Successful Application of Ex Vivo Expanded Human Autologous Oral Mucosal Epithelium for the Treatment of Total Bilateral Limbal Stem Cell Deficiency


  • Sai Kolli,

    1. Institute of Genetic Medicine, Newcastle University, Newcastle, United Kingdom
    2. Department of Ophthalmology, Royal Victoria Infirmary, Newcastle University, Newcastle, United Kingdom
    Search for more papers by this author
  • Sajjad Ahmad,

    1. Institute of Genetic Medicine, Newcastle University, Newcastle, United Kingdom
    2. Department of Ophthalmology, Royal Victoria Infirmary, Newcastle University, Newcastle, United Kingdom
    Search for more papers by this author
  • Hardeep Singh Mudhar,

    1. National Specialist Ophthalmic Pathology Service Laboratory (NSOPS), Department of Histopathology, Royal Hallamshire Hospital, Sheffield, United Kingdom
    Search for more papers by this author
  • Adam Meeny,

    1. National Specialist Ophthalmic Pathology Service Laboratory (NSOPS), Department of Histopathology, Royal Hallamshire Hospital, Sheffield, United Kingdom
    Search for more papers by this author
  • Majlinda Lako,

    Corresponding author
    1. Institute of Genetic Medicine, Newcastle University, Newcastle, United Kingdom
    • Correspondence: Majlinda Lako, Ph.D., Newcastle University, Institute of Genetic Medicine, International Centre for Life, Central Parkway, Newcastle upon Tyne NE1 3BZ, U.K. Telephone: 44-191-241-8688; Fax: 44-191-241-8666; e-mail:

    Search for more papers by this author
  • Francisco C. Figueiredo

    1. Institute of Genetic Medicine, Newcastle University, Newcastle, United Kingdom
    2. Department of Ophthalmology, Royal Victoria Infirmary, Newcastle University, Newcastle, United Kingdom
    Search for more papers by this author


Ocular surface reconstruction with ex vivo expanded limbal stem cells (LSCs) is a widely used clinical treatment for patients with limbal stem cell deficiency (LSCD). This is not applicable to patients with bilateral LSCD where there are no remaining LSCs. Cultivated oral mucosa epithelium (OME) has been used as an alternative source of autologous epithelial stem cells for ocular reconstruction in few clinical trials. However, successful generation of stratified OME epithelium has only been achieved in the presence of animal feeder cells and/or animal-derived products in the culture media, likely to contribute to increased risk of pathogen transmission and graft rejection. In this study, we report generation of multilayered OME epithelium that shares many of the characteristics of corneal epithelium using a fully compliant good manufacturing practice, feeder- and animal product-free method. Proof of concept was achieved by transplantation of autologous ex vivo expanded OME in two patients with histologically confirmed bilateral total LSCD that resulted in successful reversal of LSCD in the treated eye up to 24 months. Stem Cells 2014;32:2135–2146


The cornea is covered by a stratified squamous epithelium which serves the dual function of protection of the ocular surface and also allowing high acuity vision. There is now a substantial body of evidence pointing to the basal layer of the limbal epithelium as the location for corneal epithelial stem cells, also known as limbal stem cells (LSCs) [1, 2]. The limbal epithelium is thought to have two main functions, first acting as a reservoir for the replacement of corneal epithelial cells and secondly acting as a barrier preventing the migration of conjunctival epithelium and its blood vessels on to the surface of the cornea. Significant damage to the LSC population can lead to the condition of limbal stem cell deficiency (LSCD) characterized by the inability to maintain a healthy corneal epithelium resulting in chronic ocular pain and loss of vision, corneal neovascularization and conjunctivalization, chronic corneal inflammation, and subsequent scarring.

The only treatment option for total LSCD is surgical and involves transplantation of LSCs to the affected eye, usually in the form of whole tissue limbal grafts [3-5]. The large amount of transplanted tissue and application of long-term systemic immunosuppression (in the case of allogeneic transplants) makes this procedure less desirable for clinical transplantation [6, 7]. Our group and others have reported transplantation of autologous ex vivo expanded LSCs with a high degree of success in treatment of patients diagnosed with unilateral LSCD [8-12]. This treatment however is not applicable for a significant proportion of patients with bilateral and total LSCD. Recently, a novel approach has been to use other autologous sources of epithelial stem cells to replenish the LSC population and of those oral mucosa epithelium (OME) cells have been considered as the most suitable for a variety of reasons including their high proliferative potential, ability to be maintained in culture for long period of time without evidence of keratinization, easy access, and lack of scaring following biopsy [13-17]. To date, a number of clinical studies have reported transplantation of ex vivo expanded autologous OME for ocular surface reconstruction in patients with total bilateral LSCD, including patients with Stephen-Johnson syndrome with reported success rate of 43%–67% [18-25]. It is important to highlight that all these studies have used patients with mixed causes of LSCD. Lack of homogeneity in patients grouping may influence the clinical outcomes as some inherited causes of LSCD also affect development and maintenance of OME. In addition, previously published studies have relied on usage of murine 3T3 feeder layers and/or animal-derived products (fetal calf serum [FCS]) for the ex vivo expansion of OME. The inclusion of animal-derived products in the culture system increases the risk of pathogen transmission, immune reaction, and graft rejection and makes those methods incompatible with good manufacturing practice (GMP) requirements. To date, there is only one published report describing the generation of stratified OME using a defined and serum-free culture system [26]; however, this has not been tested as yet in clinical trials of LSCD treatment. The purpose of this study was to develop a reproducible and safe feeder- and animal product-free system which could be used to culture autologous OME ex vivo under GMP conditions. We have fully characterized the OME expanded under these conditions and as proof of principle, we transplanted two patients with bilateral total LSCD caused by chemical burns. Our results suggest that transplantation of cultured OME under our feeder- and animal product-free system results in successful reversal of LSCD in the treated eye up to 24 months post-transplantation.

Materials and Methods

Ethics and Regulatory Statements

The procedures required for the ex vivo expansion of autologous OME for the treatment of LSCD were approved by the human tissue authority (HTA), local research ethics committee, and Newcastle upon Tyne Hospitals NHS Foundation Trust's Research and Development Department and New Procedures Committee.

Oral Biopsy Procurement

All potential donors/patients were requested to observe strict oral hygiene for a week prior to the biopsy. All potential donors were examined for normal appearing oral mucosa. Immediately preoperatively, the donor was asked to rinse their mouth with oral hexetidine (0.1%) for 1 minute. The proposed area of oral mucosa to be biopsied was anesthetized with a small amount of submucosal 1% lidocaine (0.25–0.5 ml) (Supporting Information Fig. S1A). In all cases, healthy non-keratinized oral mucosa on the inside of the cheek 20 mm behind the angle of the mouth was used. A standard tissue biopsy punch (3-mm diameter) was used to produce a superficial cut on the infiltrated mucosa (Supporting Information Fig. S1B). The delineated superficial mucosa (Supporting Information Fig. S1C) was then gently removed using a fine blade (Supporting Information Fig. S1D) and placed in standard epithelial medium supplemented with penicillin and streptomycin (1%) and amphotericin B (2.5 µg/ml). The underlying excess submucosal tissue was dissected away with fine scissors under a dissecting microscope. The resulting epithelium was peeled away from the adjacent stroma and was further processed for epithelial culture as summarized in Supporting Information Figure S2 using techniques similar to those which have been used in ex vivo limbal biopsy culture [12, 27, 28].

Oral Mucosa Epithelial Culture

Several techniques were set up for the culture of OME, either as single-cell suspension or as explants. For single-cell suspension technique on murine feeder layers, OME was digested with 0.05% trypsin, resuspended in epithelial medium, and then plated onto six-well plates at a density of 5 × 104 cells per square centimeter which had been inoculated the day before with inactivated J2–3T3 cells (at a density of 2.4 × 104 cells per square centimeter) [29, 30]. The cells were then fed on alternate days with standard epithelial progenitor medium (low glucose Dulbecco's modified Eagle's medium 75% and Ham's F-12 medium 25%) supplemented with FCS (10%), penicillin-streptomycin (1%), hydrocortisone (0.4 µg/ml), insulin (5 µg/ml), triiodothyronine (1.4 ng/ml), adenine (24 µg/ml), cholera toxin (8.4 ng/ml), and epidermal growth factor (10 ng/ml).

We also used single suspension cultures of OME on intact human amniotic membrane (HAM) which has been shown to support LSC ex vivo expansion and maintain SC properties [12, 31, 32]. The HAM (obtained from NHSBT Tissue Services, London & Liverpool, U.K.) was prepared by carefully stretching it flat, epithelial side up on glass coverslips as previously described [12, 27], in the absence of 3T3 feeder cells.

These 3T3 and HAM coculture experiments were repeated with an explant culture technique, where the oral mucosal epithelium was not digested to single cells, but placed directly on the preplated irradiated J2–3T3 or HAM (Supporting Information Fig. S2B). Five healthy individuals donated four biopsies each, allowing every culture (explant or suspension culture with 3T3 fibroblast or HAM coculture) to be carried out five times.

Characterization of ex vivo expanded cultures was carried out using histology, immunohistochemistry, electron microscopy, colony-forming efficiency (CFEs) assays, and real-time RT-PCR for putative stem cell markers (Table 1), proliferation, and differentiation markers, exactly as described in our recent publications [12, 33] (for more detail refer to Supporting Information Annex 1).

GMP Validation and Clinical Transplantation

Oral mucosa biopsies were obtained under strictly sterile conditions of the operating room and the stroma was partially dissected. The explant was placed in epithelial medium and taken in sealed containers to the GMP laboratory for further processing. The explant was placed on the surface of the previously prepared HAM and pressed gently into place at the center of the HAM and then fed on the second day and then every second day until the culture outgrowth had covered the entire HAM. Three separate cultures from three volunteers were performed for validation purposes.

Clinical Studies

The patients underwent full informed consent. Following full clinical history and ophthalmic examination, the subjective score of ocular pain (0–10) and visual impairment (0–10) as previously defined was made [12]. Oral biopsy and GMP ex vivo culture technique were carried out as described above followed by transplantation of the cultured autologous OME to the worst affected eye. Postoperatively, the patients were treated with a combination of preservative-free 0.5% chloramphenicol antibiotic eye drops (Minims, Bausch & Lomb U.K. Limited), 1% prednisolone acetate steroid eye drops (Moorfields Eye Hospital, London, UK), and autologous serum (produced from the patient's own blood by the supervising physician or long-term by the National Blood Service). The patients were reviewed at day 1, week 1, week 2, and months 1, 2, and 3 and then at least 3 monthly thereafter. Full clinical examination and clinical photography were carried out by two observers at each visit.

DNA Fingerprinting Analysis

Genomic DNA was extracted from blood and ex vivo expanded OME and subjected to amplification with 10 independent DNA markers and the sex chromosomes (Amelogenin) using the Ampf/STR kit (Applera, Warrington, U.K.). Genescan analysis on the amplified fragments was performed from both samples using the following microsatellite markers: D3S1358, vWA, D16S539, D2S1338, Amelogenin, D8S1179, D21S11, D18S51, D19S433, TH01, and FGA-then analyzed on an ABI 377 sequence detector using Genescan version 3.1.2 and Genotyper version 2.5.2 software (Applied Biosystems, Foster City, CA).


Characterization of Normal Human Oral Mucosa

Hematoxylin and eosin staining demonstrates that native oral mucosa is significantly thicker than the corneal epithelium (Supporting Information Fig. S3A/panel, S4). The basal cells are much smaller than cells closer to the surface in keeping with a stem cell phenotype [34] (Supporting Information Fig. S3A/panel b). Five consecutive fields on high power were examined and showed that at the thinnest areas there were a mean number of 16.4 cells (±1.2; range 12–19) and at the thickest 29.2 cells (±2.1; range 24–36).

Immunohistochemical analysis showed that the putative epithelial and LSC marker p63 are predominantly expressed in the basal layer of the OME (Supporting Information Fig. S3B/panels a, b), particularly strongly in the tips of the papillae of the rete ridges; however, some expression is also observed in the suprabasal layer. CK3 is seen expressed in all but the basal layers suggesting CK3 may also be a marker of oral epithelial differentiation as well as corneal differentiation (Supporting Information Fig. S3B/panel c). Finally, CK19, a marker of conjunctival and limbal epithelial cells, was not expressed in the oral epithelium (Supporting Information Fig. S3B/panel d). Transmission electron microscopy (TEM) confirmed that intact oral epithelium was markedly thicker with many cell layers compared to corneal epithelium (Supporting Information Fig. S3C).

Ex Vivo Expansion of OME Using Single-Cell Suspension and Explant Culture Techniques

We were able to culture single-cell suspension of OME following the standard Rheinwald & Green method (Fig. 1A). The ability to produce colonies was reproducible with five donor biopsies from different normal volunteers with mean age 30.4 ± 7.6 (range: 23–42). Cell colonies were first visible around 5.2 ± 1.30 days (range: 4–7) and reaching confluence at 13.2 ± 0.84 days (range: 12–14). The OME-derived colonies showed a tightly packed small cell morphology with high nucleus to cytoplasm ratios in keeping with a stem cells or early transient-amplifying cell phenotype [35]. These colonies could be subcultured up to five times (range 4–6, n = 3); however, this resulted in appearance of cells with a more mature oral mucosal epithelial phenotype typified by a much larger squamous cell with large amounts of cytoplasm and small nuclear to cytoplasmic ratio (Fig. 1B).

Figure 1.

Microphotographs of oral mucosa cells expanded in culture using single cell or explant technique. (A): Microscope photographs of a single oral epithelial cell colony with 3T3 fibroblast coculture. The colony is formed of uniform tightly packed, small, round cells with large nuclei which gradually push out the surrounding 3T3s forming a smooth perimeter (indicated by the red arrows). All scale bars 500 µm (except high mag. image, scale bar =200 µm). (B): Tissue microscope photograph of oral epithelial cells allowed to grow beyond confluence. Beyond the stage of confluence, culture conditions are not ideal, some cells are beginning to differentiate (shown by dashed red border) into a mature phenotype with large amounts of cytoplasm. Scale bar = 100 µm. (C): Oral epithelial suspension culture on human amniotic membrane (no 3T3). Scale bar = 500 µm. Small colonies are seen later than with 3T3 coculture and demonstrate slower expansion, less defined, and more irregular borders. Abbreviations: C, colony; E, amniotic epithelium.

We aimed to replace murine 3T3 feeders with HAM and for this reason the culture of single-cell suspensions of OME was attempted on intact HAM. Small colonies were observed later at day 7.6 ± 1.14, (range: 7–10, n = 5) and were rather sparse and poorly defined compared to those produced with 3T3 coculture (Fig. 1C). In addition, the colonies began to break down and float off the surface of the HAM by 4 weeks prior to a confluent sheet being obtained. This corroborates other reported findings in the field which suggest that generation of stratified epithelia from OME is possible only in presence of murine 3T3 feeder layers [18-26]. To overcome this technical difficulty, explant techniques were attempted using biopsies from the same donors as above. All biopsies produced outgrowths of cells using standard 3T3-J2 coculture in a reproducible manner (Fig. 2A). Small cell colonies were usually first seen on average on day 5.4 (±0.51, range 4–7, n = 5); however, it took longer for the explants biopsies to reach the edge of the culture well of a standard six-well plate (average of 25.4 days ± 0.87, range 23–28, n = 5). To avoid application of feeder cells, we then attempted the explant technique culture on HAM which resulted in excellent outgrowths in all cases (Fig. 2B). They were first observed at day 5.2 ± 0.84 (range: 4–6, n = 5) and covered the whole HAM by day 18.4 ± 2.70 (range: 14–21, n = 5).

Figure 2.

Morphological appearance and characterization of the explant co culture on feeder layers (3T3). (A): Microscopic appearance of an oral explant with 3T3 coculture system. The culture is shown at days 5 and 7 at both low power (scale bar = 500 µm) and high power (scale bar = 200 µm). An oral explant (E) is shown with its corresponding epithelial outgrowth (O). The edge of the outgrowth is clearly seen due to the compression of the surrounding 3T3s (shown by the red arrows). The outgrowths are seen to contain uniform round cells which are tightly packed and contain very little cytoplasm. (B): Microscopic appearance of an oral explant with human amniotic membrane (HAM) coculture system. The culture is shown at day 7 at (a) low power (scale bar = 1,000 µm), (b) medium power (scale bar = 200 µm), and (c) high power (scale bar = 100 µm). (a) An oral explant (E) is shown with its corresponding epithelial outgrowth (O). Closer microscopic analysis of the culture shows that the outgrowth is made up of regularly arranged small tightly packed round cells with little cytoplasm. (C): Epithelial sheet histology demonstrated by hematoxylin and eosin staining of 10-day cultures from the same size tissue biopsy of the same patient grown on intact HAM in the absence of 3T3. The left panel (explant culture) shows a well formed ordered stratified epithelium with appropriate polarity. In contrast, the right panel (single-cell suspension) shows an irregular epithelium with poorly attached cells and lack of clear cell polarity and stratification. (D): Expression of putative epithelial SC markers and cytokeratin differentiation markers measured by RT-PCR analysis following a 10-day culture period. Limbal = Standard Rheinwald & Green culture of limbal stem cells. Oral 1 = Oral epithelial cells as single-cell suspension on 3T3. Oral 2 = Oral epithelial cells as explant on 3T3. Oral 3 = Oral epithelial cells as single-cell suspension on HAM. Oral 4 = Oral epithelial cells as explant on HAM. This is a representative example of at least three independent biological replicates.

Air lifting technique was used to produce epithelial sheets; however, we were unable to produce good quality epithelial sheets using the single-cell suspension technique on HAM in the absence of mouse 3T3 fibroblasts. In fact, all other previous publications using this technique for clinical application have used 3T3 coculture [21, 24, 25, 36]. In contrast, good sheet formation was however observed with our explant/HAM coculture technique. The epithelial sheets formed with this culture method showed well-stratified epithelial layers (three to seven layers) where cells were firmly attached to each other. In addition, those sheets were physically durable and could be easily handled (Fig. 2C).

To prove the existence of limbal-like stem and progenitor cells within our oral mucosa cultures, we tested the expression of putative epithelial stem cell markers, namely p63 [37, 38], ABCG2 [39, 40], C/EBPδ [41], and specific differentiation marker CK3 [15, 42]. In addition, we also assessed the expression of unique corneal epithelial markers, namely CK12 and PAX6 [43]. These two markers are expected to be present in corneal epithelium only; hence their presence in our oral mucosa cocultures would suggest a marked change in differentiation of these cells toward a limbal phenotype. Four different oral mucosa cocultures (using all methods delineated in Supporting Information Fig. S2) were tested by RT-PCR for the expression of above markers. This analysis indicated that: (a) cultured OME cells express similar markers despite the method that was used for their cultivation; (b) cultured OME cells expressed several putative epithelial stem cells markers including: ΔNp63α, ABCG2, and C/EBPδ; (c) cultured OME cells express the corneal and oral mucosa differentiation marker CK3, but lack the expression of CK12 and PAX6 (Fig. 2D). These results suggest that cultured oral epithelial cells retain a gene expression profile that can be attributed to epithelial stem cells in general, but have not yet acquired a typical limbal expression pattern after 10 days in culture.

Since the explant culture on intact HAM and presence of FCS in the absence of feeder layers produced excellent colonies and was the most promising for clinical implementation studies, we studied the expression and distribution of these putative epithelial and limbal markers using immunohistochemistry (Supporting Information Fig. S5). Morphological analysis of the cultured cells with hematoxylin and eosin staining revealed that an intact stratified epithelial sheet (five to seven layers thick) firmly adherent to the underlying HAM was produced. The putative epithelial stem cell marker, p63, was expressed in the basal and suprabasal layer in keeping with their expected location as progenitor cells. Although stem cells in vivo are only slowly cycling, they can demonstrate high proliferative capacity in certain conditions such as during ex vivo culture conditions which is shown by the expression of Ki67. We also observed a mild diffuse expression of CK3 in all epithelial layers (Supporting Information Fig. S5) as with the parent oral tissue (Supporting Information Fig. S3B). In conclusion, ex vivo expansion of OME using explant culture on HAM fed with FCS supplemented epithelial medium produced a stratified epithelial sheet in the absence of a 3T3 feeder layer, which has not been possible with most other culture systems reported. Most importantly, this epithelial sheet differs significantly from the native parent tissue (OME) in several ways, namely being thinner (5–7 layers compared to approximately 20), avascular, firmly attached to the underlying basement membrane with loss of rete ridges.

Animal and Feeder-Free Ex Vivo Expansion of OME Explants on Intact HAM

To test whether we could replace FCS with human autologous serum (HAS), we set up duplicate cultures of single-cell suspensions of OME on mitotically inactivated J2–3T3 fibroblasts. One culture was fed with FCS supplemented epithelial medium and the other one with HAS supplemented epithelial medium (n = 3). Successful culture of oral epithelium was achieved with both FCS and HAS supplemented media (Fig. 3A) in all cases. Both cultures behaved remarkably similarly with small colonies appearing on days 4–6 in each case, gradually enlarging to form colonies with holoclone-like appearance. CFE assays (Fig. 3B) and qRT-PCR (Fig. 3C) were used to compare the two culture conditions. This showed that the CFE is higher in fresh limbal compared to fresh OME culture. There were no significant changes in CFE potential between HAS cultivated OME cells compared with FCS cultivated cells (Fig. 3B/panel c). RT-PCR analysis also showed significantly higher expression of all three markers in cultures supplemented with HAS when compared with FCS supplemented media (Fig. 3C).

Figure 3.

Comparison of oral mucosa explants culture on FCS and HAS containing media. (A): The appearance of colonies formed using a Rheinwald and Green system of single-cell culture on J2 3T3 using FCS versus HAS-supplemented epithelial medium. The appearance of colonies on days 5 and 8 is virtually identical using either serum type. Scale bar = 200 µm. (B): CFE assay to compare the stem cell properties of single-cell suspension cultures of buccal oral epithelium produced with autologous serum versus FCS. (a) The appearance of oral SC colonies seen with phase-contrast microscopy. (b) The appearance of fixed colonies stained with Rhodamine B at day 12 of assay. (c) A chart to show the CFE of oral SC culture cultured with HAS versus FCS. The CFE of fresh cadaveric limbal tissue is shown for comparison. There is no significant difference between HAS and FCS (p = .5, n = 3). (C): Quantitative real-time RT-PCR analysis of comparison of putative epithelial stem cell marker expression in HAS versus FCS-supplemented media. Bar chart showing the expression of ΔNp63α, ABCG2, and C/EBPδ in oral epithelial cells grown in epithelial medium supplemented with HAS compared to FCS. In all cases, the data are standardized to an external limbal stem cell control known to express the markers in question. The statistical difference between HAS and FCS cultures is shown in each case. This is a representative example of at least three independent biological replicates. (D): Outgrowth rate of ex vivo cultured oral epithelium using an explant technique cultured with growth medium containing human autologous serum versus FCS. The graph shows similar growth rates even when the FBS was completely replaced by the HAS (p = .73, ANOVA). (E): Histology and immunohistochemistry of ex vivo expanded oral epithelium cultured from an explant on HAM with either FCS or HAS-containing growth medium. The epithelium was grown from similar biopsies from the same individual and cultured under identical conditions. The culture was stopped at day 14 in each case and the epithelium placed in formalin from both hematoxylin and eosin staining and immunohistochemistry. The appearance of the FCS cultures is shown in the left side panels whereas the HAS cultures are shown in the right side panels. Scale bar = 100 µm. Abbreviations: CFE, colony-forming efficiency; FCS, fetal calf serum; HAS, human autologous serum.

Finally, the feasibility of ex vivo culture of oral epithelium using an explant on HAM in the absence of 3T3s with HAS instead of FCS was measured. The culture systems were set up as described above using identical sized biopsies and placed on HAM then fed with either FCS or HAS and the culture continued for 21 days. The outgrowths produced were measured by marking the underside of the culture vessel on alternate days at the time of each feed. When using explant on HAM cultures, excellent outgrowths were obtained with HAS as well as with FCS. A similar growth rate was seen using both HAS and FCS in the growth medium (Fig. 3D) showing no significant difference (p = .73).

Hematoxylin and eosin staining and immunohistochemistry are virtually identical with FCS and HAS (Fig. 3E). Both produced a three to seven layer well-formed epithelium that was well attached to the underlying HAM. The basal layer consisted of regular, tightly packed small cuboidal cells with prominent nuclei that strongly express the putative epithelial stem cell marker p63 and the nuclear proliferative market Ki67 as well as the cytoplasmic putative stem cell marker ABCG2. In addition, both epithelia expressed the cytokeratin CK3.

Validation of Animal and Cell-Free OME Culture System Under GMP Conditions

The studies presented above show that our animal product- and feeder-free explant culture system can produce multilayered epithelial sheets which express putative epithelial stem cell markers. Their phenotypic appearance was similar to corneal epithelium although the expression profile (at least after ex vivo expansion) remained “oral” rather than “corneal.”

Prior to clinical application, we needed to ensure that the process was reproducible using fully GMP compliant culture processed in the HTA compliant GMP stem cell laboratory at the Royal Victoria Infirmary (Newcastle upon Tyne Hospitals NHS Foundation Trust, U.K.). Three cultures of three volunteers were undertaken under GMP conditions. Excellent outgrowths were observed in each case and the rates of outgrowth were comparable to those produced in our previous laboratory based studies. Interestingly, the cultures grown under GMP conditions were seen to have a faster growth rate than cultures grown under standard laboratory conditions (p = .046, ANOVA, n = 3; Supporting Information Fig. S6). This is probably mainly due to the fact that the culture manipulations during feeding were carried out in a GMP laboratory with strict monitoring of environmental conditions which may have produced an environment more conducive to cell growth. The faster growth rate under GMP conditions together with higher expression of putative stem cell markers obtained from HAS expanded culture makes this combination clinically desirable because of the higher cell quality produced and shorter timing.

Patient Selection

For our clinical translation studies we selected two patients with total bilateral LSCD caused by chemical burns. Patient 1 was a 76-year-old male with longstanding total bilateral LSCD resulting from a severe bilateral caustic soda injury. Patient 1 displayed all the hallmark signs of bilateral total LSCD (Fig. 4A) including: (a) significant epithelial defect as assessed with 0.25% sodium fluorescein staining; (b) significant peripheral and central corneal vascularization; (c) marked corneal opacity (Right eye-Grade 4 and Left eye-Grade 3) [12], and (d) total loss of Palisades of Vogt. The best corrected visual acuity was markedly reduced bilaterally at hand movements in the right and 6/12 in the left eye (with a combination of soft extended wear bandage and rigid gas permeable contact lenses). The remaining ophthalmic examination was normal in both eyes including normal intraocular pressures, normal Schirmer's test scores (normal tear production) and good lid function ensuring an appropriate environment for stem cell transplantation.

Figure 4.

Pre and post-operative clinical appearance for patient 1. (A): Clinical appearance and fluorescein staining of both eyes of patient 1 with total LSCD. (a) Right and (b) left eye clinical photographs showing total conjunctivalization of each eye with 360° vascularization, total loss of palisades of Vogt, and marked corneal opacity and scarring. The corresponding photographs following staining with 2% fluorescein (c and d) indicate significant delayed staining on the corneal surface which should be completely absent with a normal corneal epithelium. (B): Histological and immunohistochemistry analysis of the ex vivo oral epithelium of Patient 1. The hematoxylin and eosin stained tissue specimen shows the early formation of a stratified epithelium while the PAS-stained specimen shows an absence of mucin secreting goblet cells. Immunohistochemistry shows the expression of p63 and Ki67 of the nuclei of the basal epithelial sheet and ABCG2 and CK3 expression in the cytoplasm. Scale bar = 100 µm. (C): Transmission electron microscopy (TEM) of cultured oral epithelium from patient 1. (a) TEM of cultured oral epithelium of patient 1 showing a primitive stratified epithelium with a clear basal layer of cuboidal cells. (b) Higher magnification shows the basal layers to consist of large nuclei and little cytoplasm in keeping with a stem cell phenotype. Higher magnification reveals the presence of (c) hemidesmosomes attaching cells to basement membrane, (d) desmosomes attaching cells to each other, and (c) surface microplicae all indicated by the arrows. (D): Pre- and 18/12-month postoperative clinical appearance and fluorescein staining pattern of the right eye of patient 1. Upper panels show the clinical appearance of the patient's right eye preoperatively and postoperatively.

Patient 2 was a 44-year-old male with longstanding total LSCD in the right eye and subtotal in the left eye (therefore unsuitable for limbal biopsy from this eye), resulting from a severe bilateral ammonia eye injury, worse in the right eye. The best corrected visual acuity was markedly reduced in the right eye at 3/60 and 6/6 in the left eye. At the time of autologous oral mucosa stem cell transplantation, the patient had all the clinical hallmark signs of total LSCD in the right eye confirmed by impression cytology.

In view of the above, we decided to perform autologous OME cell transplantation to the patients' worse eye. Transplantation of ex vivo expanded OME on the previously prepared corneal stromal bed of the diseased right eye of both patients was carried out using a similar technique as described in our previous publication [12]. The main stages of the surgical procedure were photographed and are shown in Supporting Information Figure S7. As the first part of transplantation procedure involved removal of all the abnormal corneal epithelium, we used part of this tissue to confirm the diagnosis of total LSCD which indicated the complete replacement of the normal avascular five to seven cell layered corneal epithelium by a hypercellular highly vascularized epithelium suggestive of conjunctivalization (Supporting Information Fig. S8). In addition, immunohistochemistry shows that there is a complete absence of CK3 expression in the epithelium and strong expression of CK7, CK19, and MUC5AC confirming complete replacement of the corneal epithelium phenotype by conjunctival phenotype on the corneal stromal surface (Supporting Information Fig. S8). We also analyzed the excess cultured OME by histology and TEM to provide an internal control allowing an accurate analysis of the epithelium actually transplanted. As with our previous laboratory findings, the ex vivo expansion of OME cells produced an epithelium which looked remarkably like that produced from ex vivo expansion of limbal epithelium and very different from parent oral mucosa (Fig. 4B). Histological analysis of the cultured epithelium shows a three to five layer thick epithelium with a basal layer of small cuboidal cells in keeping with a stem cell phenotype. There is early stratification with layers of flattened superficial cells. Periodic acid–Schiff (PAS) staining shows the absence of any PAS-positive mucin-secreting cells as found in conjunctival epithelium. Some workers have advocated air lifting the epithelium during culture to produce a stratified epithelium [44]; however, our studies showed that we could achieve stratification in the absence of air lifting.

Immunohistochemical analysis confirmed that the transplanted sheet contains basal cells which express the putative stem cell markers p63 and ABCG2 and that many of the cells in this layer continue to actively divide shown by the expression of Ki67. The cells of the whole epithelium also express CK3 as shown above (Fig. 4B). TEM also demonstrated that within the cultured OME, cells adhered to each other (through desmosomes), to the basement membrane (through hemidesmosomes), and also to appropriate surface specializations allowing interaction with the tear film (microplicae, Fig. 4C), very similar to LSC. DNA fingerprinting analysis also confirmed an identical DNA profile between the blood sample taken from the patient and cultivated OME, fully confirming that ex vivo transplanted cells were derived entirely from the biopsied explants and not the HAM (Supporting Information Fig. S9).

Clinical Translation Results

Patient 1

Within the first 2 months, fine vessels were seen encroaching into the peripheral cornea. However, those did not progress beyond the first 2 months (Fig. 4D, upper panels). The surface showed no epithelial defect even as the superficial HAM melted at 12 weeks. This would indicate that the transplanted cells were successful in producing a stable epithelium that contained junctional complexes which prevented the free influx of topical fluorescein (Fig. 4D, lower panels). Subjectively, the patient felt there had been a significant and obvious reduction in vision impairment score from preoperative 9/10 to 7/10 at 6 months. The patient's pain score reduced from 3/10 to 0/10 within 1 week of surgery. The patient was last reviewed 21 and 9 months after the original oral mucosa stem cell and corneal transplantation, respectively. At that stage, his best corrected visual acuity was 6/36 in the right eye (improving to 6/24 with pinhole), and 6/18 in the left eye, suggesting a significant visual improvement in the operated right eye.

Patient 2

Within the first 2–3 months, fine vessels were seen encroaching into the peripheral cornea; however, these did not progress beyond the first 2–3 months. Similarly to patient number 1, these were quite unlike the large active vessels seen in the previously conjunctivalized surface. However, at 5 months postoperatively the patient developed a central corneal epithelial defect that was treated with a contact lens combined with intensive topical autologous serum treatment that gradually resolved within 10 weeks. Subjectively, the patient felt there had been a significant and obvious reduction in pain score from preoperatively 8/10 to 5/10 at 6 months post-op. The patient was last reviewed 41 and 21 months after the original oral mucosa stem cell and corneal transplantation, respectively. At that stage, his best corrected visual acuity was 6/60 in the right eye and 6/6 in the left eye. The pain score reduced to 4/10; however, the vision impairment score remained the same at 8/10 due to central subepithelial haze in the grafted cornea secondary to a few initial episodes of recurrent epithelial erosion.

Both patients underwent HLA matched corneal grafting approximately at 13 and 20 months after the OME transplant, respectively. Analysis of corneal button taken during keratoplasty indicated the presence of stable epithelium with superficial expression of CK3, basal expression of p63 and Ki67, and complete absence of conjunctival marker MUC5A (Supporting Information Fig. S10). Together these data suggest that transplantation of autologous ex vivo expanded OME can contribute to restoration of corneal surface of patients with total bilateral LSCD.


This mainstay treatment of bilateral total LSCD still requires allogeneic transplantation of large whole limbal tissue grafts from living or cadaveric donors [6, 7]. This leads to the requirement of long-term systemic immunosuppression with potential serious side effects as well as the risk of creating iatrogenic LSCD in living donors. As a way of circumventing these problems, oral mucosal epithelium has shown great promise as an alternative source of autologous epithelium that can be potentially used in severe bilateral ocular surface disease [21, 24, 26, 44, 45]. In keeping with previous literature, our study demonstrates that oral mucosal epithelium contains progenitor cells that display all the features inherent in other epithelial stem cells [23]. The main issue with published clinical studies is lack of homogeneity in patient selection and clinical assessment as well as the requirement for murine 3T3 feeder layers and/or FCS which increases the risk of pathogen transmissions and also precludes long-term application of these protocols for GMP compliant clinical studies. For this reason, we set out to devise a GMP compliant culture system which requires no feeder cells or animal-derived products in the culture media, by replacing 3T3 feeder cells with HAM and FCS with HAS. Although HAM and HAS have been used in other OME expansion studies, this has always been together with 3T3 cells [18-25] and for this reason we believe that this is the first report of generation of multilayered OME-derived epithelium under feeder- and animal product-free conditions that was able to restore ocular surface of bilateral LSCD patients.

We have demonstrated that animal product- and feeder-free cultures can be consistently produced by plating OME explants on intact HAM, thereby removing the risks associated with transplantation of animal-derived product and tissue while retaining the advantages of HAM as a biological substrate. On continued culture, these sheets look morphologically like corneal epithelium rather than parent OME making them suitable for transplantation onto the surface of the eye, although their gene expression remains oral rather than corneal. In addition, analysis of removed corneal epithelium as part of the keratoplasty procedure shows a weak CK3 staining, suggesting that transplanted OME cells may have not fully differentiated to corneal epithelium. We hypothesize that the change in environment from the oral cavity to HAM as a substrate induces this change in morphology and prevents the formation of multiple layered, highly vascular epithelia seen in the parent tissue. Our analysis demonstrate that these sheets contain cells which express putative epithelial stem cell marker including ΔNp63α, ABCG2, and C/EBPδ and morphologically appear like corneal epithelium with typically a five to seven cell layer thick epithelium on a basement membrane. Replacing the FCS with HAS resulted in production of epithelial sheets which were morphologically similar and moreover displayed the same gene expression profile. It was interesting to notice that expression of putative stem cells markers was actually higher in OME cultured in presence of HAS when compared with FCS, which could be most likely due to presence of growth factors in HAS that may promote maintenance of stem cell phenotype. This however is a speculative observation and needs to be further investigated by performing comparative expression analysis between HAS and FCS in a larger number of patients.

Most importantly these ex vivo expanded OME sheets using an explant techniques on HAM with HAS supplemented culture media can be generated under GMP conditions in a reproducible manner. Our experience with two patients has shown that the ex vivo expanded OME tissue can successfully restore the integrity of the ocular surface of eyes with total bilateral LSCD (at least in the short term: 24 months) and act as a suitable substitute for allogeneic corneal epithelial stem cell grafts eliminating the need for long-term systemic immunosuppression. To the best of our knowledge to date this is the first report of successful expansion of animal product- and feeder-free system for the expansion of OME for the clinical treatment of total bilateral LSCD. We are aware that our study only includes two patients and definitive conclusions on the safety and efficiency of this technique can only be drawn after performing this procedure on a larger number of patients. Indeed, this work is ongoing in our group.


In conclusion, the main problems associated with traditional treatments of bilateral LSCD treatments include long-term immunosuppression with whole tissue allografts and the use of non-human animal products including 3T3 fibroblasts and FCS when using standard ex vivo expansion protocols. These problems can be overcome with the culture and treatment method described in this study. This technique has two major advantages over the other available treatment modalities. First, it allows the exclusion of all animal products and feeder cells and the safety profile of the end product is further enhanced by the introduction of a GMP compliant system. Second, it allows the use of only autologous tissue, thereby precluding the need for systemic immunosuppression with its associated side effects. By tackling these challenges associated with the current treatments available for the total bilateral LSCD, this novel study represents a significant advance in ocular tissue regeneration and transplantation.

Table 1. Sequences of oligonucleotides used for the RT-PCR analysis
GenePrimerSequenceAnnealing temperature (°C)


We would like to thank all those in the Department of Ophthalmology at the Royal Victoria Infirmary (Newcastle upon Tyne NHS Hospitals Foundation Trust) who enabled the progression of this work to the clinical stage. In addition, we are very grateful to Anne Dickinson, Dr. Andy Gennery, Janice Dunn, and Ken Brigham for help in the good manufacturing practice laboratory. Without their help, the transplants would not have been possible. We would also like to thank the Newcastle Healthcare Charity, Life Knowledge Park, One North East Developmental Agency, U.K. NIHR Biomedical Research Centre for Ageing and Age-related disease and the Newcastle upon Tyne NHS Hospitals Trust for their financial support for this research and the United Kingdom Eye Banks for providing access to human limbal tissue donated for research. Finally, we would like to thank the donors of limbal and oral mucosa tissue for donating their tissue for research. Due to their generosity, patients with limbal stem cell deficiency will be able to have their sight restored and lead much more comfortable lives. S.K. is currently affiliated with Department of Ophthalmology, Queen Elizabeth Hospital, University Hospitals Birmingham NHS Foundation Trust, Birmingham, U.K. S.A. is currently affiliated with Department of Eye and Vision Science, Institute of Ageing and Chronic Disease, University of Liverpool and St. Paul's Eye Unit, Royal Liverpool University Hospital, Liverpool, U.K.

Author Contributions

S.K.: involved in conception and design of the study, collection and assembly of data, data analysis and interpretation, manuscript writing, and clinical work; S.A.: involved in conception and design of the study and manuscript writing; H.S.M. and A.M.: responsible for the histopathological and transmission electron microscopy interpretation; F.C.F.: involved in conception and design of the study, fund raising, collection and assembly of data, data analysis and interpretation, manuscript writing, and clinical work; M.L.: involved in conception and design of the study, fund raising, data analysis and interpretation, and manuscript writing.

Disclosure of Potential Conflicts of Interest

The authors declare no conflict of interest.