N-Cadherin Is Expressed by Putative Stem/Progenitor Cells and Melanocytes in the Human Limbal Epithelial Stem Cell Niche



Corneal epithelial stem cells are known to be localized to the basal layer of the limbal epithelium, providing a model system for epithelial stem cell biology; however, the mechanisms regarding the maintenance of these stem cells in their specialized niche remain poorly understood. N-cadherin is a member of the classic cadherin family and has previously been demonstrated to be expressed by hematopoietic stem cells. In the present study, we demonstrate that N-cadherin is expressed by putative stem/progenitor cells, as well as melanocytes, in the human limbal epithelial stem cell niche. In addition, we demonstrate that upon in vitro culture using 3T3 feeder layers, loss of N-cadherin expression occurs with cell proliferation. These results indicate that N-cadherin may be a critical cell-to-cell adhesion molecule between corneal epithelial stem/progenitor cells and their corresponding niche cells in the limbal epithelium.


Tissue-specific stem cells have been found in numerous adult tissues, including the bone marrow, epidermis, hair follicle, intestine, brain, testis, and cornea [1, [2]–3]. These cells, with their potential for self-renewal and the ability to differentiate into multiple lineages, have great therapeutic potential as a cell source for regenerative therapies in treating damaged or diseased tissues [4].

The concept of a stem cell niche was first proposed by Schofield in 1978 [5] as a specific microenvironment in which adult stem cells reside in their tissue of origin. Within the niche, stem cells are able to maintain their ability for self-renewal, as well as their multipotentiality. Consequently, detachment from the niche compartment induces stem cell differentiation and loss of self-renewal. In general, the niche is thought to consist of a highly organized microenvironment in which various factors, such as secreted cytokines, extracellular matrix interactions, and intercellular adhesion, are thought to work cooperatively to maintain the undifferentiated stem cell phenotype [2, 6].

Of all adult lineages, hematopoietic stem cells are the most well-established; we have significant understanding of the factors that regulate stem cell maintenance in the niche environment of the endosteal lining of the bone marrow cavity [7]. After the hematopoietic stem cell niche, epithelial stem cells, including the epidermis, intestinal epithelium, and corneal epithelium, are likely the most commonly examined adult stem cell systems [1]. In the epidermis, it has been established that the bulge area of the hair follicle functions as a niche. Likewise, for the corneal epithelial system, stem cells are thought to be localized to the epithelial basal layer of the limbus, the transitional zone between the cornea and the more peripheral bulbar conjunctiva [3, 8]. In particular, the corneal epithelial system is well-known as a model system for the study of all epithelial stem cells because of the localization of stem cells to the basal layer of the limbal epithelium and their spatial separation from their more differentiated progeny of the central corneal epithelium [9]. However, even though the localization of the stem cell niche has been determined, the factors and interactions that regulate the maintenance of epithelial stem cells remain poorly understood.

N-cadherin is a member of the classic cadherin family that mediates cell-to-cell adhesion via homophilic binding interactions [10, 11]. Recent reports regarding the bone marrow niche have demonstrated that hematopoietic stem cells expressing N-cadherin are attached to spindle-shaped N-cadherin+/CD45 osteoblasts that line the bone surface [12, [13]–14]. In terms of cell-to-cell interactions, the number of N-cadherin+ osteoblasts has been shown to control the quantity of hematopoietic stem cells. In addition, N-cadherin is known to maintain hematopoietic stem cell quiescence via angiopoietin-1 (Ang-1)/Tie-2 signaling [15, 16]. These findings suggest that the homophilic adhesion of N-cadherin between hematopoietic stem cells and osteoblasts plays an important role in the long-term maintenance of these tissue-specific stem cells within the bone marrow niche.

In the present study, we demonstrate that N-cadherin is expressed by both putative epithelial stem/progenitor cells and melanocytes in the limbal epithelium. These findings suggest the possibility that N-cadherin also plays an important role in the interactions between stem cells and their corresponding niche cells for the epithelial system.

Materials and Methods

Immunofluorescence Staining

Human corneoscleral rims (Northwest Lions Eye Bank, Seattle, http://www.nlfoundation.org/eyebank.cfm) (n = 3) were from cadaveric donor eyes (age range, 37–58) with death-to-preservation times of 2.8–5.5 hours and storage in Optisol (Bausch & Lomb, Rochester, NY, http://www.bausch.com) for 5–7 days prior to experimental use. Tissues were fixed in 4% paraformaldehyde (PFA) and processed into 10-μm-thick frozen sections. After drying for 1 hour at room temperature, tissue sections were washed three times with Tris-buffered saline (TBS) (Takara Bio, Shiga, Japan, http://www.takara-bio.com) and incubated with TBS containing 5% donkey serum and 0.3% Triton X-100 for 1 hour to block nonspecific reactions. Sections were then incubated with either a 1:100 dilution of anti-N-cadherin antibody (H-63; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), a 1:200 dilution of anti-E-cadherin monoclonal antibody (clone 36; BD Biosciences, San Diego, http://www.bdbiosciences.com), or a 1:30 dilution of anti-P-cadherin monoclonal antibody (clone 56; BD Biosciences) at 4°C overnight. Slides were again washed twice with TBS and incubated with a 1:200 dilution of Alexa Fluor 488-conjugated secondary antibodies (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com).

For double staining, N-cadherin-stained sections were incubated with a 1:25 dilution of anti-Melan-A antibody (M2-7C10/M2-9E3; Progen, Heidelberg, Germany, http://www.progen.de) or a 1:100 dilution of anti-keratin 15 (K15) antibody (LHK15; Chemicon, Temecula, CA, http://www.chemicon.com) at 4°C overnight and then a 1:200 dilution of Alexa Fluor 568-conjugated secondary antibodies (Molecular Probes) for 2 hours at room temperature. Finally, all tissue sections were counterstained with Hoechst 33342 (Molecular Probes) and observed by confocal laser scanning microscopy (Fluoview FV1000; Olympus, Tokyo, http://www.olympus-global.com). For all immunostaining experiments, slides treated with isotype-matched nonspecific IgG antibodies were used as isotype controls.

Cell Preparation

Limbal tissues were isolated from human corneoscleral rims (n = 11; donor age range, 30–66 years; death-to-preservation times, 3.2–8.0 hours; storage period before experimental use, 5–8 days) using scissors, and 8.0-mm-diameter portions of central corneas were obtained by trephination. Excised tissues were incubated separately in Dulbecco's modified Eagle's medium (DMEM) (Nikken Biomedical Laboratory, Kyoto, Japan, http://www.nikken-bio.co.jp) containing 2.4 units/ml dispase II (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) at 37°C for 1 hour. All epithelial layers were separated under a dissecting microscope and treated with 0.05% trypsin solution (Nacalai Tesque, Kyoto, Japan, http://www.nacalai.co.jp/en) containing 2.4 mM CaCl2 for 30 minutes at 37°C to create single-cell suspensions. Enzymatic activity was stopped by adding an equal volume of DMEM containing 10% fetal bovine serum (FBS) (Moregate Biotech, Brisbane, Queensland, Australia, http://www.moregatebiotech.com).

Flow Cytometric Analysis and Cell Sorting

Isolated epithelial cells from the limbal or corneal epithelium were resuspended in ice-cold Dulbecco's phosphate-buffered saline (PBS) (Nikken Biomedical Laboratory) containing 2% FBS (washing buffer) and kept on ice for all subsequent procedures. Limbal and corneal epithelial cells were incubated with anti-N-cadherin monoclonal antibody (GC-4; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) at 1 μg per 106 cells for 30 minutes. After two washes, cells were incubated with Alexa Fluor 488-conjugated anti-mouse IgG antibody (Molecular Probes) for 30 minutes on ice. For triple-color flow cytometric analysis, after incubation with anti-N-cadherin monoclonal antibody, limbal epithelial cells were treated with allophycocyanin (APC)-conjugated secondary antibody (Santa Cruz Biotechnology). Cells were then stained with phycoerythrin (PE)-conjugated anti-CD71 and fluorescein isothiocyanate-conjugated anti-integrin α6 monoclonal antibody (BD Biosciences). Prior to analysis, cells were incubated with 7-amino-actinomycin D (7′AAD) (BD Biosciences) to distinguish live from nonviable cells.

For flow cytometric analysis of intracellular proteins, after staining by APC-N-cadherin, cells were fixed with formalin (Nacalai Tesque), permeabilized with the IntraPrep Permeabilization Reagent (Immunotech, Marseille, France, http://www.immunotech.com), and incubated for 30 minutes with PE-conjugated anti-Melan-A monoclonal antibody (A103; Santa Cruz Biotechnology). Prior to analysis, 7′AAD was applied to stain cell nuclei. Appropriate background stainings were performed with the corresponding control antibodies. For flow cytometry, a FACSCalibur or FACSAria (BD Biosciences) system was used. Cell sorting was performed on a FACSAria instrument. Sorted cells were finally resuspended in medium and observed by phase-contrast microscopy (Axiovert 200; Carl Zeiss, Jena, Germany, http://www.zeiss.com).

Colony Forming Assays

Feeder layers were prepared by the seeding of mitomycin-C-treated NIH 3T3 cells at a density of 3 × 104 cells per square centimeter on 35-mm-diameter tissue culture dishes as described previously [17]. Separate preparations of N-cadherin+, N-cadherin, and all sorted limbal epithelial cells were seeded at densities of 250 and 500 cells per dish. After 10–14 days, the dishes were fixed and stained with rhodamine B (Sigma-Aldrich). Colony formation was then scanned under a dissecting microscope.

Gene Expression Analyses

Total RNA was obtained from equal numbers of N-cadherin+ or N-cadherin cells using the RNeasy total RNA kit (Qiagen, Valencia, CA, http://www1.qiagen.com). Reverse transcription was performed with the SuperScript First-Strand Synthesis System for reverse transcription-polymerase chain reaction (RT-PCR) (Invitrogen), according to the manufacturer's suggested protocol, and cDNA was used as a template for PCR. Quantitative real-time RT-PCR was performed using the ABI Prism 7900HT Sequence Detection System (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com), according to the manufacturer's suggested protocol. Primer pairs and TaqMan MGB probes labeled with 6-carboxyfluorescein (FAM) at the 5′-end and with nonfluorescent quencher at the 3′-end were designed with Assay-by-Design (Applied BioSystems). Quantitative PCR was performed a Prism 7900HT Sequence Detection System (Applied BioSystems). Thermocycling programs consisted of an initial cycle at 50°C for 2 minutes and 95°C for 10 minutes and 45 cycles at 95°C for 15 seconds and 60°C for 1 minute. Negative controls using non-reverse-transcribed total RNA as template strands were performed for all experiments. All assays were run in duplicate for six or more individual samples.

Melanocyte Culture

After sorting, N-cadherin+ or N-cadherin cells were centrifuged onto glass chamber slides (Nunc, Naperville, IL, http://www.nuncbrand.com) coated with type IV collagen at densities of 250–1,000 cells per well. These cells were maintained in a modified MCDB-153 medium (M254S; Kurabo, Osaka, Japan, http://www.bio.kurabo.co.jp/English) with 0.2% bovine pituitary extract, 5 μg/ml insulin, 3 ng/ml human basic fibroblast growth factor, 5 μg/ml transferrin, 0.5 μM hydrocortisone, and antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin), 3 μg/ml heparin, and 10 ng/ml phorbol 12-myristate 13-acetate. Cultures were incubated at 37°C in a humidified atmosphere of 5% CO2 for 5 days. Slides were washed with PBS and fixed in 4% PFA for 3 days at 4°C. Cells were finally stained by anti-Melan-A and anti-N-cadherin antibodies.

Statistical Analysis

Data are expressed as mean ± SE. Statistical analysis was performed using the Mann-Whitney rank sum test. All statistics were calculated using SigmaStat 2.0 (SPSS, Inc., Chicago, http://www.spss.com).


Localization of N-Cadherin-Expressing Cells in Human Limbal Epithelium

Immunofluorescence staining revealed that N-cadherin was expressed in the basal layer of the limbal epithelium and not in either suprabasal limbal cells or any epithelial layers of the central cornea (Fig. 1A, 1B). Although expression of N-cadherin was discontinuous in the limbal epithelial basal layer, N-cadherin was localized to intercellular areas between basal cells (Fig. 1C). In contrast, E-cadherin showed strong staining in all layers of the central corneal epithelium, as well as in superficial cells of the limbal epithelium, with weak expression detected in the middle and basal layers of the limbal epithelium (Fig. 1D, 1E). The expression of P-cadherin was not observed in either the corneal or the limbal epithelium (Fig. 1F, 1G).

Figure Figure 1..

Immunofluorescence staining of N-, E-, and P-cadherin in limbal and corneal epithelium. Distribution of cells expressing N-, E-, and P-cadherin proteins was examined by immunofluorescence staining (green). (A, B): N-cadherin was inconsistently and discontinuously stained in the basal cells of the limbal epithelium, but not in the corneal epithelium. (C): Magnified view of (A). (D, E): E-cadherin was mainly expressed in limbal superficial layers and in all layers of corneal epithelium. (F, G): P-cadherin was not detectable in limbal and corneal epithelium. Nuclei were labeled with Hoechst 33342 (blue). Scale bar = 50 μm.

N-Cadherin Is Expressed by Putative Epithelial Stem/Progenitor Cells

When limbal epithelial cells were subjected to flow cytometric analysis, results revealed that a small portion of cells expressed N-cadherin (Fig. 2A; 9.9% ± 0.3%; n = 7). In contrast, almost no N-cadherin+ cells were present in the epithelium of the central cornea (Fig. 2A; 0.33% ± 0.03%; n = 3). Forward scatter analysis showed that N-cadherin+ cells were smaller than N-cadherin cells (Fig. 2B). When limbal epithelial cells were subjected to triple-color flow cytometric analysis for N-cadherin, integrin α6, and CD71, the majority of N-cadherin+ cells were α6bright6bri)/CD71dim (Fig. 2C; 80.3% ± 2.0%; n = 3), which is believed to be a phenotypic trait of immature epithelial stem/progenitor cells [18, 19]. In addition, a smaller proportion of N-cadherin+ cells were α6dim/CD71dim (17.9% ± 2.0%) cells. In contrast, N-cadherin cells were mainly α6dim/CD71dim (70.5% ± 1.0%), although some cells were α6bri/CD71dim (18.3% ± 1.0%) or CD71bri (11.3% ± 0.2%) (Fig. 2C).

Figure Figure 2..

Flow cytometric analysis of N-cadherin expression in limbal and corneal epithelial cells isolated from eye bank tissues. (A): Isotype-control IgG (green lines) and anti-N-cad antibody GC-4 (blue filled plots). Limbal epithelium contained 9.9% ± 0.3% (n = 7) N-cad+ cells. In contrast, corneal epithelium contained 0.33% ± 0.03% (n = 3) N-cad+ cells. (B): Flow cytometric analysis of limbal and corneal epithelium for N-cad expression and FSC. N-cad+ cells showed a smaller FSC value than N-cad cells. (C): Triple-color flow cytometric analysis for expression of cell surface marker integrin α6 and CD71 in N-cad+ and N-cad cells. The majority of N-cad+ cells were α6bri/CD71dim (80.3% ± 2.0%; n = 3). In contrast, N-cad cells were mainly in the α6dim/CD71dim fraction (70.5% ± 1.0%; n = 3). Abbreviations: bri, bright; FSC, forward scatter; N-cad, N-cadherin.

N-Cadherin+ Cells from the Limbal Epithelium Have High In Vitro Proliferative Potential

To further examine the possibility that N-cadherin+ cells were enriched for epithelial stem/progenitor cells, colony forming assays were performed following isolation of N-cadherin+ cells using fluorescence-activated cell sorting (FACS) (Fig. 3A). Microscopic analysis showed that N-cadherin+ cells were significantly smaller than N-cadherin cells (Fig. 3A). When cells were isolated and subjected to the colony forming assay, all sorted cells demonstrated colony forming efficiency (CFE) (4.85% ± 0.58%; n = 6) similar to previously reported results [20]. In contrast, N-cadherin+ cells exhibited significantly higher CFE (22.0% ± 3.5%; n = 6) than N-cadherin and all sorted cells, whereas N-cadherin cells had little clonogenic capacity (1.0% ± 0.41%; n = 6) (Fig. 3B, 3C). These results indicated that via sorting for N-cadherin+ cells, enrichment for colony forming cells could be achieved. Although N-cadherin+ cells maintained their expression of N-cadherin at 24 hours after seeding on 3T3 feeder layers, no N-cadherin was detected within the cell colonies after 9 days of in vitro culture (Fig. 3D).

Figure Figure 3..

Colony forming efficiency (CFE) of limbal N-cadherin+ epithelial cells. To investigate the CFE of isolated limbal N-cad+ epithelial cells, a colony forming assay was performed. (A): Limbal N-cad+ and N-cad cells were isolated by cell sorting and observed on a phase-contrast microscope. N-cad+ cell consisted of smaller cells than N-cad cells (B): Representative dishes of 250 or 500 viable, sorted isolated epithelial cells, N-cad, or N-cad+ epithelial cells cultured on 3T3 fibroblast feeder cells. (C): The graph shows CFEs (mean ± SE) of all viable cells, N-cad, and N-cad+ cells from six cultures. ∗, p value <.05 between N-cad+ cells and all cells or N-cad cells. (D): Immunofluorescence staining for N-cad protein of N-cad+ epithelial cells cultured on 3T3 feeder layers after 24 hours and 9 days in vitro. N-cad was detected in N-cad+ cells after 24 hours culture (green; arrows), but not within cell colonies after 9 days culture in vitro. Nuclei were costained blue with Hoechst 33342. Scale bars = 50 μm. Abbreviations: h, hour(s); N-cad, N-cadherin.

N-Cadherin+ Cells Express Stem/Progenitor Cell-Related Markers

Cells isolated from the limbal epithelium were subjected to FACS for N-cadherin and then subjected to quantitative real-time RT-PCR. N-cadherin+ cells showed significantly higher expression of the stem/progenitor cell-related markers ΔNp63, K15, Bmi-1, and ABCG2 compared with N-cadherin cells (Fig. 4A). On the contrary, N-cadherin+ cells demonstrated almost no expression of both K3 and K12, which are known markers of differentiated corneal epithelial cells [3] (Fig. 4B).

Figure Figure 4..

Quantification of stem cell-related markers and differentiation markers in N-cadherin+ and N-cadherin cells. Quantitative real-time reverse transcription-polymerase chain reaction was performed for mRNA expression of the stem cell-related markers ΔNp63, K15, Bmi-1, and ABCG2 (A) and the differentiation markers K3 and K12 (B) in N-cad+ and N-cad cells. The relative gene expression was normalized to that of GAPDH in each sample. N-cad+ cells showed significantly higher expression of ΔNp63, K15, Bmi-1, and ABCG2 compared with N-cad cells. On the contrary, N-cad+ cells demonstrated almost no expression of both K3 and K12. The graphs show the mean ± SE of six or more independent samples. ∗, p value <.05. Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; K, keratin; N-cad, N-cadherin.

N-Cadherin Expression Is Associated with Both K15 and Melan-A in the Human Limbal Epithelium

K15, which has previously been reported as a marker of putative epithelial stem cells that is present in the bulge region of the hair follicle [21, 22], showed diffuse staining in the basal layer of the limbal epithelium (Fig. 5A) and could not be detected in the epithelium of the central cornea (data not shown). When limbal tissues were stained for both N-cadherin and K15, the results demonstrated that a large proportion of K15+ cells in the basal layer of the limbus also expressed N-cadherin (Fig. 5A–5D). However, some K15+/N-cadherin cells were also present.

Figure Figure 5..

Double staining of N-cadherin with K15 or Melan-A in the limbal epithelium. Colocalization of N-cad with K15 or Melan-A was examined by double immunofluorescence staining. (A): K15 (red); (B): N-cad (green); (C): N-cad/K15 merge; (D): Hoechst 33342 (blue). K15 was detected in the basal layer of limbal epithelium. Large parts of N-cad expression corresponded to K15 expression in limbus. (C): Asterisks (∗) indicate the areas where N-cad and K15 were colocalized. (E, J): Melan-A (red); (F, K): N-cad (green); (G, L): N-cad/Melan-A merge; (H, I): Hoechst 33342 (blue). Melan-A+ cells (melanocytes) discontinuously existed in basal layer of limbal epithelium. N-cad expression approximately corresponded with Melan-A. (G): Asterisks (∗) indicate the areas where N-cad and Melan-A were colocalized. (I–L): Magnified views of enclosed box shown in (G). Scale bar = 50 μm (A–H), 10 μm (I–F). Abbreviations: K, keratin; N-cad, N-cadherin.

Because our findings indicated that N-cadherin+ cells were enriched for epithelial stem/progenitor cells, we sought to examine the possibility of N-cadherin interactions of stem cells with niche cells in the limbal epithelium, analogous to the hematopoietic system. As melanocytes are known to be present within the limbal epithelium [23, 24] and have been shown to express N-cadherin [25], we examined the presence of Melan-A in relation to N-cadherin. Similarly to N-cadherin, Melan-A was only detected in the basal layer of the limbal epithelium (Fig. 5E) and could not be detected within the epithelium of the central cornea (data not shown). Interestingly, a large proportion of Melan-A+ areas were colocalized with N-cadherin+ areas (Fig. 5E–5H). Under higher magnification, limbal melanocytes appeared to adhere to epithelial cells in the basal layer via N-cadherin+ regions (Fig. 5I–5L).

N-Cadherin+ Cells from the Limbal Epithelium Contain a Population of Melanocytes

To confirm the presence of melanocytes within the fraction of N-cadherin+ cells, limbal epithelial cells were double stained with anti-Melan-A and anti-N-cadherin antibodies and subjected to flow cytometric analysis. Results showed that the limbal epithelium contained 4.9% ± 0.9% (n = 3) melanocytes, and within the fraction of Melan-A+ cells, 11.5% ± 2.6% (n = 3) were also positive for N-cadherin (Fig. 6A). Moreover, when N-cadherin+ and N-cadherin cells were subjected to quantitative real-time RT-PCR for Melan-A, both groups expressed approximately the same levels of Melan-A, with no significant difference (Fig. 6B). In addition, when N-cadherin+ cells were isolated and cultured for 5 days under serum-free conditions, N-cadherin+ cells maintained N-cadherin expression during in vitro culture, and Melan-A+ multipolar cells were observed from both N-cadherin+ (Fig. 6C) and N-cadherin cells (data not shown). As Melan-A is a specific marker of melanocytes, these results indicated that the fraction of N-cadherin+ cells contained melanocytes, in addition to epithelial stem/progenitor cells.

Figure Figure 6..

Melanocytes in limbal N-cadherin+ cells. (A): Flow cytometric analysis for N-cad and Melan-A in limbal epithelium. The limbal epithelium contained 4.9% ± 0.9% (n = 3) melanocytes, and 11.5% ± 2.6% (n = 3) of them were N-cad+. (B): Melan-A mRNA expression was detected to be approximately equivalent in N-cad+ and N-cad cells. (C): Sorted N-cad+ cells were cultured on chamber glass slides coated with type IV collagen under serum-free conditions. Cultured N-cad+ cells maintained N-cad expression (green), and multipolar cells were stained for Melan-A (red). Nuclei were labeled with Hoechst 33342 (blue). Scale bar = 50 μm. Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; N-cad, N-cadherin.


In this study, we present evidence that N-cadherin is expressed in a subpopulation of human limbal epithelial cells and that these N-cadherin+ cells are not only enriched for putative stem/progenitor cells but also contain melanocytes. These data suggest that N-cadherin is involved in homotypic interactions between neighboring stem cells, as well as heterotypic contact between stem cells and melanocytes, in the limbal epithelium. We also hypothesize that these N-cadherin+ melanocytes act as niche cells in the limbal epithelium.

Our immunofluorescence staining results showed that the expression of N-cadherin was localized with diffuse and discontinuous staining only in the basal layer of the limbal epithelium. In contrast, both E-cadherin and P-cadherin, which are known to be expressed in the epidermis [26, 27], were hardly detected in limbal basal layer. These data suggested that N-cadherin, but not E-cadherin or P-cadherin, is a major mediator of the intercellular interactions of basal limbal epithelial cells.

Likewise, the results of flow cytometry also indicated that N-cadherin expression was localized only in the limbal epithelium, with almost no N-cadherin+ cells present in the central corneal epithelium. In addition, forward scatter analysis and phase-contrast microscopy revealed that N-cadherin+ cells were smaller than N-cadherin cells, indicating their immature cell phenotype. To confirm the possibility that N-cadherin+ limbal epithelial cells were enriched for epithelial stem/progenitor cells, they were subjected to cell sorting for cell surface marker integrin α6 and CD71. It has been shown that cells with an α6bri/CD71dim phenotype correspond to slow-cycling keratinocyte stem cells in the skin and that they exhibit high CFEs [18, 19]. Likewise, we have also observed that α6bri/CD71dim cells are localized to the basal layer of the limbal epithelium and have significantly higher CFEs than the other cell fractions (data not shown). In the present study, we were able to confirm that a major fraction of N-cadherin+ cells was also α6bri/CD71dim, with significantly higher CFE compared with N-cadherin cells, indicating their stem/progenitor phenotypes. Similarly, N-cadherin+ cells also showed significantly higher mRNA expression of the stem cell-related markers for ΔNp63, K15, Bmi-1, and ABCG2, with significantly reduced expression of the differentiated cell markers K3 and K12. ΔNp63 is a well-recognized stem cell marker for both keratinocytes and corneal epithelial cells [28, 29], and Bmi-1 is required for the maintenance of self-renewal by both hematopoietic and neural stem cells [30, 31]. ABCG2 is expressed by side-population cells, which have recently been correlated to stem cell phenotypes in several tissues, including corneal epithelium [32, 33]. In addition, K15, which is a marker of the bulge region of the hair follicle that is known to be enriched for epithelial stem cells, was also colocalized with N-cadherin+ cells. These results therefore strongly suggest that N-cadherin is a marker of epithelial stem cells in the corneal system.

Immunofluorescence staining showed that a large portion of N-cadherin+ areas were colocalized with Melan-A+ areas, and melanocytes and N-cadherin+ cells seemed to adhere to each other via N-cadherin under higher magnification of N-cadherin+/Melan-A+ areas. When N-cadherin expression in limbal melanocytes was examined by flow cytometry, the results showed that the N-cadherin+ limbal cell fraction contained 4.9% ± 0.9% melanocytes, in a similar proportion to previously reported data [24]. In addition, 11.5% ± 2.6% of limbal melanocytes were shown to express N-cadherin. Melanocytes are known to exist in the epidermis or dermis [34], and epidermal melanocytes have been shown to express mainly E-cadherin and P-cadherin [35, 36]. In contrast, within the dermis, melanocytes are known to express N-cadherin [25]. Although the cadherin expression of human limbal melanocytes is uncertain, it appears that limbal melanocytes may have cadherin expression patterns different from those of epidermal melanocytes.

It is well-known that melanocytes protect keratinocytes from oxidative DNA base damage by synthesizing the pigment melanin and transferring it into epithelial cells [37]. Specifically, in the corneal epithelial system, epithelial stem cells that maintain the clear and unpigmented central cornea are known to be localized to the limbus, which can be identified physically by the palisades of Vogt, where melanin is deposited [38]. Recently, it has also been reported that melanocytes in limbus interact with K19+ basal epithelial cells [24]. K19 is thought by some to be an epidermal stem cell marker, similar to K15 [39, 40]. In the present report, we also show data that suggest that melanocytes likely interact with K15+ limbal basal cells, but importantly, N-cadherin+ melanocytes also appear to colocalize with basal limbal epithelial cells that represent putative stem/progenitor cells.

Recently, Zhang et al. have demonstrated that spindle-shaped N-cadherin+ osteoblasts fulfill the function of bone marrow niche cells for hematopoietic stem cells [12]. These hematopoietic stem cells also expressed N-cadherin, which likely acts as an anchor molecule for their attachment to N-cadherin+ osteoblasts. Our finding that N-cadherin+ corneal epithelial stem cells and melanocytes were likely colocalized in the basal layer of the limbal epithelium suggests the strong possibility that melanocytes have direct contact with epithelial stem/progenitor cells in the limbus via N-cadherin.

Interestingly, it has been shown that the Ang-1/Tie-2-signaling pathway maintains hematopoietic stem cells in the quiescent state in the niche, presumably by activating cell adhesion molecules such as N-cadherin [15]. In addition, previous studies have shown that melanocyte-keratinocyte contact inhibits keratinocyte proliferation in vitro [41]. We have also observed that the use of melanocytes as feeder cells cannot induce limbal epithelial cell proliferation in vitro (data not shown), indicating that melanocyte-epithelial stem cell interactions may also play a role in maintaining epithelial stem cell quiescence in vivo.

It should be noted that although the N-cadherin+ cell population exhibited a high CFE, only 22% of the N-cadherin+ cells were clonogenic in vitro. We hypothesize that N-cadherin+ cells with the ability for in vitro proliferation largely represent cells that are slow cycling in vivo. These slow-cycling cells may have high proliferative potentials, but their cell division is carefully regulated within the stem cell niche. However, upon in vitro culture, such as a colony forming assay on 3T3 feeder layers, they can proliferate and form colonies (Fig. 7). Interestingly, limbal N-cadherin+ cells maintained their expression of N-cadherin up to 1 day after seeding on 3T3, but they lost this expression over time in culture. These results therefore suggest that the downregulation of N-cadherin expression under in vitro culture conditions may allow for cells that are slow cycling in vivo to form large cell colonies in vitro. Therefore, N-cadherin may also be a critical cell-to-cell junction molecule in regulating stem cell quiescence, not only in the bone marrow but also in the limbal niche. However, the function of melanocytes and N-cadherin in the stem niche remains a subject of further investigation.

Figure Figure 7..

A model for the limbal epithelial stem cell niche. Schematic diagram shows that N-cad is expressed by a melanocyte and a limbal epithelial stem/progenitor cell in stem cell niche. N-cad mediates an intracellular interaction between the stem/progenitor cell and the melanocyte, which functions as niche cell. Detached from its niche, the epithelial stem/progenitor cell proliferates and differentiates into TA cells in vivo or forms colonies on 3T3 feeder layers in vitro. This proliferative capability appears to be correlated with a downregulation in N-cad expression. Abbreviations: CFE, colony forming efficiency; K, keratin; N-cad, N-cadherin; TA, transient amplifying.

In summary, our results demonstrate that a subpopulation of N-cadherin+ limbal epithelial cells contained putative corneal epithelial stem/progenitor cells. In addition, these N-cadherin+ cells also contained a fraction of melanocytes, and these two cell types appeared to colocalize via homophilic N-cadherin interactions. These findings therefore imply that N-cadherin plays an important role in the epithelial stem cell niche via interactions between melanocytes acting as niche cells and epithelial stem/progenitor cells.


The authors indicate no potential conflicts of interest.


This work was supported in part by Grants-in-Aid for Scientific Research 15390539, 16200036, and 16300161; the High-Tech Research Center Program; and the Center of Excellence Program for the 21st Century from the Ministry of Education, Culture, Sports, Science, and Technology in Japan and by Core Research for Evolution Science and Technology from the Japan Science and Technology Agency.