SEARCH

SEARCH BY CITATION

Keywords:

  • RHAMM/HMMR;
  • Aldefluor;
  • ALDH1;
  • Limbal stem cells;
  • p63;
  • Cytokeratin 3/12;
  • Epithelial lineage;
  • SP

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

The corneal epithelium is maintained by stem cells located at the periphery of the cornea in a region known as the limbus. Depletion of limbal stem cells (LSCs) results in limbal stem cell deficiency. Treatments for this disease are based on limbal replacement or transplantation of ex vivo expanded LSCs. It is, therefore, crucial to identify cell surface markers for LSCs that can be used for their enrichment and characterization. Aldehyde dehydrogenases (ALDHs) are enzymes which protect cells from the toxic effects of peroxidic aldehydes. In this manuscript, we show for the first time that ALDH1 is absent from the basal cells of the limbal and corneal epithelium. We separated limbal epithelial cells on the basis of ALDH activity and showed that ALDHdim cells expressed significantly higher levels of ΔNp63 and ABCG2 as well as having a greater colony forming efficiency (CFE) when compared to ALDHbright cells. Large scale transcriptional analysis of these two populations led to identification of a new cell surface marker, RHAMM/HMMR, which is located in all layers of corneal epithelium and in the suprabasal layers of the limbal epithelium but is completely absent from the basal layer of the limbus. Our studies indicate that absence of RHAMM/HMMR expression is correlated with properties associated with LSCs. RHAMM/HMMR- limbal epithelial cells are smaller in size, express negligible CK3, have higher levels of ΔNp63 and have a higher CFE compared to RHAMM/HMMR+ cells. Taken together these results suggest a putative role for RHAMM/ HMMR as a negative marker of stem cell containing limbal epithelial cells. Cell selection based on Hoechst exclusion and lack of cell surface RHAMM/HMMR expression resulted in increased colony forming efficiency compared to negative selection using RHAMM/HMMR alone or positive selection using Hoechst on its own. Combination of these two cell selection methods presents a novel method for LSC enrichment and characterization.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

The cornea is the clear front of the eye, and it is covered by a stratified epithelium. This corneal epithelium is devoid of its own stem cells, which are located at the periphery of the cornea in a region known as the limbus [1]. Indeed radio-labeling studies have shown that the limbal stem cells (LSCs) maintain and renew the corneal epithelium [1]. Identification of LSCs is currently hindered due to a lack of LSC specific surface markers which could be used to isolate viable LSCs. The transcription factor p63 is known to be present in LSCs [2] and the cytokeratin (CK) 3/12 is known to be absent in LSCs [3, 4], but isolation strategies which require cell permeabilization cannot isolate viable LSCs on the basis of the presence or absence of these two markers. For this reason, cell surface markers for LSCs need to be used to isolate viable cells, and several possible candidate markers include the integrins [5], epidermal growth factor receptor [6] and ATP binding cassette G2 (ABCG2) [7, 8].

Aldehyde dehydrogenases (ALDHs) are a superfamily of 17 intracellular enzymes which protect cells from the cytotoxic effects of peroxic aldehydes [9]. The human cornea has been shown to contain high levels of ALDH3A1 and ALDH1A1, comprising 5% and 3% of soluble corneal proteins respectively [10]. Immunohistochemistry (IHC) on sections of mouse cornea have shown that ALDH3A1 expression is localized to the corneal epithelium, with highest levels at the central cornea and decreasing levels towards the corneal periphery [11]. The corneal epithelium is exposed to very high levels of ultraviolet light. ALDH3A1 has been shown to have a protective role against ultraviolet light through two mechanisms, the direct absorption of ultraviolet light and the detoxification of cytotoxic aldehydes [12, 13]. Although not as predominant in the human cornea as ALDH3A1, a major role for ALDH1A1 is the oxidation of malondialdehyde, a cytotoxic aldehyde which is not a substrate for ALDH3AI.

Hematopoietic progenitors display high levels of ALDH activity, and this property has been used by our group and others to purify viable and engraftable hematopoietic progenitors from both bone marrow and cord blood of humans and mice [14, [15]16] as well as neural stem cells [17, 18]. This method uses a visible light excitable fluorophore (DIPYrromethene BOron Difluoride conjugated aminoacetaldehyde, Aldefluor [StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com]) which is metabolized by ALDH to a carboxylate ion retained within the cell, allowing cells with high levels of ALDH activity (ALDHbright cells) to be isolated by fluorescence activated cell sorting (FACS) because of their high fluorescence [14]. This reagent, more commonly known as Aldefluor, therefore allows the identification and subsequent isolation of viable cells. The purpose of this study was to investigate the use of Aldefluor for purification or enrichment of LSCs which would enable the identification of further cell surface markers using large scale transcriptional analysis.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Human Donor Tissue

Cadaveric adult human limbal tissue from 20 donors (age range 25–98 years old, mean 70.25 years) was obtained from the UK Transplant Service. Informed consent for the use of this human tissue for experimental research was obtained, in accordance with the Declaration of Helsinki.

Immunohistochemistry on Sections of Human Cornea and Limbus

IHC for ALDH3A1 is well-characterized [11] and this was not repeated. However, IHC for ALDH1 was performed. Human cadaveric limbal tissue was fixed in 4% paraformaldehyde (Sigma Aldrich, Gillingham, U.K., www.sigma-aldrich.com) at 4°C overnight and subsequently processed and embedded in paraffin wax. Tissue sections (7 μm) were then prepared for antigen retrieval and staining. Samples were de-paraffinised, incubated with 3% hydrogen peroxide solution (Sigma Aldrich) in phosphate buffered saline (PBS) at room temperature for 30 minutes. After three washes in PBS, the sections were immersed in 10 mM tri-sodium citrate (Sigma Aldrich) in water and placed in the microwave at full power for 5 minutes. The sections were allowed to stand for a further 20 minutes in the hot tri-sodium citrate solution at room temperature. Following three washes in PBS, the sections were incubated with a 1:50 dilution of sheep serum (Sigma Aldrich) in PBS at room temperature for 30 minutes. The sections were then incubated with 1.0 μg/ml mouse anti-human ALDH1 (clone 44, BD Biosciences, San Diego, http://www.bdbiosciences.com) in PBS overnight at 4°C. For the control sections, this primary antibody was replaced with PBS. After three washes in PBS, the sections were incubated with 2.0 μg/ml biotinylated anti-mouse immunoglobulins (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com) in PBS at room temperature for 30 minutes. Following three washes in PBS, the sections were incubated with avidin/biotinylated horseradish peroxidase complex (DakoCytomation) in PBS at room temperature for 30 minutes. The sections were washed in PBS, and then incubated with diaminobenzidine solution (Sigma Aldrich) at room temperature for 45 seconds. After washing in water, the sections were counterstained with 0.5% methyl green (Sigma Aldrich), before being dehydrated in ethanol and mounted in distyrene/plasticiser/xylene (DPX) complex (Raymond A Lamb, East Sussex, U.K., http://www.ralamb.co.uk). The sections were viewed using a Carl Zeiss microscope and photographs taken using Axiovert software (Carl Zeiss, Jena, Germany, http://www.zeiss.com).

Culture of Human Limbal Epithelium

Human limbal epithelial cells were cultured on murine 3T3-J2 fibroblast feeder layers. The co-cultures were fed with epithelial medium on the third day, and then every other day thereafter. The primary limbal epithelial cultures were sub-cultured after removal of the 3T3 fibroblasts by incubation with 0.02% EDTA solution (Sigma Aldrich) for 30 seconds at room temperature. The remaining limbal epithelial cells were trypsinized and plated at 0.6 × 104 cells/cm2 in wells pre-plated with inactivated 3T3 fibroblasts. The sub-cultures (passaged cells) were maintained in a similar way to the primary cultures.

Analysis and Cell Sorting of Human Limbal Epithelial Cells Using Aldefluor

Aldefluor reagent (BD Biosciences) was prepared as per manufacturer's instructions. A 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 re-suspended in Aldefluor assay buffer (BD Biosciences), to obtain a final cell density of no more than 1 × 106 cells/ml. The Aldefluor reagent was added to this limbal epithelial cell suspension to obtain a final dilution of 1.5 μM. One/tenth of this Aldefluor + limbal epithelial cell mixture which was immediately transferred to a tube containing 10 fold molar excess ALDH inhibitor, diethylaminobenzaldehyde (DEAB). Both the Aldefluor + limbal epithelial cell mixture and the DEAB + Aldefluor + limbal epithelial cell mixture were incubated in a water bath at 37° for 45 minutes in the dark. The cell mixtures were kept on ice until further analysis. Flow cytometry was performed using the FACSCalibur flow cytometer (BD Biosciences). The cells were analyzed on the basis of side scatter and fluorescence intensity in the FL1 channel using the CellQuest software (BD Biosciences). The DEAB containing sample was analyzed first and all the cells were gated. This gate was used to highlight the cells taking up low amounts of Aldefluor (ALDHdim cells). The sample without inhibitor was then analyzed, and another gate was used to highlight those cells taking up large amounts of Aldefluor (ALDHbright cells). The numbers of cells in each gate was quantified using CellQuest software. At least 50,000 events were acquired each time. The limbal epithelial cells labeled with Aldefluor were sorted into ALDHdim and ALDHbright cell populations using a Digital Vantage (DiVa) or FACS Aria cell sorter (BD Biosciences). The sorted cells were then further analyzed to check the purity of the sorted populations.

Cytospins of Sorted Limbal Epithelial Cells

Cytospins of the ALDHdim and ALDHbright sorted cell populations were made on microscope slides (Griffiths & Nielsen, West Sussex, U.K.,http://www.g-and-n.com) using 50 μl of each cell sample using a Shandon Cytospin four cytocentrifuge (Thermo Electron Corporation, Waltham, MA, http://www.thermo.com) at 1,000 rpm for 3 minutes. These cytospins were immediately fixed by immersion in methanol (VWR, Leicestershire, U.K., http://www.vwr.com) at room temperature for 10 minutes. The slides were then immersed in PBS three times at room temperature for 3 minutes. Approximately 50 μl of a 1 μg/ml solution of propidium iodide (Sigma) in distilled water was placed on each cytospin at room temperature for 1-minute. The slides were then immersed in PBS for 1 minute and mounted in distyrene/plasticiser/xylene complex with a coverslip. The cytospins were viewed using a Carl Zeiss microscope and photographs taken using Axiovert software (Carl Zeiss).

Side Population Analysis

Side population (SP) analysis was carried out using the DNA binding dye Hoechst 33342 as described in [36]. The cell population was analyzed using FACS Aria using the DIVA software (BD Biosciences). For combined SP and RHAMM staining, the staining with Hoechst 33,342 was carried out first. The cells were kept at 4°C for the second step which comprised staining with the unconjugated primary RHAMM antibody followed by fluorescein isothiocyonate (FITC) conjugated secondary antibody.

Colony Forming Efficiency of Sorted Limbal Epithelial Cells

In order to assess the colony forming efficiency (CFE) of the cells, sorted cells (1,000–2,500) were cultured on mitotically inactivated 3T3 pre-plated wells, and maintained for 12 days as previously described [19]. All experiments were carried out in triplicates. The cultures were fixed on day 12 by incubation with a 3.7% solution of formaldehyde (Sigma Aldrich) in PBS at room temperature for 15 minutes. The fixed cultures were then stained for 15 minutes by incubation with a 0.2% solution of Rhodamine B (Sigma Aldrich) in methanol. Limbal epithelial colonies in each well were counted under a dissecting microscope, and photographs taken. CFE (%) was calculated using the following formula: number of colonies formed/number of cells inoculated × 100.

Real Time Reverse Transcription Polymerase Chain Reactions on Sorted Limbal Epithelial Cells Using Known Markers

Sorted cells from both populations were lysed in TRizol reagent (Gibco, Grand Island, NY, http://www.invitrogen.com), and RNA was extracted as per manufacturer's instructions. The concentration of isolated RNA was assessed by analyzing 1 μl of RNA using a NanoDrop (LabTech International, Sussex, U.K., http://www.labtech.com). Reverse transcription was then performed as follows. A 10 μl final solution containing 2 μg RNA, 1× DNase reaction buffer, 1 unit DNase per μg of RNA used, and sterile water (all 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 (Promega), the mixture was incubated at 65°C for 10 minutes. One μg 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, 0.5 mM deoxynucleotide phosphate (dNTP) mix, 25 units of RNasin ribonuclease inhibitor, and 200 units of reverse transcriptase (all from Promega). The resulting mixture was incubated at 37°C for 60 minutes and then 99°C for 5 minutes. Real time reverse transcription polymerase chain reactions (RT-PCRs) were then performed for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ALDH1A1 and known putative LSC markers (ΔNp63 and ABCG2) on the resulting ALDHdim and ALDHbright sorted limbal epithelial cell samples. LightCycler capillary tubes (Roche, Basel, Switzerland, http://www.roche.cpm) were placed in cooled centrifugation tubes (Roche) and each capillary tube was filled with 1 μl of the reverse transcribed mixture, 10 μl QuantiTect SYBR Green PCR Master Mix, 1 μl each of specific 10 μM forward and reverse primers (see supplemental online Table 1), and 7 μl sterile water (all from Qiagen). The filled LightCycler tubes were briefly centrifuged at the lowest setting of a micro-centrifuge (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 of 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). The gene to GAPDH ratio was calculated and the LightCycler Relative Quantification software (Roche) was used to analyze the data. The primer sequences are shown in supplemental online Table 1.

Microarray Analysis of Sorted Limbal Epithelial Cells

Due to the small quantities of RNA isolated following sorting, RNA amplification (using approximately 100 ng RNA) was performed prior to performing the microarray hybridization. This was performed using the MEGAscript T7 Kit (Ambion, Austin, TX, http://www.ambion.com) as per manufacturer's instructions. The copy RNA (cRNA) was then cleaned and biotin labeled as per manufacturer's instructions (Affymetrix, Santa Clara, CA, http://www.affymetrix.com). The biotin labeled cRNA was quantified and fragmented (Affymetrix) prior to hybridization. A test array of housekeeping controls using the fragmented cRNA samples was performed and analyzed to assess sample suitability for GeneChip arrays (Affymetrix). Following suitability, the biotin labeled fragmented cRNA samples were hybridized to the GeneChip Human Genome U133 Plus 2.0 Array (Affymetrix) and washed as per manufacturer's instructions. Microarrays were performed on ALDHdim and ALDHbright samples (each sample in duplicate). The MAS 5.0 software detection algorithm (Affymetrix) was used to determine presence (P), absence (A) or marginal expression (M) of each gene in the array. The data obtained from MAS 5.0 were then normalized and further analyzed in the GeneSpring software 6.2 (Silicongenetics, Redwood, CA, http://www.chem.agilent.com). Per chip normalization was done as follows: values below 0.01 were set to 0.01 and then each measurement was divided by the 50th percentile of all measurements in that sample. A gene was defined as significantly upregulated if the signal fold change (FC) between the target and reference sample was larger than two and the target sample was present. Alternatively, a gene was defined as significantly downregulated if the FC was less than −2 and the reference sample was present. As suggested by the manufacturer, the probe sets were excluded if the detection call for both the target and the reference were absent or if no change call (NC) was detected (change p value >0.05). Only the genes that fulfilled all the filtering criteria reproducibly in two biological replicates were considered significant. Clustering the genes that showed significant fold change to various molecular pathways was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG).

Flow Cytometry on Limbal Epithelial Cells

Following trypsinization of limbal epithelial cells from the culture well, the cell samples was centrifuged. The supernatant was removed and the cells were re-suspended in a 1:50 dilution of the rabbit anti-human receptor for RHAMM/HMMR in PBS. The sample was incubated for 30 minutes at 4°C. After this incubation, the cell sample was centrifuged and the supernatant removed. The cells were then re-suspended in 5% FCS in PBS and re-centrifuged. The supernatant was removed and the cell pellet was re-suspended in 5% FCS in PBS. Following centrifugation, the cell pellet was re-suspended in 10 μg/ml FITC conjugated goat anti-rabbit immunoglobulins in PBS. The sample was incubated for 30 minutes at 4°C. After this incubation, the cell sample was centrifuged and the supernatant removed. The cells were then re-suspended in 5% FCS in PBS and re-centrifuged. The supernatant was removed and the cell pellet was re-suspended in 5% FCS in PBS. The cells were then analyzed using a FACS Calibur flow cytometer (BD Biosciences) and the data analyzed using CellQuest software (BD Biosciences). Propidium iodide was added to the cell sample to eliminate non-viable cells from the analysis. At least 10,000 events were acquired in each experiment. The source of antibodies and their dilutions are shown in supplemental online Table 2.

Tumor Formation in SCID Mice

10,000 cells from ALDHbright and ALDHdim were injected into the testis of adult severe combined immunodeficient (SCID) mice. After 10 weeks, SCID mice were sacrificed, tissues were dissected, fixed in Bouins fixative (Sigma Aldrich) overnight, processed and sectioned according to standard procedures and counterstained with either hematoxylin and eosin or Weigerts stain. Sections (5–8 μm) were examined using bright-field light microscopy and photographed as appropriate. As a control, 10,000 human embryonic stem cells were injected into the testis of two adult SCID mice.

Statistical Analysis

For two groups of data, the two paired Student's t-test was used to obtain probability (p) values. For three or more groups of data, one- way analysis of variance (ANOVA) was used to obtain p values. Results with p values of less than 0.05 were considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Lack of ALDH1 Immuno-Staining in Human Limbal Epithelial Basal Cells

Immunoreactivity for ALDH1 was present in the suprabasal layers of the corneal and limbal epithelium, but not at the basal layer of both epithelia (Fig. 1A and 1B). ALDH1 staining was cytoplasmic and stronger in the cornea compared to the limbus. The lack of ALDH1 staining in the basal layers of cornea and limbus could be used to enrich LSCs by negative selection. This would, however, require the physical dissection of the limbal region from the corneal region in order to exclude the negatively ALDH1 negative cells of the corneal epithelial basal layer. This is technically difficult due to the lack of a clear anatomical separation of the limbal and corneal epithelia.

thumbnail image

Figure Figure 1.. Expression of ALDH1 in cornea and limbus. Immunohistochemistry on sections of peripheral cornea (A) and limbus (B) for aldehyde dehydrogenase one. Positive staining can be seen in brown. The slides are counter-stained with hematoxylin. Staining can be seen in most layers of the corneal and limbal epithelium but is absent in the basal layer. The basal layer is indicated by black arrows, the suprabasal layers by red arrows and stained cells with blue arrows.

Download figure to PowerPoint

The Cell Population Lacking Aldehyde Dehydrogenase Activity Is Enriched for LSCs

Cornea-limbal rings were disassociated into single cells and subjected to Aldefluor assay followed by flow cytometry to identify ALDH bright cells (Figs. 2A and 2B). The flow cytometry analysis indicated that 20.69% of the cells entrap Aldefluor reagent when compared to the diethylaminobenzaldehyde containing control sample (n= 7, Fig. 2A and 2B). It must be emphasized that mitotically inactivated 3T3 fibroblasts do not entrap Aldefluor (data not shown); therefore, expansion of the limbal epithelial cells by this co-culture does not affect the Aldefluor staining or analysis. Interestingly, the cells taking up relatively more Aldefluor tended to be larger than those taking up relatively less Aldefluor, which showed a small or intermediate size (Figs. 2C and 2D).

thumbnail image

Figure Figure 2.. A sub-population of cultured limbal epithelial cells entraps Aldefluor. (A): A representative example of flow cytometry dot plot for the control limbal epithelial cell sample containing inhibitor (diethylaminobenzaldehyde) as well as Aldefluor reagent. The fluorescent intensity (FL1-H), directly proportional to the amount of Aldefluor within the cell, is shown in the x-axis. Side scatter is shown on the y-axis. (B): A representative example of flow cytometry dot plot for the positive limbal epithelial cell sample containing only Aldefluor reagent. There is a significant increase in the cell population within the gate (shown in green) as compared to figure A. (C): Phase contrast micrograph of a limbal epithelial cell suspension. (D): Fluorescent micrograph of cells shown in figure C for Aldefluor staining (shown in green). Cells entrapping more Aldefluor (as shown in green) are indicated by the green arrows, and those showing basal levels of Aldefluor (that is moving freely in and out of the cells) are indicated by the white arrows. The 50 μm scale-bar is shown. Abbreviations: FL1-H, fluorescent intensity; SSC-H, side scatter.

Download figure to PowerPoint

Aldefluor analysis was performed on freshly isolated limbal epithelial cells as well as primary cultures and subcultures. This indicated that the percentage of ALDHbright limbal epithelial cells increases from primary cultured human limbal epithelial cells through to the sub-cultures 1 and 2 (Fig. 3A). The percentage of ALDHbright limbal epithelial cells increased from 13.3% at primary passage to 29.2% at subculture two (p= .001145; n=3). This was accompanied by a decline in colony forming efficiency (Fig. 3B; p= 6.07E-14; n=3) and ΔNp63 expression (Fig. 3C; p= .04; n= 3). Our own observations had indicated that the frequency of differentiated cells (with elongated morphology) increases when primary cultures are left to become confluent. It is for this reason that we investigated any correlations between loss of stem cell marker expressions and ALDH staining during this process. Following culture confluence, the percentage of ALDHbright limbal epithelial cells increased (Fig. 4A; p= .042; n=3) while the colony forming efficiency (Fig. 4B; p= .049, n=3) and ΔNp63 expression decreased (Fig. 4C; p= .027; n=3). Taken together these data suggest a correlation between loss of LSCs following ex vivo expansion and confluency and increase in percentage of cells with high aldehyde dehydrogenase activity.

thumbnail image

Figure Figure 3.. Subculture results in loss of limbal stem cells and increase in number of ALDHbright cells. (A): Staining of uncultured and ex vivo expanded limbal epithelial cells with the Aldefluor reagent. (B): CFE analysis of uncultured and ex vivo expanded limbal epithelial cells. (C): Real time reverse transcription polymerase chain reactions on uncultured and ex vivo expanded limbal epithelial cells. Sub-culturing was carried out every seventh day for example, primary culture represents 7 days of culture, sub-culture one represents 14 days of culture, and so on. The ΔNp63 to glyceraldehyde-3-phosphate dehydrogenase ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the positive control (uncultured cell population) was used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results shown in (A–C) was assessed using ANOVA one way. Abbreviations: CFE, colony forming efficiency; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Download figure to PowerPoint

thumbnail image

Figure Figure 4.. Confluency results in loss of LSCs and increase in number of ALDHbright cells. (A): Staining of limbal epithelial cells with the Aldefluor reagent prior to and post confluency. Statistical significance of the results was assessed using paired Student t-test. (B): CFE analysis of limbal epithelial cells prior to and post confluency. Statistical significance of the results was assessed using paired Student t-test. (C): Real time reverse transcription polymerase chain reactions and prior to and post confluency. The ΔNp63 to glyceraldehyde-3-phosphate dehydrogenase ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the positive control (uncultured cell population) was used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results shown in (C) was assessed using paired Student t-test. Abbreviations: CFE, colony forming efficiency; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Download figure to PowerPoint

Cultured human limbal epithelial cells were sorted on the basis of Aldefluor retention activity into ALDHdim and ALDHbright populations (Figs. 5A). The sorted samples were then analyzed by flow cytometry to confirm the efficiency of the sorting procedure (data not shown). Following cell sorting, the CFEs were carried out on sorted ALDHdim and ALDHbright cells (Fig. 5B) after 12 days in culture. The CFEs analysis showed that the CFE of ALDHdim cells was significantly higher than the CFE of ALDHbright cells (3.6 ± 0.115% vs. 0.6 ± 0.05%; p= .00198; n=3), suggesting that ALDHdim population contains most of the LSCs/limbal progenitor cells. Both cell populations were injected into the testis of SCID mice to check for tumor formation. We found no evidence of tumor formation after 10–12 weeks of injection (data not shown) suggesting that neither cell population possessed a tumorigenic phenotype.

thumbnail image

Figure Figure 5.. Fluorescence activated cell sorting of limbal epithelial cells on the basis of Aldefluor entrapment. Flow cytometry dot plot of the limbal epithelial sample prior to cell sorting. (A): The cells marked in red represent the cell population with low aldehyde dehydrogenase activity which enables Aldefluor reagent to move freely in and out of the cells (ALDHdim). The cells marked in green represent the cells with high aldehyde dehydrogenase activity which entrap Aldefluor reagent within the cells. The fluorescent intensity, directly proportional to the amount of Aldefluor within the cell, is shown in the x-axis. Side scatter is shown on the y-axis. (B): Colony forming efficiency analysis of limbal epithelial cells sorted on the basis of Aldefluor entrapment (n= 3) showing a significantly higher CFE in the ALDHdim population compared to ALDHbright. (C): Real time reverse transcription polymerase chain reactions on ALDHdim and ALDHbright sorted limbal epithelial cells. The gene of interest to glyceraldehyde-3-phosphate dehydrogenase ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the ALDHbright was used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results was assessed using paired Student t-test. Abbreviations: ABCG2, ATP binding cassette G2; CFE, colony forming efficiency; CK, cytokeratin; FITC, fluorescein isothiocyonate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; SSC-A, side scatter.

Download figure to PowerPoint

The expression of cytokeratin 3 (CK3), ΔNp63 and ABCG2 in the ALDHdim and ALDHbright sorted limbal epithelial cells was investigated using real time RT-PCR (Fig. 5C, n=3). The ratio of the gene to GAPDH was then calculated for each sample and this was compared to the expression level in the ALDHbright sample. Expression of the putative LSCs markers, ΔNp63 and ABCG2 was significantly higher in the ALDHdim cell population than ALDHbright. The expression of CK3 was not significantly different between the two cell populations. The real time RT-PCR data along with the IHC (which showed a lack of ALDH1 in the limbal epithelial basal cells) and CFE analysis, suggest that the ALDHdim cell population is enriched for limbal progenitor cells.

Large Scale Transcriptional Comparison of ALDHBright and ALDHDim Cell Populations

The cRNAs obtained from ALDHbright and ALDHdim populations were hybridized to the Human U133 Plus two array. The microarray experiments were performed in duplicate for each sample. 21,751 and 22,178 genes were identified as present or marginally expressed in the ALDHdim and ALDHbright samples respectively. Genes up-regulated twofold or more in both ALDHdim samples compared to ALDHbright samples (650 genes) and genes up-regulated twofold or more in both ALDHbright samples compared to ALDHdim samples (421 genes) were identified using the filter on fold change tool provided by GeneSpring software. KEGG analysis was then used to cluster these genes according to their biological function (see supplemental online Fig. 1). This analysis revealed an important cluster of genes involved in extracellular matrix (ECM) interactions. Our ongoing work is focused on understanding the role of ECM in limbal stem cell maintenance and its role in a number of signaling pathways that we have identified using antibody array technology (S. Kolli and M. Lako, unpublished data). It is for this reason that we focused our attention on this gene cluster. A few examples include collagen IV α2 (an isoform of collagen IV found in the limbal but not corneal basement membrane which was up-regulated 2.147-fold in the ALDHdim compared to the ALDHbright), collagen type XVIII, alpha one (that showed 2.138-fold upregulation in the ALDHdim compared to ALDHbright) and the receptor for hyaluronan mediated motility, RHAMM (which was down-regulated 2.057-fold in the ALDHdim population compared to the ALDHbright). Our own work [20] and others has shown an important role for RHAMM in the maintenance of pluripotency and viability in human embryonic stem cells and proliferation, trafficking and mobilization of hematopoietic stem cells [33, [34]35]. However the transcriptional data obtained herein suggests a role and/or expression for RHAMM in differentiated limbal epithelial cells which has not been described previously. In view of this knowledge and cell surface expression which enhances its utility as a potential marker, the role of RHAMM/HMMR in limbal epithelial cell biology was further studied.

RHAMM/HMMR Is a Negative Marker for Limbal Stem Cells

RHAMM/HMMR expression in the ALDHdim and ALDHbright sorted limbal epithelial cells was investigated using real time RT-PCR (Fig. 6A). This analysis showed that RHAMM/HMMR expression was three times higher in the ALDHbright population as compared to the ALDHdim population corroborating the microarray analysis results (p= .022; n=3). Immunohistochemistry for RHAMM/HMMR was performed on fresh cryosections of human cornea and limbus. This revealed the presence of RHAMM/HMMR staining throughout the layers of the cornea (Fig. 6B). In contrast, RHAMM/HMMR expression in the limbus was localized to the suprabasal layers and completely absent from the basal cells of the limbal epithelium (Fig. 6C).

thumbnail image

Figure Figure 6.. RHAMM expression in the central cornea and limbus. (A): Real time reverse transcription polymerase chain reactions on ALDHdim and ALDHbright sorted limbal epithelial cells for RHAMM expression. The RHAMM/HMMR to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the ALDHbright was used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results was assessed using paired Student t-test. (B and C): Immunohistochemistry on sections of cornea and limbus for RHAMM/HMMR. Positive staining can be seen in green with propidium iodide nuclear counterstaining. Note the lack of staining in the basal layers of the limbus. The basal layer is indicated by white arrows while blue arrows are used to mark labeled cells. (D): Colony forming efficiency analysis for sorted RHAMM/HMMR + and RHAMM/HMMR− populations (n= 3). Abbreviations: CFE, colony forming efficiency; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; RHAMM/HMMR, hyaluronan mediated motility receptor.

Download figure to PowerPoint

Flow cytometry was performed to quantify the number of limbal epithelial cells expressing surface RHAMM/HMMR. This analysis revealed that 40 ± 2.4% of the viable cells expressed RHAMM/HMMR after the second subculture (n= 3). The RHAMM/HMMR positive (+) and RHAMM/HMMR negative (−) cell populations were separated using flow cytometry (98% sort purity; data not shown). The two sorted cell suspensions of RHAMM/HMMR− and RHAMM/HMMR+ cells were then observed under the phase contrast microscope. Indeed the RHAMM/HMMR− cells were smaller than the RHAMM/HMMR+ cells (data not shown). The CFE analysis indicated that a higher number of colonies were obtained from the RHAMM/HMMR− versus RHAMM/HMMR+ cell populations (Fig. 6D). Real time RT-PCR analysis for RHAMM/HMMR, CK3, ΔNp63 and ALDH1A revealed negligible RHAMM/HMMR, ALDH1A and CK3 expression in the RHAMM/HMMR− sample as compared to the RHAMM/HMMR+ sample (n= 3) and higher ΔNp63 expression in the RHAMM/HMMR− sample as compared to the RHAMM/HMMR+ sample (Fig. 7A). Collectively these results indicate that RHAMM/HMMR expression is predominantly found in differentiated cells, thus suggesting that RHAMM/HMMR can be used as a negative marker of a stem cell-containing population of human limbal epithelial cells.

thumbnail image

Figure Figure 7.. RHAMM/HMMR as a negative marker for human LSCs. (A): Real time reverse transcription PCR for RHAMM/HMMR, ΔNp63, ALDH1A and CK3 in sorted RHAMM/HMMR+ and RHAMM/HMMR− cell populations. The gene to GAPDH ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the RHAMM/HMMR+ used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results was assessed using pairwise Student t-test. (B): Photomicrograph of RHAMM/HMMR+ and RHAMM/HMMR− cell populations taken at two subcultures after FACS purification procedure. (C): Real time reverse transcription PCR for ABCG2, ΔNp63 and CK3 in sorted autocultured RHAMM/HMMR+ and RHAMM/HMMR− cell populations. The gene to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) ratio is shown on the y-axis. The data represent the mean ± SD from three independent experiments. The value for the RHAMM/HMMR+ used as the calibrator (therefore= 1) and all other values were calculated with respect to this. Statistical significance of the results was assessed using pairwise Student t-test. (D): Flow cytometry dot plot of the limbal epithelial sample stained with the Hoechst 33,342 DNA binding dye. The cells marked in green represent the SP (gate P3). (E): Flow cytometry dot plot of the limbal epithelial sample stained with the Hoechst 33,342 DNA binding dye and RHAMM/HMMR antibody. The cells marked in green represent the RHAMM/HMMR-SP (left panel) and RHAMM/HMMR+SP (right panel). Abbreviations: ABCG2, ATP binding cassette G2; CK3, cytokeratin 3; RHAMM/HMMR, hyaluronan mediated motility receptor.

Download figure to PowerPoint

To investigate in more detail the properties of LSC population that can be enriched using negative cell selection based on RHAMM/HMMR expression, we purified RHAMM/HMMR positive and negative cell fractions and subjected them to two further subcultures. The RHAMM/HMMR− cells produced round colonies containing small cells of tight compact morphology, while RHAMM/HMMR+ cells differentiated extensively showing an elongated and irregular morphology that was very different to the RHAMM/HMMR− fraction (Fig. 7B). Real time RT-PCR analysis also indicated that RHAMM/HMMR− cells had a higher expression of putative LSC markers (ΔNp63 and ABCG2) and lacked CK3 expression that characterizes differentiated cells (Fig. 7C). In addition, clonal analysis of RHAMM/HMMR− cells indicated presence of holoclone type colonies, while 10% holoclone, 20% meroclones and 70% paraclone colonies were obtained from RHAMM/HMMR+ cells. We also compared the efficiency of LSC selection using negative enrichment based on RHAMM/HMMR expression or Hoechst exclusion as described in [36]. We were able to detect 0.02% of cells within the SP (gate P3, Fig. 7D) which has been described to contain cells with LSC properties [36]. CFE analysis indicated that SP population had a much higher colony forming efficiency (4%, n=3) compared to the G0/G1 population (gate P4, Fig. 7D) which showed 1% efficiency (n= 3). The CFE of SP population was very similar to RHAMM/HMMR negative population (Fig. 6D) suggesting that both sorting methods are well comparable to each other. To investigate whether combination of negative and positive cell selection could result in higher CFE, we performed FACS using both cell surface RHAMM expression and Hoechst exclusion (Fig. 7E). This analysis indicated that half of the SP population was marked by RHAMM/HMMR expression (Fig. 7E). CFE analysis indicated an increase in colony forming efficiency (up to 8%, n=3) when RHAMM−/SP cell population was plated on feeder cells. No colonies were observed when RHAMM+SP population was subcultured on 3T3 cells.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

The Aldefluor reagent has been used successfully for enrichment of hematopoietic stem cells from both human and murine bone marrow [15, 16]. Since this had not been investigated previously for limbal epithelial cells, the purpose of this study was to investigate the usefulness of such strategy for enrichment of putative LSCs. Staining of human limbal epithelial cells with Aldefluor reagent followed by flow activated cell sorting allowed us to purify two cell populations, namely cells that expressed ALDH at low levels (ALDHdim) and cells that showed high levels of expression (ALDHbright). Analysis of these two populations indicated that the ALDHdim cell population is enriched in putative LSCs. It must be emphasized, however, that other cell types in addition to LSCs are present in the ALDHdim population in view of the lack of ALDH1 staining in basal layers of both peripheral cornea and limbus.

Using comparative large scale transcriptional profiling of ALDHdim and ALDHbright populations, we were able to identify a new cell surface marker, RHAMM/HMMR that showed its highest expression in the ALDHbright cells. Immunocytochemistry and flow cytometry analysis indicated that RHAMM/HMMR can be used as a negative marker of a stem cell-containing population of human limbal epithelial cells. Our data strongly suggests that RHAMM/HMMR is a better negative marker than ALDH1 since its expression is seen throughout most layers of the cornea including the basal layer, however it is completely absent in the basal layer of the limbal epithelium. In view of this, no physical dissection of limbus from cornea would be necessary prior to enrichment of LSCs using RHAMM/HMMR as negative marker.

Limbal stem cell deficiency (LSCD) is a painfully blinding disease caused by the loss or dysfunction of LSCs [21]. It is caused by chemical or thermal burns to the eye, hereditary diseases (such as aniridia and ectodermal dysplasia), contact lens related eye disease, or inflammatory eye diseases (such as Stevens-Johnson Syndrome and ocular cicatricial pemphigoid). In LSCD, the loss of LSCs leads to degradation of the corneal epithelium which becomes replaced by the conjunctival epithelium and its accompanying blood vessels that surround the cornea and limbus. This results in the loss of corneal clarity and hence visual loss, and an epithelial surface which is prone to breaking down and causing pain. The treatment of LSCD relies upon the transplantation of live LSCs [22, 23].

LSCs have in the past been identified on the basis of being p63 positive and CK3/12 negative. However, the use of these markers to identify LSCs makes it impossible to isolate and characterize live LSCs since cells have to be permeabilized which results in loss of cell viability. In the last few years, it has been possible to isolate a population of clonogenic limbal epithelial cells by FACS using ABCG2 transporter as a marker or Hoescht 3,342 dye staining. Limbal epithelial cells isolated by this method showed a high CFE and stronger expression of p63 [7, 24].

We investigated in particular the expression of ALDH1 in corneal, limbal and conjunctival sections. A strong expression pattern was noticed in the cytoplasm of corneal and limbal suprabasal layers. Its expression was completely absent in the basal limbal and corneal epithelial cells. We were able to separate the ALDHdim and ALDHbright using one of the most recent substrates, Aldefluor (BD Biosciences), previously used by our group for the isolation of hematopoietic stem cells from the bone marrow [16]. Previous work from our laboratory and others have shown that embryonic stem cells and some somatic stem cells (such as the hematopoietic stem cells and intestinal stem cells) have enhanced defense mechanisms against oxidative stress and high DNA repair efficiency [14, 25]. This is based on high expression of several types of enzymes such as ALDHs and superoxide dismutases which are in turn downregulated upon differentiation of stem cells to more committed progenitors [25]. These properties can be used to isolate the stem cells by using fluorescent substrates, which can be trapped intracellularly under the action of specific enzymes, and render the cells separable by flow cytometry [16]. In particular, corneal epithelial cells have been shown to express high levels of ALDH which is thought to provide protection from harmful UV radiation [10].

We then investigated the two populations for the expression of putative LSCs markers (ΔNp63 and ABCG2) as well as the terminally differentiated corneal markers, CK3. Using real time RT-PCR, we were able to show that ΔNp63 and ABCG2 expression was significantly higher in the ALDHdim population. Most significantly, the ALDHdim population showed a higher CFE compared to the ALDHbright cells. Collectively, these findings suggest that ALDHdim population is enriched for putative LSCs. This is opposite to what has been found for bone marrow and neural stem cells where the stem cells are present in ALDHbright population [17, 26]. We believe that it does reflect a slightly different role for aldehyde dehydrogenases in the cornea than neural stem cells, which is to protect them from light induced oxidative stress.

The stem cell niche is well known to be vital for the maintenance of the LSC state [27]. As part of this niche, the ECM interactions are also clearly an important component. For this reason, analysis of the microarray data was used to identify such genes important in ECM interactions. This analysis revealed collagen type IV α2 isoform as being up-regulated in Aldefluordim cells. Indeed collagen type IV α2 isoform has been localized to the epithelial BM of the human limbus [28]. The microarray analysis also identified RHAMM as another important gene involved in ECM interactions as being up-regulated in the Aldefluorbright limbal epithelial cell population. The studies outlined in this manuscript indicate for the first time that RHAMM/HMMR is a useful negative cell surface marker for LSCs.

RHAMM/HMMR expression was absent in basal cells of the limbal epithelium. Our studies indicate that absence of RHAMM/HMMR expression correlated to properties associated with LSCs. RHAMM/HMMR− limbal epithelial cells were smaller in size, (a property associated with LSCs [29, 30]; expressed negligible CK3 (a terminal differentiated corneal cell marker) and higher levels of ΔNp63 (a putative LSC marker); and had a higher CFE as compared to RHAMM/HMMR+ cells.

RHAMM//HMMR is the receptor for hyaluronan, a glycosaminoglycan and important component of the ECM [31, 32]. Interestingly, the loss of β1 integrin expression after the up-regulation of RHAMM/HMMR characterizes the differentiation of thymocyte progenitors [33, 34]. It is indeed possible that a similar event occurs in the differentiation of LSCs. Since β1 integrin expression is seen in LSCs, it is possible that the up-regulation of RHAMM results in loss of β1 integrin expression and subsequent LSC differentiation [5]. Integrins are thought to anchor stem cells in their niche and therefore loss of expression is associated with differentiation. In addition to this, various studies have shown that the up-regulation of RHAMM results in increased mobilization of cancer cells and hematopoietic progenitors [33, 35]. A similar mechanism may be at play with increased RHAMM expression resulting in mobilization of LSCs from their niche and subsequent differentiation.

A recent study has reported that the number of ABCG2 positive cells (thought to correlate with LSCs) in human limbal epithelium is less than 1% [36]. The small number of stem cells makes it technically difficult to isolate them by positive selection using just one surface marker alone (e.g., ABCG2). Lineage depletion of differentiated cells has proved successful in purification of hematopoietic stem cells [37]. In this manuscript we report that a combined sorting strategy built on positive selection using Hoechst dye exclusion and negative selection based on cell surface RHAMM/HMMR expression results in purification of an enriched progenitor population displaying higher colony forming efficiency than enriched stem cell like populations obtained using each method alone.

Summary

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Our studies described for the first time the use of RHAMM/HMMR as a negative LSC marker. The proper identification and easy separation of these cells using flow cytometry based methods will allow better understanding of LSC biology and improvement of in vitro and in vivo procedures for successful clinical LSC isolation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

We thank UK Transplant for providing the donor tissue, Dr. Colin Jahoda and Dr. Frans Nauwelaers for helpful discussion, Dr. Brian Shenton and Ian Harvey for help with FACS, and Professor Fiona Watt for providing the 3T3-J2 fibroblasts. We are grateful to the Newcastle Healthcare Charity, Life Knowledge Park, One North East Developmental Agency and BD Biosciences for financial support. We also thank Dr. Volker Assmann for providing us with the RHAMM/HMMR antibody.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Summary
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
SC-07-0782_Supplemental_Data.pdf5KSupplemental Data
SC-07-0782_Supplemental_Figure_1.tif404KSupplemental Figure 1
SC-07-0782_Supplemental_Table_1.tif421KSupplemental Table 1
SC-07-0782_Supplemental_Table_2.tif344KSupplemental Table 2

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.