbal-derived stromal cells. Flow cytometry analysis combined with RT-PCR and immunocytochemistry using p63-, CK3/12-, or CK12-specific antibodies revealed that human limbal fibroblasts (LFs) and human ESCs do not express any of the markers expressed by the limbal stem cells, such as ΔNp63, or more terminally differentiated limbal epithelial cells, such as CK3 and CK12 (Fig. 2D–2F; supplemental online Fig. 1).
Morphological Changes in the Differentiated Human Embryonic Stem Cells
The hESC colonies attached downwell to the collagen IV-coated culture wells within a few hours of plating (Fig. 2A) and then spread out by the following day. Phase contrast observation of the differentiating cultures showed that, within the first 3 days of differentiation, the undifferentiated colonies began to change morphology substantially into much flatter and more spread-out-looking cells (Fig. 3B). After 6 days of differentiation, the majority of the hESC culture was composed of these flatter cells (Fig. 3C). Moreover, on top of these flat cells, other large round elevated cells could be seen, forming a meshwork-type pattern. The number of these large round elevated cells increased from day 9 to day 15 of differentiation (Fig. 3D–3F). By the third week of differentiation, there was a significant absence of the flatter-looking cells, and the majority of the culture was composed of the large round cells that formed a meshwork pattern (data not shown). Morphological changes in both the hES-NCL1 and H1 cell colonies during the differentiation time period studied were very similar.
Scanning electron microscopy was performed on hESCs differentiated for 1 week using collagen IV- and limbal fibroblast-conditioned medium (Fig. 3G) and on cultured human limbal epithelial cells cocultured for 1 week with 3T3 fibroblasts (Fig. 3H). The first significant similarity between the two types of cells which was observed using scanning electron microscopy was the presence of microcilia. Limbal epithelial cells, similar to other epithelial cells, have multiple microcilia. The differentiated hESCs also had multiple microcilia, although these cilia were more in number and also appeared much longer than did those seen on the limbal epithelial cells. Importantly, undifferentiated hESCs did not possess microcilia (data not shown). The most significant difference between the two types of cultures was that the differentiated hESCs were still much smaller than the cultured limbal epithelial cells.
Differentiation of Human Embryonic Stem Cells into Epithelial-Like Cells
The loss of hESC pluripotency during the differentiation process was confirmed by real-time RT-PCR analysis, flow cytometry, and immunocytochemistry. Real-time RT-PCR showed that the expression of undifferentiated hESC cell markers OCT4  and NANOG [30, 31] declined significantly over the 21-day differentiation time period (Fig. 4A, 4B; supplemental online Fig. 2). Despite an initial increase in OCT4 during the first 3 days of differentiation, OCT4 declined to low levels by day 9. This was the case for both hES-NCL1 and H1 hESC lines, although the initial increase was significantly more so for hES-NCL1 hESC. NANOG expression also declined to negligible levels by the sixth day of differentiation. This decline was almost identical for hES-NCL1 and H1 hESC lines. Flow cytometry on the differentiating cultures for the undifferentiated hESC marker stage-specific embryonic antigen 4 (SSEA4) [2, 3, 32, 33] also showed that this marker declined significantly to low levels by day 21 of differentiation (Fig. 4C). However, there were approximately 10% of cells in the culture that still expressed SSEA4 by day 21 of differentiation. This was confirmed by immunocytochemistry, which showed that the expression of undifferentiated hESC markers SSEA4 and OCT4 could still be seen in some regions of the differentiated hESC cultures by the end of the differentiation time period (Fig. 5).
The differentiation of hESC into epithelial-like cells was confirmed by real-time RT-PCR analysis, flow cytometry, and immunocytochemistry. For the hES-NCL1 cell line, the expression of p63 by real-time RT-PCR peaked at days 6–9 of differentiation, showing almost 130-fold increase when compared to undifferentiated hESCs (Fig. 6A). In the case of the H1 cell line, the peak in p63 expression was much less pronounced (p = .0001, day 9 between two cell lines) and more gradual, showing more than a 30-fold increase in expression. Flow cytometry for p63 showed a peak in expression at day 6 of differentiation (Fig. 6B). This was the case in both the hES-NCL1 and H1 hESC lines, although the extent of the peak differed between the two cell lines, corroborating the real-time RT-PCR data. We noticed that the differences shown in expression of p63 by real-time RT-PCR were greater than those shown by flow cytometry. These changes can be explained by the fact that the primers selected for real-time RT-PCR amplify ΔNp63 specifically, whereas the antibody that is available for flow cytometry cannot distinguish between the aforementioned isoform and the rest. Real-time RT-PCR analysis for expression of the terminally differentiated cell (TDC) marker CK3 (for corneal epithelial TDCs) showed a peak of expression during later stages of the 21-day differentiation time period (Fig. 6C). It was interesting to observe an earlier peak of expression for CK12 compared to CK3 (Fig. 6D and 6C). Because specific antibodies for CK3 only are not commercially available, we carried out flow cytometry analysis using an antibody that detects the CK3/12 dimer, thus allowing us to get some information on expression of both markers. This analysis showed a peak in expression by days 6–9 (Fig. 6E), which mostly corroborates the real-time RT-PCR data on the CK12 expression. This was the case for both the hES-NCL1 and H1 hESC lines, with approximately half of the hESCs expressing CK3/12 at their peak. In addition to expression of CK3, real-time RT-PCR also revealed the expression of CK10, a known marker of TDCs of the skin epithelial cells (Fig. 6F). This suggests that our differentiation protocol leads to generation of more than one type of epithelial-like cell.
Immunocytochemistry of fixed cells showed that the expression of both p63, a marker for epithelial SCs from various tissues (including LSCs) , and CK3/12, the TDC marker specific to corneal epithelium [35, 36], could be detected from week 1 of differentiation (Fig. 7). During subsequent stages of the differentiation time period, immunocytochemistry analysis showed that the numbers of cells expressing p63 declined, but the expression of CK3/12 continued into the second week of differentiation, after which time it also declined (data not shown). It was interesting to observe that p63 expression was also noticeable in the cytoplasm of these cells as well as in the nucleus; this was, however, observed in some of the differentiating cultures, but not all (Fig. 7B, 7C). Cytoplasmic and nuclear expression of p63 has been observed in lung adenocarcinoma cells and suggested to be important for the oncogenic effect of p63, probably through interaction with other molecules in the cytoplasm . The significance of these findings at present is unclear and merits further investigations. Double-staining experiments at weeks 1 and 2 of differentiation suggested no colocalization between the limbal stem cell marker, p63 and more terminally differentiated cell markers (CK12 or CK3/12; Fig. 7G–7K) or colocalization of skin epithelial cell markers (CK10) to corneal epithelial cell marker (CK12 or CK3/12, data not shown), although all these markers were detected individually in the same cultures between day 7 and 14 of differentiation.
CFEs were performed on undifferentiated and differentiated hESC (at days 0, 7, 14, and 21 of differentiation). CFEs using 3T3 fibroblast coculture are indicative of the epithelial clonogenic capacity of the plated cells. A total of 5,000 cells were plated for each CFE assay, and the hES-NCL1 differentiated cells were used mainly because of the greater p63 expression as shown by real-time RT-PCR and flow cytometry. The CFE analysis revealed that the CFE increased from day 0 to day 7 and then started to decline rapidly to a very low level by day 21 (Fig. 7L, 7M), corroborating the real-time RT-PCR and flow cytometry data for p63.
Studies outlined in this article highlight the importance of the SC niche for inducing differentiation of human ESC into the corneal epithelial-like cells, which can be achieved by replicating factors present in the LSC niche within the in vitro culture system [8, 10]. Our experiments showed that the culture of hESC on collagen IV-coated plates with conditioned medium from corneal limbal fibroblasts resulted in differentiation of these cells into corneal and skin-like epithelial cells as confirmed by various methods. During the course of differentiation we observed a peak in p63 expression, which is likely to indicate the emergence of early epithelial progenitor cells. This was followed by the appearance of more differentiated corneal-like epithelial cells marked by the expression of CK3/12 and CK12 and skin-like epithelial cells noted by the expression of CK10 at the later stages of differentiation. On the basis of our flow cytometry and real-time RT-PCR data, it appears that, although the major differentiating lineage is more than likely to be the corneal-like epithelium, other epithelial lineages (such as skin) are also likely to have been formed.
Scanning electron microscopy reveals that, like epithelial cells and unlike undifferentiated hESCs, the differentiated hESC have ciliated structures on their cell surface. Notwithstanding this, important differences certainly exist between these and adult epithelial cells. The scanning electron microscopy images firstly show that the differentiated hESCs are much smaller than cultured limbal epithelial cells. Secondly, cilia on the differentiated hESCs are more numerous and characteristically much longer than those of limbal epithelial cells. These data suggest that some differences exist between LSCs and hESC-derived corneal progenitors, and necessitate comparative studies that can address the differences/similarities between these two cell types. In addition, our data necessitates in vivo functional studies in animal models of limbal stem cell deficiencies (LSCs) to investigate whether the hESC-derived cells function in the same way as do LSCs.
Although similar morphological changes are noted in both of the hESC lines within the first few days of differentiation, the flow cytometry and real-time RT-PCR studies reveal differences in their ability to differentiate in the same way. The H1 cell line appears to show a less pronounced peak of p63 expression by real-time RT-PCR than does hES-NCL1. However, the H1 hESC line does still form TDC-like cells of the corneal epithelium and skin. It has been noted in various studies that characteristic differences exist between different hESC lines [38, –40]. This can result from various factors such as the embryonic stage from which the hESCs were derived, culture conditions used to derive the hESC lines, or inherent propensity to differentiate toward selected lineages. However, recent work from our group has indicated that the embryonic stage at which the hESC line is derived is not as significant, at least when transcriptional profiling of the two cell lines has been investigated . Because the two cell lines were maintained under the same conditions and by the same investigator in our lab, differences between the two cell lines to differentiate toward epithelial lineages are more likely to result from the inherent properties of each cell line, and can be studied more in depth by looking at their epigenetic rather than their transcriptional profiles, as suggested by recent studies .
This article describes, for the first time, the differentiation of hESCs to corneal-like epithelial lineages, and this provides a first step toward refinement of protocols to produce these cells for potential therapeutic purposes. Notwithstanding this success, several improvements to the technique need to be carried out before their therapeutic potential can be investigated. Firstly, functional studies to prove the ability of hESC-derived corneal-like cells to reconstruct a corneal epithelium in vitro or in vivo has to be carried out to prove that they can function in the same way as LSCs. Secondly, a low but not negligible fraction of cells that express hESC cell surface marker SSEA4 still remain by the end of the differentiation process. This can result in tumor formation on transplantation, and, therefore, steps have to be taken to either enhance the differentiation process or selectively purify these cells before transplantation to eliminate this risk. Thirdly, our differentiation protocol still relies on the presence of FCS and conditioned medium from limbal fibroblasts, and efforts are under way in our group to understand the nature of the secreted factors produced by limbal stromal cells that can be used to refine this protocol to good manufacturing practice standards. Fourthly, the lack of cell surface markers for LSCs and early epithelial progenitors renders difficult the selection and molecular profiling of these cells by flow cytometry. Efforts are under way in our group to create hESC clones in which the expression of endogenous p63 is mimicked by a green fluorescent protein reporter. This will allow the selection of early epithelial-like progenitors and investigation of transcriptional machinery that drives their further differentiation into corneal or skin differentiated cells. Fifthly, the differentiated hESCs remain immunogenic and will not avoid the requirement of immunosuppression in the recipient . Having said this, it has recently been suggested that 150 random hESC lines could provide a suitable human leukocyte antigen match for most recipients . In view of these, the differentiation model we have provided in this article is more likely at present to provide an invaluable tool for understanding early developmental changes that occur during embryonic formation of ectoderm and epithelium and, in particular, corneal epithelium .
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
We thank Vivian Thompson and Tracey Scott-Davey at the Biomedical Electron Microscopy Unit (University of Newcastle upon Tyne, U.K.), Ian Dimmick for help with flow cytometry, and Dennis Kirk for technical assistance. This work was supported by the Newcastle Healthcare Charity, Life Knowledge Park and One North East Regional Development Agency. M.S. is currently affiliated with Centro de Investigación Príncipe Felipe, Valencia, Spain.