• Open Access

PEDF promotes self-renewal of limbal stem cell and accelerates corneal epithelial wound healing


  • Tsung-Chuan Ho,

    1. Department of Medical Research, Mackay Memorial Hospital Taipei, Taiwan, Republic of China
    Search for more papers by this author
  • Show-Li Chen,

    1. Department of Microbiology, School of Medicine National Taiwan University, Taipei, Taiwan, Republic of China
    Search for more papers by this author
  • Ju-Yun Wu,

    1. Department of Medical Research, Mackay Memorial Hospital Taipei, Taiwan, Republic of China
    Search for more papers by this author
  • Mei-Ying Ho,

    1. Department of Medical Research, Mackay Memorial Hospital Taipei, Taiwan, Republic of China
    Search for more papers by this author
  • Lee-Jen Chen,

    1. Department of Ophthalmology, Mackay Memorial Hospital, Taipei, Taiwan, Republic of China
    2. Mackay Medicine, Nursing and Management College, Taiwan, Republic of China
    Search for more papers by this author
  • Jui-Wen Hsieh,

    1. Department of Ophthalmology, Mackay Memorial Hospital, Taipei, Taiwan, Republic of China
    2. Mackay Medicine, Nursing and Management College, Taiwan, Republic of China
    Search for more papers by this author
  • Huey-Chuan Cheng,

    1. Department of Ophthalmology, Mackay Memorial Hospital, Taipei, Taiwan, Republic of China
    Search for more papers by this author
  • Yeou-Ping Tsao

    Corresponding author
    1. Department of Medical Research, Mackay Memorial Hospital Taipei, Taiwan, Republic of China
    2. Department of Ophthalmology, Mackay Memorial Hospital, Taipei, Taiwan, Republic of China
    • Mackay Memorial Hospital, Number 92, Sec. 2, Chung Shan North Road, Taipei, Taiwan, Republic of China. E-mail: yptsao@yahoo.com Telephone: 886-2-28094661, ext. 3076; Fax: 886-2-28085952

    Search for more papers by this author

  • Author contributions: T.-C.H.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; S.-L.C.: conception and design, data analysis; J.-Y.W.: provision of study material, collection and assembly of data; M.-Y.H.: conception and design, manuscript writing; L.-J.C.: provision of study material, conception and design, data analysis; J.-W.H.: conception and design, data analysis; H.-C.C.: data analysis and interpretation; Y.-P.T.: conception and design, data analysis, interpretation, manuscript writing; and final approval of manuscript.


Limbal epithelial stem cell (LSC) transplantation is a prevalent therapeutic method for patients with LSC deficiency. The maintenance of stem cell characteristics in the process of culture expansion is critical for the success of ocular surface reconstruction. Pigment epithelial-derived factor (PEDF) increased the numbers of holoclone in LSC monolayer culture and preserved the stemness of LSC in suspension culture by evidence of ΔNp63α, Bmi-1, and ABCG2 expression. BrdU pulse-labeling assay also demonstrated that PEDF stimulated LSCs proliferation. In air-lift culture of limbal equivalent, PEDF was capable of increasing the numbers of ΔNp63α-positive cells. The mitogenic effect of PEDF was found to be mediated by the phosphorylations of p38 MAPK and STAT3 in LSCs. Synthetic 44-mer PEDF (residues 78–121) was as effective as the full length PEDF in LSC expansion in suspension culture and limbal equivalent formation, as well as the activation of p38 MAPK and STAT3. In mice subjecting to mechanical removal of cornea epithelium, 44-mer PEDF facilitated corneal wound healing. Microscopically, 44-mer PEDF advanced the early proliferative response in limbus, increased the proliferation of ΔNp63α-positive cells both in limbus and in epithelial healing front, and assisted the repopulation of limbus in the late phase of wound healing. In conclusion, the capability of expanding LSC in cell culture and in animal indicates the potential of PEDF and its fragment (e.g., 44-mer PEDF) in ameliorating limbal stem cell deficiency; and their uses as therapeutics for treating corneal wound. STEM Cells 2013;31:1775–1784


Limbus anatomically locates between cornea and conjunctiva on ocular surface. The basal layer of limbal epithelium is enriched with a special cell population, named limbal epithelial stem cells (LSCs) [1]. The cornea contains a stratified squamous epithelium that turnovers rapidly. The renewal of corneal epithelium is supported by the transient amplifying cells (TACs) generated from asymmetric division of LSCs [[1, 2]]. The proliferation of LSCs remains relatively slow, but can be activated upon wounding to the corneal, which enables epithelial damage repaired process [[1, 2]]. Loss or dysfunction of LSCs (clinically termed LSC deficiency) may result from various pathological conditions such as Stevens Johnson syndrome, chemical or thermal injuries, severe dry eye syndrome, and contact-lens induced ocular surface disease. Once limbus is damaged, remaining LSCs may not be sufficient enough to sustain cornea epithelial cell turnover; and with the invasion of conjunctival epithelium, severe visual loss may be inevitable [3].

Presently, transplantation of the ex vivo expanded limbal epithelial sheet has become the most widely used therapy for LSC deficiency [4]. This therapeutic approach generally involves placing a small limbal biopsy removed from either the patient or a donor on transplantable carriers such as denuded human amniotic membrane (AM) to support the migration of limbal cells out of the biopsy and form a limbal-like epithelial sheet [[4, 5]]. Limbal transplantation failure is often resulted from the depletion of LSCs in culture [6], and transplantation using culture in which more than 3% LSCs was preserved has recently been reported to result in a better prognosis after transplantation [7]. Some modified approaches, such as enzymatically isolating the limbal cells from limbal tissue and expanding the isolated LSCs by suspension culture systems, contribute to maintain the stem cell population and prevent the expanded LSCs from spontaneous differentiation [[8-10]].

Pigment epithelial-derived factor (PEDF) is a 50-kDa secreted glycoprotein with multiple biologic effects on various types of cells. PEDF is effective in suppressing neovascularization in animal models of choroidal neovascularization (CNV) [11]. It is also a neuroprotective factor, protecting cultured neurons from glutamate toxicity [12]. It has been identified that the amino acid positions Val78-Thr121 of human PEDF (also termed 44-mer) determines the neurotrophic activity and binds receptors on the surfaces of different types of neurons [13]. The 44-mer has been reportedly not having sufficient antiangiogenic activity on CNV lesions [14]. On the other hand, another functional PEDF motif mapped at positions Asp44-Asn77, the 34-mer, is responsible for PEDF antiangiogenic activity and does not exhibit the neurotrophic activities of the 44-mer PEDF [15]. The effects of these PEDF functional peptides on LSCs and cornea have not yet been reported.

Previously, PEDF was only reported to stimulate the proliferation of neuronal progenitor cells. In mouse brain, PEDF is associated with the division of neural stem cell derived from the subventricular zone [16]. In cell culture, PEDF sustains the expansion of hippocampal progenitor cells [17] and retinal progenitor cells derived from ciliary body [18]. However, whether PEDF may serve as a mitogen for stem cell of other origin is not known. In this study, we identify that PEDF and its 44-mer derivative are effective in expanding LSCs in monolayer cell culture and thereby facilitating the construction of progenitor cell-enriched limbal equivalents. Moreover, in mouse eyes wounded by epithelial removal, 44-mer speeds up the healing process by stimulating the proliferation of progenitor cells in the epithelium that results in healing cornea and limbus.

Materials and Methods


HEPES-buffered Dulbecco's modified Eagle's medium (DMEM), Ham's/F12 medium, trypsin-EDTA, fetal bovine serum (FBS), antibiotic–antimicotic solutions and trypsin were purchased from Invitrogen (Carlsbad, CA). Hydrocortisone, dimethyl sulfoxide (DMSO), insulin–transferrin–sodium selenite (ITSE) media supplement, β-nerve growth factor (NGF), mitomycin C (MMC), bovine serum albumin (BSA), 5-bromo-2′-deoxyuridine (BrdU), Triton X-100, Hoechst 33,258 dye, and formalin were all from Sigma-Aldrich (St. Louis, MO). Dispase II and epidermal growth factor (EGF) were obtained from Roche (Indianapolis, IN). ΔNp63α polyclonal antibody and all the fluorescent dye-conjugated secondary antibodies were purchased from BioLegend (San Diego, CA). Mouse anti-Bmi-1 antibody (clone F6; 05-637) and keractin-3 (clone AE5; CBL218) were purchased from Millipore Corporation (Bedford, MA). Rat monoclonal anti-ABCG2 antibody (ab24115) was from Abcam (Cambridge, MA). Mouse anti-BrdU antibody (GTX42641) was from GeneTex (San Antonio, TX). SB203580 and STAT3 inhibitor (No. 573096) were purchased from Calbiochem (La Jolla, CA). PEDF was purified from human plasma via collagen I–sepharose resin as previously described [19] and was analyzed by SDS–polyacrylamide gel electrophoresis (PAGE) and immunoblotting using an anti-PEDF antibody. Peptides (34-mer and 44-mer) were synthesized and modified with acetylated at the NH2 termini and amidated at the COOH termini for stability and characterized by mass spectrometry (>95% purity) to order at GenScript (Piscataway, NJ).

Preparation of Feeder Cells

Confluent NIH-3T3 cells were incubated with 4 μg/ml MMC for 2 hours at 37°C under 5% CO2, trypsinized, and plated onto trans-well culture dishes (BD Biosciences) at a density of 2.2 × 104 cells/cm2.

Isolation and Culture of LSCs

LSCs were isolated from 6-month-old New Zealand white rabbits and used for cell-suspension culture by modifying a previously described method [[20, 21]]. Briefly, rabbit limbal tissues were washed in phosphate-buffered saline (PBS) containing 50 μg/ml gentamicin. After carefully removing the iris, and excessive sclera, the limbal rings were exposed to dispase II (1.2 IU/ml in Hanks' balanced salt solution free of Mg2 + and Ca2 +) at 4°C for 16 hours. The loosened epithelial sheets were removed with a cell scraper and separated into single cells by treating with 0.5 ml trypsin (0.25% and 0.01% EDTA) for 15 minutes at 37°C with gentle shaking. Cells were transferred to 9 ml of 10% FBS/DMEM/F-12 medium and then collected by centrifugation (400g for 5 minutes).

To determine the proliferative capacity of LSCs, approximately 1 × 105 cells were seeded on each well of a six-well plate and incubated with a DMEM/F-12 basal medium (10 mM HEPES, 5 ng/ml human EGF, 1% ITSE, 1% antibiotic–antimicotic solutions, 0.5% DMSO and 0.5 μg/ml hydrocortisone) supplemented with 10% FBS for 2 days, and then shifted to a basal medium or basal medium respectively containing 4.5 nM PEDF, 25 nM 34-mer, and 25 nM 44-mer for 3 more days until they reach confluence (passage 0). Cultures were incubated at 37°C under 5% CO2. LSCs were cocultured with MMC-treated NIH-3T3 fibroblast feeder cells located within transwell (0.4 μm pore, BD Biosciences, Bedford, MA). For passage, near-confluent cells were harvested with 0.25% trypsin and then 1 × 105 subcultured cells were cultured in the respective medium described above.

Colony-Forming Efficiency

Approximately 1 × 103 limbal cells were seeded in a 3.8-mm2 dish and cocultured with MMC-treated NIH-3T3 feeder cell located within the trans-well. The medium was changed every 2–3 days. Colonies were fixed by 4% paraformaldehyde (room temperature [RT] for 1 hour) 14 days later for immunostaining or crystal violet staining and photographed. The colony-forming efficiency (CFE) (%) was calculated using the formula: number of colonies formed/number of cells plated × 100.

Human Limbal Suspension Cells Cultured on Human AM

Human tissue was handled according to the tenets of the Declaration of Helsinki. This study was approved by the Institutional Review Board of the Mackay Memorial Hospital (Taipei, Taiwan). Corneoscleral tissues were obtained immediately after the central corneal button was used for corneal transplantation. The tissue was rinsed three times with DMEM/F-12 containing 50 μg/ml gentamicin and 1.25 μg/ml amphotericin B. After careful removal of excessive sclera, conjunctiva, iris, and corneal endothelium, the corneoscleral rim was trimmed to obtain limbal tissue cubes (explants; approximately 3 × 5 mm size). For cell-suspension culture, the cells of limbal explants were enzymatically separated by digestion with dispase II/trypsin-EDTA as described above.

Human AMs were obtained from healthy mothers underwent elective cesarean section with properly written informed consent. The membranes were washed with sterilized PBS containing antibiotics (50 μg/ml gentamicin), and stored at −80°C in DMEM and glycerol at a ratio of 1:1 (vol/vol). To prepare denuded AM, the AM was thawed, washed three times with sterilized PBS, and cut into pieces, each was approximately 2.5 cm2. Membranes were then deprived of their amniotic epithelial cells by incubating in 0.02% EDTA at 37°C for 2 hours to lessen cellular adhesion, followed by gentle scraping with a cell scraper.

Airlift Culture

Approximately 5–10 × 104 human limbal cells were suspended in 3 ml of basal medium supplemented with 10% FBS and then seeded onto denuded AM spread on the bottom of culture insert, and cocultured with MMC-inactivated 3T3 fibroblasts located within the transwell. After 2 days, cells were adhered to AM, the cells were then cultured in a basal medium or a basal medium containing 4.5 nM PEDF, 25 nM 34-mer, and 25 nM 44-mer, respectively, for another 7 days. Then, the level of the culture medium was lowered to expose the upper surface of the cells to air for additional 7 days. Basal medium was used and changed daily during airlifting. The epithelial sheets were embedded in OCT compound (Sakura Finetek Japan Co., Tokyo) and 5 μm sections were cut.


Deparaffinized tissue sections or 4% paraformaldehyde-fixed LSCs were blocked with 10% goat serum and 5% BSA in PBS containing 0.1% Tween-20 for 1 hour. Staining was done using primary antibodies respectively against ΔNp63α (1:150 dilution), BrdU (1:250 dilution), keratin-3 (1:250 dilution), ABCG2 (1:150 dilution) and Bmi-1 (1:150 dilution) at 37°C for 2 hours, followed by incubation with the appropriate rhodamine- or FITC-conjugated donkey IgG (1:500 dilution) for 1 hour at RT. Images were captured using a Zeiss epifluorescence microscope with a CCD camera and photographs taken using the Axiovert software.

Immunoblot Analysis

Cell lysis, SDS–PAGE, and antibodies use for immunoblotting were performed as described in our previous study [22]. The band intensity in immunoblots was evaluated with a Model GS-700 imaging densitometer (Bio-Rad Laboratories, Hercules, CA) and analyzed using Labworks 4.0 software.

BrdU Labeling

2 × 105 passage one limbal cells were seeded at FNC coating mix solution (Athena Enzyme Systems, Baltimore, MD)-coated slide and incubated in culture medium for 1 day. BrdU (final, 10 μM) was added to the culture for 2 hours. After fixing with 4% paraformaldehyde, cells were exposed to cold methanol for 2 minutes, and then treated with 1 N HCl at RT for 1 hour before performing immunofluorescence. For animal study, BrdU was reconstituted in DMSO as stock (80 mM). 10 μl of BrdU mixed with 90 μl of PBS was intraperitoneally injected into mouse at 3 hours before euthanasia.

RNA Extraction and Quantitative Real-Time PCR

Experiments were performed as previously described [23]. The sequence of specific polymerase chain reaction (PCR) primers was Ki67 sense, 5′- agaggctatgcctttgtgga −3′; antisense, 5′- tgatggctgatgaaatggaa −3′ (PCR product: 170 bp).

Preparation of Polyclonal Antibody to PEDF

The PEDF rabbit antiserum was raised by immunizing animals with bacterially expressed His-tagged PEDF [24]. Preimmune serum samples were collected as a negative control before immunizations. The production and specificity of the antibodies in each rabbit serum were determined using immunoblotting of a recombinant human PEDF derived from stable baby hamster kidney (BHK) cell transfectants (Chemicon, Temecula, CA), before their use for corneal wounding assay in animal.

Corneal Epithelial Wounding and Treatments

The experiments were performed using 8-week-old female C57BL/6 mice with the approval from the Mackay Memorial Hospital Review Board for animal investigation. Animals were anesthetized by an intraperitoneal injection of a mixture of zoletil (6 mg/kg) and xylazine (3 mg/kg). One filter paper (0.9 mm diameter) soaked with 20% ethanol was placed on the central cornea of right eye for 1 minute and then irrigated extensively with PBS. Subsequently, the mechanical epithelial scrape was performed by using a punch under a dissection microscope that created a circular injury (2 mm diameter) at the entire corneal region of the mouse eye without encroaching the corneal stroma, limbus, or conjunctiva [[25, 26]].

A 44-mer was reconstituted in DMSO as stock (5 mM) and mixed with TOBREX eye ointment (Alcon; containing 0.3% Tobramycin and 0.5% Chlorobutanol) to a concentration 50 μM. After the scrape injury, the mice were randomly assigned to different experimental groups and the right eye received 20 μl 44-mer-mixed ointment or DMSO-mixed ointment once a day. Wound size was determined by staining with topical fluorescein (Fluor-I-Strip, Ayerst Laboratories, Philadelphia, PA) and photographed with a digital camera. The area of defect was quantified from the photographs by using a computer-assisted image analyzer (Adobe Photoshop CS3 10.0) and was calculated as the percentage of residual epithelial defect at each time point / initial wound area.


Results were expressed as the mean ± SEM. One-way analysis of variance was used for statistical comparisons. A p < .05 was considered significant.


PEDF Enhances the Proliferation Potential of Limbal Stem Cells In Vitro

To determine the influence of PEDF or its fragments on proliferative capacity of LSCs, cells were cultured in the medium supplemented with PEDF or its peptide derivatives, and passaged, split culture at 4-day intervals. At passage 3, only the cells exposed continuously to PEDF or 44-mer grew to reach near-confluence; whereas cells cultured in basal medium or in 34-mer never achieved confluence, an indication that the growth potential of LSC in such culture environment were exhausted. (Fig. 1A; 3.8 ± 0.8 × 105 cells grown in basal medium vs. 7.9 ± 0.7 × 105 PEDF-treated cells and 8.3 ± 0.6 × 105 44-mer-treated cells). In addition, cell morphology under phase-contrast microscope showed that PEDF or 44-mer exposure was more uniformly small, clustered in most areas and a large nuclear–cytoplasmic ratio up to passage two (Supporting Information Fig. S1). However, cells cultured in basal medium or with 34-mer showed the enlargement of the cell size associated with a lessened cell density beyond the first passage. This suggests that PEDF is helpful to maintain the morphology of progenitor cells. Passage two cells (1 × 103 seeded in well of 12-well plate and with growth-arrested 3T3 cells placed in transwell) cultivated in medium supplemented with PEDF or 44-mer for 14 days showed a higher clonogenic capacity than that of the control cells (approximately fourfold; Fig. 1B). Moreover, when colony formation assay was performed by growing limbal epithelial cells on top of growth-arrested 3T3 feeder layer, limbal epithelial cells exposed to PEDF or 44-mer formed more holoclones than those exposed to 34-mer or basal medium (Supporting Information Fig. S2). Colonies immunostained by keratin-3 (i.e., a differentiation marker of cornea epithelium) confirmed that spontaneous differentiation of LSC in vitro is not affected by the treatment of 34-mer as the staining for LSCs grown in basal medium or in medium containing 34-mer are strong and homogeneous (Fig. 1C); whereas most PEDF or 44-mer treated cells remained undifferentiated or keratin-3 stained negative. The keratin-3 expression in cells subject to either treatment was also confirmed by flow cytometry, in which attenuated keratin-3 staining was observed in majority of cells exposed to PEDF and 44-mer PEDF (Supporting Information Fig. S3).

Figure 1.

Proliferative capacities of limbal cells under pigment epithelial-derived factor (PEDF) and its peptide derivatives. (A): Time course analysis. After subculture, 1 × 105 limbal cells were seeded in culture dish and cultivated in a conventional basal medium or the basal medium supplemented with either PEDF or PEDF peptides (34-mer and 44-mer) for 4 days, then the expanded cells were trypsinized and counted. Data represent four independent experiments with similar results. (B): Colony-forming efficiency assay. Four independent experiments were performed (n = 3, *, p < .002). (C): Representative immunocytochemical images of keractin-3 (green). The cell nuclei were stained with Hoechst 33,258 (blue). Scale bar = 50 μM. (D, E) Limbal cells were continuously cultivated in different media by a time period of 14 days. The expressions of limbal epithelial stem cell markers including ΔNp63α, ABCG2 and Bmi-1 were detected by immunofluorescence microscopy (original magnification, ×400). Representative images from three independent experiments. Scale bar: 10 μM. Abbreviations: CFE, colony-forming efficiency; PEDF, pigment epithelial-derived factor; UT, untreated.

To substantiate this finding, we further investigated the expressions of LSC markers such as ΔNp63α, Bmi-1 and ABCG2 by immunocytochemistry. Cells continuously treated with PEDF or 44-mer for 14 days were positive for cytoplasmic label ABCG2 and nuclear label ΔNp63α and Bmi-1 in most of cells, whereas expression levels of respective labels in cells grown in basal medium or in medium containing 34-mer were minimum (Figs. 1D, 1E). Furthermore, immunoblot analysis confirmed that the expressions of Bmi-1 and ΔNp63α in LSCs were enhanced by PEDF treatment (Supporting Information Fig. S4). These results demonstrate that LSCs cultured in basal medium would spontaneously differentiate toward cornea epithelium, whereas LSCs cultured in medium containing either PEDF or 44-mer will remain undifferentiated or without losing their stemness.

PEDF Promotes LSC Proliferation In Vitro

To determine the effects of PEDF and its short derived peptides on the self-renewal activity of LSCs, LSCs were pulse-labeled with BrdU (2 hours; Fig. 2A), then analyzed by immunostaining. β-NGF (100 ng/ml), a LSC mitogen, was used as an assay control [6]. It was found that proliferating ratio for LSCs cultivated in medium containing β-NGF, PEDF or 44-mer is higher than that of LSCs cultured in control medium (16.1 ± 1.5%, 17.6 ± 3.4%, and 20.4 ± 2.2% vs. 3.0 ± 0.5%; Fig. 2B), but 34-mer PEDF does not exhibit such effect. In addition, the levels of Ki67 transcript (a proliferation-associated nuclear antigen) detected by real-time Q-PCR was 2.1 ± 0.2-fold, 2.6 ± 0.4-fold, and 2.8 ± 0.2-fold increased in LSCs grown in medium supplemented with β-NGF, PEDF, and 44-mer (Fig. 2C). Collectively, LSC proliferation in culture is enhanced by PEDF or 44-mer PEDF.

Figure 2.

Pigment epithelial-derived factor and 44-mer enhance proliferation of limbal epithelial stem cells (LSCs). (A, B): LSC proliferation was determined by 5-bromo-2′-deoxyuridine (BrdU) labeling for 2 hours. LSCs (ΔNp63α) and BrdU were detected by immunofluorescence microscopy (original magnification, ×400). Scale bar = 10 μM. A representative picture of three independent experiments is shown. Ten randomly selected fields in each group were photographed, and the percentage of BrdU and ΔNp63α-double positive cells (pale pink) per total ΔNp63-positive cells was calculated. *, p < .002 versus untreated cells. (C): Ki67 mRNA in LSCs assayed by quantitative real-time polymerase chain reaction (PCR). The cycle threshold (Ct) value of the Ki67 PCR product and a control mRNA (GAPDH) were used to calculate relative quantities of mRNA between samples. Data represent three independent experiments. *, p < .05 versus untreated cells. Abbreviations: β-NGF, β-nerve growth factor; PEDF, pigment epithelial-derived factor; UT, untreated.

The Mitogenic Effect of PEDF on LSC Is Modulated by Induction of p38 MAPK and STAT3

To explore the molecular basis of LSC proliferation induced by PEDF, LSCs were treated with PEDF for intervals ranging from 5 to 60 minutes, and immunoblots revealed that p38 MAPK and STAT3 were respectively phosphorylated at 5 minutes after PEDF treatment. The peak phosphorylation of p38 MAPK (p-p38) appeared between 5 and 10 minutes, and its nuclear substrate, ATF2, was subsequently phosphorylated. In addition, the peak phosphorylation of STAT3 (p-STAT3) occurred between 20 and 40 minutes (Fig. 3A, 3B). PEDF had no effect on the phosphorylations of ERK and JNK. On the other hand, similar to PEDF treatment, 34-mer and 44-mer peptides respectively induced phosphorylation of p38 MAPK (Fig. 3C). It was further noted that 44-mer also resulted in phosphorylation of STAT3, with the same kinetics as those observed in PEDF-treated cells. Immunofluorescence analysis revealed that the majority of p-38 and p-STAT3 in LSCs were located at nucleus after PEDF stimulation for 5 and 20 minutes, respectively (Fig. 3D). The identity of LSC phenotype was confirmed by Bmi-1 or ΔNp63α. Without PEDF treatment, the levels of p-p38 MAPK and p-STAT3 were barely detectable.

Figure 3.

PEDF activates p38 MAPK and STAT3 in limbal epithelial stem cells (LSCs). (A, B): Pigment epithelial-derived factor (PEDF) time dependently induces phosphorylations of p38 MAPK, ATF2, and STAT3. Limbal epithelial stem cells (LSCs) were exposed to PEDF for the indicated time periods. Immunoblotting was performed to detect the active phosphorylated forms (upper panels) and then unphosphorphated forms (lower panels) by reincubated with indicated antibodies. Representative blots and densitometric analysis from four independent experiments are shown. (C): LSCs were exposed to 25 μM of PEDF peptides for the times indicated. Immunoblotting was performed as described above. (D): Localization of p-p38 MAPK and p-STAT3 in LSCs. Cells were left untreated or treated with PEDF for 5 minutes to detect p-p38 MAPK and for 20 minutes to detect p-STAT3. LSC phenotype was recognized by staining of Bmi-1 and ΔNp63α. Representative images from three independent experiments. Original magnification, ×1,000. (E): p38 MAPK or STAT3 inhibitors prevent PEDF (P)-induced LSC proliferation. LSCs were pretreated with SB203580 (10 μM; p38 MAPK inhibitor) or 50 μM STAT3 inhibitor for 1 hour and then treated with PEDF for 24 hours. Subsequently, LSC proliferation was determined by 5-bromo-2′-deoxyuridine labeling for 2 hours as described in legends of Figure 2. #, p < .005 versus PEDF-treated cells. Abbreviations: BrDU, 5-bromo-2′-deoxyuridine; DAPI, 4',6-diamidino-2-phenylindole; p-ATF2, phospho-activating transcription factor-2; PEDF, pigment epithelial-derived factor; p-STAT3, phosphorylation of STAT3; UT, untreated.

Chemical inhibitors were used to examine the involvement of p38 MAPK and STAT3 in LSC proliferation induced by PEDF. BrdU pulse-labeling assay revealed that pretreatment with p38 MAPK inhibitor (SB203580) and STAT3 inhibitor suppressed the PEDF-induced cell proliferation from 17.6 ± 3.9% to 4.9 ± 2.5% and 3.2 ± 1.7%, respectively (Fig. 3E). The specificity of inhibitor was verified by the inhibition of phosphorylation of its respective kinases (Supporting Information Fig. S5). These data suggest that simultaneous activation of p38 MAPK and STAT3 are required for inducing LSCs in entering cell cycle.

PEDF Improves the Application of Air-Lift Culture by Retaining LSCs

We sought to examine the effects of PEDF and its 44-mer peptide on the expansion of limbal cells in the limbal equivalent. Air-lift culture was conducted in the presence of PEDF, 34-mer or 44-mer peptide as described in the Methods. As depicted in Figure 4, ABCG2, ΔNp63α and Bmi-1 labeling can only be detected in cells of the basal layer of epithelium formed in the presence of PEDF or 44-mer. Culture in basal medium or in medium supplemented with 34-mer did not lead to expression of these stem cell markers. This suggests that PEDF or 44-mer peptide may improve the quality of limbal equivalent by enriching stem cell/progenitor cell population. Importantly, prominent keratin-3 signal can be identified in the superficial cell layers of stratified epithelial cell sheets of air-lift culture that products from all the treatment conditions. This result indicates that PEDF and 44-mer-expanded limbal cells maintain their ability to differentiate into corneal cells.

Figure 4.

Airlift culture. Representative pictures of four independent experiments show immunohistochemical staining of epithelia by limbal epithelial stem cell markers (ΔNp63α, ABCG2, and bmi-1) and keratin-3. Nuclei were counterstained with Hoechst 33,258 (blue) to visualize cell layers. Original magnification, ×1,000. Abbreviations: ABCG2, ATP-binding cassette sub-family G member 2; PEDF, pigment epithelial-derived factor; UT, untreated.

44-Mer Promotes Corneal Wound-Healing

To investigate the effect of 44-mer on corneal epithelial wound healing, a 2-diameter epithelial wound was created in the center of mouse eyes by scraping, and subsequently treated the wound with 44-mer-contained ointment or a control ointment. The size of healing wound was evaluated by fluorescein staining (Fig. 5A). No significant differences in the size of the initial abrasion and wound healing for 8 hours were noted between 44-mer-treated and control mice (Fig. 5B). 44-mer treatment promoted re-epithelialization with significantly smaller epithelial defect as compared to that of the control mice at 16 hours (32 ± 9% vs. 84 ± 9%) and 24 hours (27 ± 5% vs. 67 ± 6%), respectively. By 48 hours, mice in 44-mer-treated group exhibited a complete wound closure, while wounds in the control group exhibited a 21 ± 6% defect. Collectively, 44-mer treatment accelerates corneal wound healing process.

Figure 5.

44-Mer promotes corneal wound-healing and limbal epithelial stem cell proliferation after corneal injury. (A): The 2 mm corneal wound at mouse eye was stained with topical fluorescein followed for 48 hours. (B): Histogram of residual epithelial defect (%) in mice corneas is presented as percentage of the original wound. Representative images are from three independent experiments and five mice per group. (C, D): Limbal cell proliferation at 5 hours post-corneal injury. Double immunofluorescence of sections of control ointment and 44-mer ointment-treated eye stained with 5-bromo-2′-deoxyuridine- and ΔNp63α-antibodies. ΔNp63α is expressed in the cell nucleus as confirmed by Hoechst 33,258 counterstaining. Representative images are from three independent experiments and four mice per group. *, p < .005. Original magnification, ×1,000. Abbreviations: PEDF, pigment epithelial-derived factor; BrDU, 5-bromo-2′-deoxyuridine.

Next, the possible physiologic function of PEDF on corneal wound healing was investigated. The corneal epithelial wound was treated with anti-PEDF antibody-containing ointment or a control ointment mixed with preimmune serum twice per day for 2-day. As shown in Figures 5A and 5B, anti-PEDF antibody (5 μg per treatment) significantly delayed wound healing at 24 and 48 hours compared with those of the preimmune serum treated mice, respectively (88 ± 5% vs. 69 ± 6% and 46 ± 8% vs. 24 ± 4%). This suggests that PEDF plays a physiological role in assisting cornea wound healing. Immunofluorescent staining of corneal limbal tissue revealed the accumulation of PEDF protein in ΔNp63α-positive limbal basal cells (Supporting Information Fig. S6). PEDF was also detectable in a few cells localized at suprabasal to superficial layers of the limbal epithelium. PEDF was not detectable in corneal epithelium. The accumulation of PEDF protein in limbal epithelia devoid of cornea was also evaluated by immunoblotting. The results confirmed that levels of PEDF protein in limbal epithelium were significantly higher than that in full corneal epithelium (12 ± 1.53-fold; Supporting Information Fig. S6). This indicates that LSCs provides at last in part of intrinsic PEDF for supporting corneal epithelial wound healing.

44-Mer Promotes LSC Proliferation During Corneal Wound Healing

After total removal of cornea epithelium, LSCs proliferate to produce TACs that migrate centripetally to regenerate corneal epithelium [27]. The stimulation of LSC proliferation in culture suggests a possible effect of PEDF in limbal cell proliferation in response to wounding. To investigate whether the LSC proliferation could be accelerated by 44-mer treatment, mice were intraperitoneally injected with BrdU and euthanized at 5 hours after corneal wounding. ΔNp63α and BrdU double-immunostaining of ocular sections revealed that the increases in the number of BrdU- and ΔNp63α-positive cells, or proliferating limbal progenitor cells, in the limbus of 44-mer treated eyes, was observed as early as 5 hours after wounding. Without PEDF peptides, limited number of cells was proliferating at this moment (Fig. 5C, 5D; 42 ± 6.5% vs. 2 ± 0.8%). This indicated that 44-mer PEDF advances limbal cell proliferation in response to wounding.

As it takes a period of time for the newly proliferated cells to accumulate in limbus and cornea, we investigated whether the number of TACs is affected by 44-mer at 22 hours after corneal injury. Immunofluorescence analysis confirmed that more ΔNp63α-positive cells (Fig. 6A, 6B; 33.6 ± 3.2% vs. 13.8 ± 2.6%) and BrdU-positive cells (Fig. 6C; 21.6 ± 2.1% vs. 10.8 ± 1.3%) were identified on the cornea surface in 44-mer-treated eyes than those in the control eyes, revealing the dramatic effect of 44-mer in maintaining the proliferation potential and undifferentiated cell status. More importantly, significant increase in the number of ΔNp63α-positive TACs was observed in the limbus of 44-mer-treated eyes (Fig. 6A, 6B; 63.4 ± 4.3% vs. 27.6 ± 4.6%). The rapid expansion of ΔNp63α-positive TACs in limbus and healing front by 44-mer in turn accelerates corneal epithelium healing.

Figure 6.

Immunofluorescence analysis of the levels of ΔNp63α and 5-bromo-2′-deoxyuridine-positive cells at limbus and regenerative cornea. (A–C): At 22 hours post-corneal wounding. Nuclei were visualized with Hoechst 33258 staining. Representative 44-mer panels show that 44-mer treatment stimulates expansion of ΔNp63α-positive TACs for corneal regeneration. Results of immunofluorescence were evaluated from six sections per mouse cornea, and six mice at the time point. A labeling index was computed as the number of labeled cells divided by the total number of cells with nuclei. *, p < 0.05 versus control treatment. (D):Comparative phenotypic analysis of limbus at day 4 postwounding. Hematoxylin and eosin staining (upper panels; original magnification: ×400). Lower panels: Immunofluorescence analysis showed the distribution and level of ΔNp63α-positive cells at limbus. Representative results from four separate experiments are shown. Normal indicates uninjured eye. Scale bar = 10 μM. Abbreviations: BrDu, 5-bromo-2′-deoxyuridine; H&E, hematoxylin and eosin.

At day 4 after wounding, when all cornea wounds were healed, ΔNp63α immunostaining and H&E stain of limbal tissue revealed that the epithelial thickness, total cell density and ΔNp63α-positive cell density in 44-mer-treated limbus were similar to those of the normal unwounded limbus. Whereas limbal epithelia of control ointment-treated eyes were overall thinner, and contained fewer ΔNp63α-positive LSCs, indicating the regeneration of limbal cell population was inadequate (Fig. 6D).

Taken together, the speedy cornea wound healing and accelerated rebuilt of limbal progenitor cell population indicate that 44-mer PEDF treatment effectively activates LSCs in vivo and thereby assists wound healing process.


Clinical evidence indicated that the prognosis of LSC deficiency is determined by whether there is adequate number of ΔNp63α-positive cells with high division potential in limbal equivalent [7]. Hence, the preservation of stem cells is a prerequisite for successful transplantation. In this study, we observed that not only may PEDF promote the growth of limbal cells, which is confirmed by CFE assay and air-lift culture, but it may also maintain the stemness of LSCs. PEDF is thus a specific mitogen for progenitor cells, as well as stem cells from limbus. Mitogens have been included in the medium for ex vivo expansion of limbal cells, for example, EGF is included in serum free medium for limbal cell culture [6]; hepatocyte growth factor (HGF) and keratinocyte growth factor (KGF) are used to stimulate limbal cell proliferation [20]. However, their respective ability on stem cell expansion remains to be compared by CFE assay.

It is interesting to us that PEDF at the same time increases limbal progenitor cell proliferation and maintains the expression of stem cell marker. This seems contradictory to the observation that the proliferation of progenitor cells leads to spontaneous differentiation. In our observation, PEDF does not stimulate limbal cell proliferation in the absence of cornea wounding. After wounding, cytokines are mobilized to activate reparative cell proliferation. One possible explanation is that PEDF maintains stem cell phenotype during the proliferation stimulated by other growth factor. In culture, PEDF stimulates LSC proliferation in the presence of EGF and insulin in culture medium. In our preliminary observation, PEDF treatment induces the expression of leukocyte inhibitory factor which is known to maintain the stemness of embryonic stem cells.

When limbal cells that were cultured on supporting material, are raised above fluid level, stratified epithelium starts to differentiate and forms an equivalent of limbal epithelium ready for transplantation. This limbal equivalent also provides clues as to the quality and composition of limbal expansion. When PEDF-expanded cells experiencing air-lift, they differentiate to similar degree as cells expanded in the absence of PEDF, this supports the proposition that limbal cell expansion by PEDF does not affect differential potential. One interesting observation is that there are many ΔNp63α-positive cells in limbal equivalent generated from PEDF- or 44-mer-expanded cells. This phenomenon is not observed in control limbal equivalent. It seems that PEDF expansion not only results in more progenitor cells being expanded in monolayer culture, but these cells also remain to be progenitors during air-induced corneal differentiation.

A mechanical corneal injury model is used to address the effect of 44-mer on LSC activation in vivo. We found that 44-mer enhances corneal resurfacing after corneal wounding. A coincidental finding is the proliferation of limbal cells in response to 44-mer. Although mitogenic effect of several growth factor has been reported in cell culture, this is the first observation that growth factor can stimulate cornea wound healing and amplify LSC proliferation. Interestingly, 44-mer did not induce limbal cell proliferation in the absence of cornea wounding (data not shown). Apparently, loss of corneal epithelium creates a change of microenvironment which is essential for 44-mer to stimulate LSCs and TACs to enter cell cycle. These observations suggest 44-mer may be a potential therapeutics for human ocular surface diseases.

The signaling mechanism of LSC proliferation induced by PEDF or its 44-mer peptide seems to involve p38 MAPK pathway and STAT3 pathway. The induction of p38 MAPK is shared by other limbal cell mitogen such as KGF and HGF [20]. The attenuation of the mitogenic activity of PEDF by specific inhibitor of p38 MAPK further confirm the role of p38 MAPK in LSC proliferation induced by PEDF or its 44-mer peptide. Ex vivo expansion of human LSCs on intact AM is in part supported by ERK-mediated survival signaling, not p38 MAPK [28]. This difference of signaling mechanism may come from difference of cell culture environment, particularly the involvement of cells and supporting matrix. This suggests that different mechanism may be involved in different ex vivo expansion support. STAT3 signaling has been shown to mediate the expansion of limbal stem cell by IL-6 [29]. Our results show that simultaneous activation of p38 MAPK and STAT3 are required for PEDF-induced cell proliferation. Moreover, by examining stem cell markers and cornea epithelial cell differentiation markers, we conclude that PEDF supports the maintenance of stem cell phenotype in progeny after cell proliferation. The mutually dependent interplay of p38 MAPK and STAT3 in arresting corneal epithelial differentiation and maintaining LSC stemness remain to be determined. Cell culture in hypoxic condition benefits expansion of LSC population [30]. It is proposed that normoxia is toxic to stem cell in culture. 44-mer peptide of PEDF protects neuronal cell from oxidative injury through activating AKT pathway [31]. It is conceivable that the expansion of LSCs requires both mitogenic signaling mediated through the p38 MAPK/STAT3 and a stem cell protection pathway. Capability in activating both pathways allows PEDF or its 44-mer peptide to act as in vivo inducer of LSC expansion, which renders them potential candidates for the development of medicaments useful for the treatment of diseases or conditions related to LSCs deficiency.


We have shown that PEDF and a 44-mer PEDF peptide may respectively enhance proliferation while suppresses spontaneous differentiation of LSCs in cultures. This may significantly improve the quality of limbal equivalent and therefore on the corneal regenerative medicine. The direct stimulation of the proliferation of limbal progenitor cells and the ensued acceleration of corneal reepthelialization suggest PEDF peptide derivative as potential agent to treat LSC deficiency-related disorders.


We thank support from Chu-Ping Ho and Chin-Min Wang for assistance with animal experiments. This study was supported by grants from the National Science Council, Taiwan (NSC 101-2314-B-195-006-MY3) and Mackay Memorial Hospital (MMH-E-101-006).

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.