The Anti-Inflammatory and Anti-Angiogenic Role of Mesenchymal Stem Cells in Corneal Wound Healing Following Chemical Injury

Abstract

To investigate the anti-inflammatory and anti-angiogenic effects of mesenchymal stem cells (MSC) in the chemically burned corneas, we mechanically removed the corneal epithelium of rats after 100% alcohol instillation. The rats were then randomized into four groups: fresh media, conditioned media derived from the MSC culture (MSC-CM), MSC applied topically to the damaged corneas for 2 hours immediately after the injury or MSC-CM applied either once or 3 times per day for 3 consecutive days. Corneal surface was evaluated every week. After 3 weeks, the corneas were stained with the hematoxylin-eosin, and the expression of interleukin (IL)-2, interferon (IFN)-γ, IL-6, IL-10, transforming growth factor (TGF)-β1, thrombospondin-1 (TSP-1), matrix metalloproteinase-2 (MMP-2), and vascular endothelial growth factor (VEGF) were analyzed. CD4⁺ cells were assessed in the corneas. We found that both MSC and three-time applied MSC-CM (1) reduced corneal inflammation and neovascularization, (2) decreased IL-2 and IFN-γ, although increased IL-10 and TGF-β1 as well as IL-6, (3) reduced the infiltration of CD4⁺ cells, and (4) upregulated the expression of TSP-1, although downregulated that of MMP-2. Interestingly, whereas three-time application of MSC-CM was partially effective, transplantation of MSC achieved a better outcome in suppressing corneal inflammation. The results of this study suggest that the anti-inflammatory and anti-angiogenic action of MSC in the chemically burned corneas might be mediated in part through paracrine pathways involving soluble factors such as IL-10, TGF-β1, IL-6 and TSP-1.

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

Angiogenesis, Cornea, Inflammation, Mesenchymal stem cell

Introduction

Limbal stem cell deficiency (LSCD), resulting from various clinical disorders such as a chemical burn causing severe inflammation, aniridia, or Stevens-Johnson syndrome, is accompanied by persistent epithelial defects, stromal opacity, or neovascularization, thereby leading to permanent vision loss. The current treatment options are prompt anti-inflammatory drug therapy in the early phase of the disease and provision of limbal stem cells (LSC) in the late stage after inflammation has subsided. However, the anti-inflammatory drugs currently available are not sufficient to inhibit neovascularization and conjunctivalization, and transplantation of LSC has its limitations such as the low availability of LSC and the high immunorejection rates. The lack of an effective treatment and the potential severity of the disease causing blindness emphasizes the importance of novel therapeutic strategies for LSCD.

Mesenchymal stem cells (MSC) can differentiate into muscle, brain, liver, cartilage, bone, fat, and the vessels [1]. They also display immunoregulatory properties [2, 3], which are important for their proposed consideration as a potential therapeutic option for limbal stem cell deficiency (LSCD). Indeed, two reports previously described the favorable effects of MSC on the reconstruction of the chemically injured corneas [4, 5]. However, little is known about the anti-inflammatory properties of implanted MSC in the cornea, not to mention the mechanisms mediating such effects.

Accordingly, the purpose of this study was (1) to investigate whether transplantation of MSC improves the ocular surface reconstruction by modulating corneal inflammation and angiogenesis in rats with a chemical burn, which is one of the main causes of LSCD, and (2) to explore the underlying mechanisms responsible for such effects. We herein focused on the anti-inflammatory capacity of MSC and its mechanism of action in a chemical burn model of the rat cornea.

Materials and Methods

Animals

Six-week-old male Sprague-Dawley rats weighing 200–250 g (Orient Bio Inc., Seongnam, Korea, http://www.orient.co.kr) were used in all experiments. The experimental protocols were approved by the Institutional Review Board and Institutional Animal Care and Use Committee of the Seoul National University Hospital.

Preparation of MSC

Rat primary mesenchymal stem cell lines were purchased from Millipore (cryopreserved rat mesenchymal stem cells, Part Number 2004005; Temecula, CA, http://www.millipore.com) and thawed according to the manufacturer's instructions. Cells were resuspended in mesenchymal stem cell expansion media (Millipore). The cells were plated in a flask at a density of 1 × 10⁶ cells/cm² and incubated at 37°C with 5% CO₂. The media was exchanged every two days, and after 10–14 days, the nonadherent cells and supernatant were discarded. Adherent cells were treated with Accutase (Millipore) for three minutes and replated in a new flask at 1 × 10⁶ cells/cm². Adherent cells obtained after the second subculture, which was the third passage of cells, were used for the experiment. On the day of transplantation, MSC were labeled with PKH 26 (Sigma, St. Louis, http://www.sigmaaldrich.com) following the manufacturer's protocol, and washed twice in fresh media.

Animal Model

Rats were anesthetized with an intramuscular injection of 10 mg/kg zolazepam and 10 mg/kg xylazine hydrochloride to create the chemical burn and to apply the treatments mentioned below. To create the chemical burn, a 6-mm filter paper soaked in 100% ethanol was applied to the whole cornea including the limbus for 30 seconds followed by a rinse of 10-ml of balanced salt solution, and then the corneal and limbal epithelium were removed with a surgical blade.

Application of MSC or MSC Culture Conditioned Media

Forty rats were randomly divided into four groups. A 6-mm-diameter hollow plastic tube, which was customized as an applicator, was put on the cornea in order to keep the eye open (Fig. 1A). Next, media or cells were put into the customized applicator and allowed to remain in the damaged cornea for two hours. The following four groups were studied with different treatments applied by the customized applicator: (1) 200 μl of new media containing DMEM supplemented with 10% FBS, 100 IU/ml penicillin, 100 μg/ml streptomycin was put into the tube (control group, n = 10), (2) 200 μl supernatants collected from the MSC culture (MSC-conditioned media; MSC-CM) (MSC-CM I group, n = 10), (3) 200 μl MSC-CM applied for 2 hours per day over 3 consecutive days (MSC-CM II group, n = 10), (4) 200 μl media containing 2 × 10⁶ MSC (MSC group, n = 10). After application, the eyelids were tightly sutured for rats not to blink in order to prevent possible mechanical shedding of the transplanted cells.

Figure 1.

A 6-mm-diameter hollow plastic tube (A) was emplaced to keep the eye open and the cells or media were directly applied to the cornea into the customized applicator. (B) MSC engraftment in the wounded cornea three weeks after topical application. PKH26-labeled MSC were observed beneath the overlying corneal epithelium with red fluorescein. Magnification, ×200.

Ocular Surface Evaluation and Clinical Outcome Analysis

Each eye underwent slit-lamp examination and photography once a week after the injury. Corneal epithelial integrity, clarity, and new vessels (NV) were assessed and graded using the slit-lamp by a blinded investigator who was an ophthalmologist. A fluorescein sodium solution was used to evaluate the degree of the corneal epithelial defects. The corneal epithelial integrity was quantified by the ratio of the epithelial defect area to the total corneal area using an image analyzer (Image Pro Plus, Media Cybernetics, Bethesda, MD, http://www.mediacy.com). The corneal clarity was graded as follows: grade 0, totally clear with no opacity seen by any method of slit-lamp examination; grade 1, haze of minimal density seen with difficulty on direct and diffuse illumination; grade 2, mild haze easily visible on direct focal illumination; grade 3, moderately dense opacity that partially obscured the iris details; and grade 4, severely dense opacity that completely obscured the details of the intraocular structures [6]. The corneal NV were quantified by calculating the wedge-shaped area of vessel growth with the following formula: A = C/12 × 3.1416 [r² − (r − l)²], where A is the area, C is time (in hours), l is the radius from the center to the border of vessel growth, and r is the radius of the cornea [7]. The degree of NV was compared between the groups using the ratio of the NV area to the whole corneal area.

Histolopathology

Three weeks after the transplantation the corneas were excised after the rats were sacrificed. Portions of the cornea were sectioned and stained with hematoxylin-eosin or subjected to immunofluorescent study. Inflammatory cells were counted in five randomly selected fields (×200) of the hematoxylin-eosin-stained slides. The immunofluorescent staining was performed to evaluate the presence of CD4⁺ cells in the corneas. The obtained cornea was fixed in 10% neutral buffered formalin and incubated overnight at 4°C. It was cut to 4-μm thickness and dried at 60°C for 1 hour and deparaffinized with ethanol. Proteinase K (20 μg/ml; Sigma-Aldrich), 3% H₂O₂, and 0.3% Triton X-100 were used for serial treatments, and then 1% serum was added. The monoclonal mouse antibodies against rat CD4 1:100 (Millipore) were used as primary antibodies and phosphate buffered saline as negative controls. The anti-rabbit IgG, 1:5000 (SouthernBiotech, Birmingham, AL, http://www.southernbiotech.com), anti-streptavidn 1:5000 (USBiological, Swampscott, MA, http://www.usbio.net), and anti-mouse IgG 1:5000 (Sigma) were used as secondary antibodies. The chromophore of Alexa 488, 1:500 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and Hoechest 33,342 (Sigma) were used for counterstaining. The numbers of CD4-stained cells were counted in five randomly selected fields (×400) under confocal microscopy (LSM 510 Meta, Zeiss, Yena, Germany, http://www.zeiss.de).

Cytokine Quantification by an Enzyme Linked Immunosorbent Assay

Protein for interleukin IL-2, interferon (IFN)-γ, IL-6, IL-10, and transforming growth factor (TGF)-β1 in the corneas was determined using an Enzyme Linked Immunosorbent Assay (ELISA) kit (R&D Systems, Minneapolis, http://www.rndsystems.com). Corneas excised from the eyeballs were sliced with scissors into small pieces, and the tissue was homogenized with a sonicator (Tomy-Seiko, Tokyo, http://bio.tomys.co.jp). The samples were lysed in 350 μl of extraction buffer (100 nM/L Tris HCL PH 8.0 and 10 mM/L lice) and were centrifuged at 2,500 rpm for 20 minutes. An aliquot of each supernatant was assayed. They were diluted into 10 ng/well in 0.05 M carbonate bicarbonate buffer and incubated overnight at 4°C. The primary monoclonal mouse antibodies against rat IL-2, IFN-γ, IL-6, IL-10 and TGF-β1 (R&D systems, Minneapolis) and the secondary antibodies of HRP-goat anti-mouse IgG conjugate (1:1000; AbD Serotec, Oxford, U.K., http://www.serotec.com) were used. The concentrations of each cytokine were measured using an ELISA kit (Spectra MAX 250, Molecular Devices Corporation, Sunnydale, CA, http://www.moleculardevices.com) at 490 nm. Normal healthy corneas were used for comparison of the baseline activity. ELISA experiments were performed in triplicate to ensure the reproducibility of the data.

Quantitative Real-Time Polymerase Chain Reaction

The expression of thrombospondin-1 (TSP-1), metalloproteinase-2 (MMP-2), and vascular endothelial growth factor (VEGF) in the corneas was evaluated at the mRNA level by real-time polymerase chain reaction (Real-Time PCR) assays. Total RNA was extracted from the harvested corneas with an extraction reagent (TRIzol; Gibco, Grand Island, NY, http://www.invitrogen.com), following the manufacturer's instructions. The total RNA was dissolved in water treated with diethylpyrocarbonate (DEPC), and the concentration was measured by a spectrophotometer (UV-1601; Shimadzu, Kyoto, Japan, http://www.shimadzu.com). The samples were treated with DNase I (0.2 U/μl; Ambion, Austin, TX, http://www.ambion.com) to remove possible DNA contamination. For the SDS 7,000 system reactions, a master mix of the following components was prepared with the indicated end concentration of 2.5 μl water, 2.5 μl forward primer (9 μM) and reverse primer (9 μM), 2.5 μl probe (2.5 μM), 12.5 μl TaqMan PCR 2x master mixture (Applied Biosystems, Lincoln, CA, http://www.appliedbiosystems.com). Seventy-five ng of reverse transcribed total RNA in 5 μl was added as PCR template. Relative quantitative real-time PCR on 96-well optical plates was performed using the reagents and analyzed on an ABI Prism 7,000 Sequence Detection System (Applied Biosystems). The following PCR conditions were used: after initial activation of uracyl-N-glycosylase at 50°C for 2 minutes, AmpliTaq Gold (Applied Biosystems) was activated at 95°C for 10 minutes, the subsequent PCR condition consisted of 45 cycles of denaturation at 95°C for 15-second and annealing extention at 60°C for 1 minute per cycle. During the PCR amplification, the amplified products were measured continuously by determination of the fluorescence emission. The expression level of target gene was normalized to internal Rat GAPDH and represented as relative expression [8]. The PCR probe sets used are as follows: GCAACCGCATTCCAGAGTCTGGGGG (Rn01513689_g1) for TSP-1; GGTTTATTTGGCGGACAGTGACACC (Rn02532334_s1) for MMP-2; and CCTCCACCATGCCAAGTGGTCCCAG (Rn00582935_m1) for VEGF. To confirm a constant housekeeping gene expression level in the investigated total RNA extractions, a GAPDH real-time PCR was also performed. Real-time PCR was qualified in the SDS 7,000 (Applied Biosystems) with the Rat GAPD (GAPDH) Endogenous Control (RefSeq NM_017008.3; P.N 435,2338E) (VIC/MGB Probe, Primer Limited).

Statistical Analysis

Data were expressed as mean ± standard error. Comparisons of parameters among the groups were made by the Student's t-test or nonparametric Mann-Whitney test using SPSS software (SPSS 12.0. Chicago, http://www.spss.com). Differences were considered significant at p < .05.

Results

Engraftment of MSC in the Cornea

We confirmed that MSC were successfully engrafted underneath the epithelial cell layers up to 3 weeks after topical application by identification of PKH26-labeled cells in all MSC-treated corneas by fluorescein microscopy (Fig. 1B).

Clinical Improvement in the Corneal Surface by MSC

To determine the effects of the MSC or MSC-CM on the cornea after the chemical burn, we evaluated the corneal surface based on the findings of transparency and neovascularization. At 1 week post-injury, the groups did not show any differences with massive peripheral new vessels and opaque corneal surfaces in all groups (Fig. 2). However, the degree of corneal opacity and neovascularization rapidly decreased in the MSC-treated cornea as time passed after the injury, whereas they gradually increased in the vehicle-treated control. The degree of corneal opacity and neovascularization were the lowest in the MSC group and the highest in the control group at three weeks post-injury. For the MSC-CM-treated corneas, the effects on the corneal surface reconstruction were proportional to the number of MSC-CM application; that is, corneas treated with MSC-CM 3 times (MSC-CM II group) were more transparent with less neovascularization than were the controls. However, the corneas treated with MSC-CM only once (MSC-CM I group) did not differ from the control. When the MSC group and MSC-CM II group were compared, the MSC-treated corneas achieved significantly better results at three weeks post-injury (Fig. 3A, B). Re-epithelialization occurred faster in the MSC and MSC-CM II groups, but the differences were not statistically significant (Fig. 3C).

Figure 2.

Photography of cornea one (A–D), two (E–H), three weeks (I–L) post-injury. With time, neovascularization and opacity markedly decreased in the corneas with MSC (D, H, L) or MSC-conditioned media three times (C, G, K), whereas increased in the control (A, E, I). Corneas treated with MSC-conditioned media once (B, F, J) showed the intermediate outcome.

Figure 3.

Effects of MSC on cornea. Neovascularization (A), opacity (B), and epithelial defect (C) decreased in the MSC- or MSC-conditioned media (MSC-CM)-treated corneas with time, whereas increased in the control. Clinical outcome was the best in the MSC-treated corneas. Corneas treated with MSC-CM three times (MSC-CM II) achieved better outcomes than those with MSC-CM once (MSC-CM I).

Attenuation of Inflammation by MSC on Histopathology

Similar to the clinical findings, the histological analysis of the corneal sections revealed that the MSC and MSC-CM II groups had fewer inflammatory cell infiltrates than did the control and MSC-CM I groups (Fig. 4) (p < .001 in control versus MSC-CM II; p < .001 in control versus MSC; p = .032 in MSC-CM I versus MSC-CM II; p = .003 in MSC-CM I versus MSC). Comparisons between the MSC and MSC-CM II groups showed that the corneas treated with MSC had significantly fewer inflammatory cells than did those treated with application of MSC-CM 3 times (p = .043). On the immunofluorescent staining, fewer CD4⁺ cells were detected in the MSC-treated corneas, although statistically insignificant, compared to those treated with MSC-CM or fresh media as a vehicle (Fig. 5).

Figure 4.

Hematoxylin-eosin staining of cornea three weeks post-injury. Control (A) and corneas treated with MSC-conditioned media once (MSC-CM I, B) were densely infiltrated with inflammatory cells in the stroma and goblet cells in the epithelium. The infiltration was markedly reduced in the corneas with MSC-conditioned media three times (MSC-CM II, C) or MSC (D). Magnification, ×200.

Figure 5.

CD4 Immunohistochemistry of cornea three weeks post-injury. Immunofluorescent confocal images showed that more CD4⁺ cells were infiltrated in the control corneas (A), than those treated with MSC (D), MSC-conditioned media once (B), or three times (C). CD4 presented as green and the nuclei were counterstained as blue. Arrows indicate positive-staining cells. Magnification, ×400.

Modulation of Inflammatory Cytokines by MSC

In addition to the histology, anti-inflammatory effects were further studied with regard to inflammatory cytokine production. ELISA analysis showed that the production of the immunostimulatory cytokines IL-2 and IFN-γ was repressed both in the MSC- and MSC-CM-treated corneas (Fig. 6) (p = .019 in control versus MSC-CM II for both IL-2 and IFN-γ; p = .013 and 0.009 in control versus MSC for IL-2 and IFN-γ, respectively; p = .045 in MSC-CM I versus MSC-CM II for IL-2; p = .031 and 0.041 in MSC-CM I versus MSC for IL-2 and IFN-γ, respectively). The levels of IL-2 and IFN-γ were proportional to the number of MSC-CM applications for the MSC-CM groups; that is, IL-2 and IFN-γ levels were significantly lower in corneas with MSC-CM applied three times (MSC-CM II), compared to the vehicle-treated control or corneas treated with MSC-CM only once (MSC-CM I). There were no significant differences between the control and MSC-CM I group.

Figure 6.

Cytokine expression evaluated by ELISA. IL-2 (A) and IFN-γ (B) were repressed in the corneas treated with MSC or MSC-conditioned media three times (MSC-CM II), compared to the control, whereas IL-10 (C) and IL-6 (E) increased in the MSC and MSC-CM II group. TGF-β1 (D) was high only in the MSC group.

To investigate the possible mechanism involved in suppression of inflammation with the MSC or MSC-CM-treated corneas, we next assessed the production of anti-inflammatory cytokines IL-10 and TGF-β. We found large quantities of IL-10 and TGF-β in MSC-treated corneas (Fig. 6) (p < .001 and p = .001 in control versus MSC for IL-10 and TGF-β, respectively; p < .001 and p = .002 in MSC-CM I versus MSC for IL-10 and TGF-β, respectively; p = .008 and p < .001 in MSC-CM II versus MSC for IL-10 and TGF-β, respectively). The level of IL-10 was also higher in the corneas of the MSC-CM II group, compared to the control or MSC-CM I groups (Fig. 6) (p = .017 in control versus MSC-CM II; p = .077 in MSC-CM I versus MSC-CM II). We then investigated the expression of IL-6 to determine the possible mechanism of anti-inflammatory cytokine induction by MSC, because IL-6, produced by MSC, is known to mediate the immunomodulatory action of MSC in vitro. The level of IL-6 was slightly, but significantly and consistently, high in both the MSC-and MSC-CM-treated corneas, whereas its expression in the vehicle-treated control was as low as in the normal corneas without a previous injury (p = .001 in control versus MSC and control versus MSC-CM I; p < .001 in control versus MSC).

Effect of MSC on Corneal Angiogenesis

We next tried to figure out the mechanism associated with the more rapid regression of new vessels after MSC and MSC-CM application. The expression of TSP-1, MMP-2, and VEGF, known to be involved in chemical-burn-induced corneal angiogenesis, were evaluated. The level of TSP-1 was significantly upregulated at the mRNA level in the MSC and MSC-CM II groups, compared to the control (Fig. 7) (p = .01 in control versus MSC-CM II; p = .04 in control versus MSC; p = .01 in MSC-CM I versus MSC-CM II; p = .03 in MSC-CM I versus MSC). MMP-2 was expressed very high in the corneas treated with a vehicle or MSC-CM, whereas it was significantly repressed in the MSC-treated corneas (Fig. 7) (p = .01 in control versus MSC-CM II; p = .04 in control versus MSC; p = .01 in MSC-CM I versus MSC-CM II; p = .03 in MSC-CM I versus MSC). The expression of VEGF was high in all groups, showing no significant differences among the groups (Fig. 7) (p = .023 in control versus MSC; p = .034 in MSC-CM I versus MSC; p = .004 in MSC-CM II versus MSC).

Figure 7.

Real-time PCR. Upregulation of TSP-1 (A) was observed in the corneas treated with MSC or MSC-conditioned media three times (MSC-CM II), compared to the control. MMP-2 was downregulated in the MSC group (B). There were no differences in the expression of VEGF (C). Values were expressed as folds relative to fresh corneas without an injury.

Discussion

In the cornea, two previous studies have evaluated the role of MSC demonstrating that MSC help to reconstruct the ocular surface damaged by a chemical injury [4, 5]. However, the exact mechanism of those effects has never been focused in those studies. We believe that an effective inhibition of inflammation during the acute phase is mandatory for the prognosis of LSCD. That is why we investigated the anti-inflammatory activity of MSC for LSCD using a chemical burn model.

Many reports have already shown that MSC have immunoregulatory properties in various diseases. They have been shown to: (1) promote the survival of skin grafts and reduce the incidence of graft-versus-host disease (GVHD) [9, 10, 11]; (2) ameliorate experimental autoimmune encephalomyelitis [3]; (3) reduce myocardial inflammation [12, 13]; and (4) decrease the inflammation in bleomycin-induced lung injury [14, 15]. In our study, we found the anti-inflammatory property and therapeutic potentials of MSC for restoration of the pre-existing limbal stem cell niche in the acute phase of a chemical burn, where inflammation actively occurred, thereby resulting in corneal opacity and neovascularization on the loss of LSC. The main observation of this study was that (1) topical application of MSC or multiple applications of MSC-derived conditioned media contributed to restoration of anti-angiogenic and anti-inflammatory property of the cornea, as determined by corneal transparency and attenuated neovascularization, and (2) MSC increased the level of IL-10, TGF-β, IL-6, and TSP-1, which reduced infiltration of inflammatory cells and the expression of IL-2, IFN-γ, and MMP-2 in the chemically burned cornea.

To date, the mechanism of MSC-mediated immunosuppression has not been fully explained. MSC have been shown to: (1) inhibit the function of mature, naïve and memory T-cells [16, 17]; (2) decrease tumor necrosis factor secretion and increase IL-10 secretion from dendritic cells (DC) [2, 18, 19]; (3) increase the proportion of regulatory T-cells [2, 20]; (4) block the antigen-presenting cells (APC) maturation [2, 18, 21, 22, 23], and these immature APC, in turn, secrete immunosuppressive cytokine IL-10 [18, 23]. Consistent with these in vitro findings, our in vivo study showed that the application of MSC increased IL-10 and TGF-β, and reduced IL-2 and IFN-γ in the damaged cornea, which was accompanied by significantly improved clinical outcomes.

Meanwhile, in our in vivo setting, we found an increment in IL-6 in MSC- and MSC-CM-treated corneas. IL-6 is known to have dual action in inflammation. On the one hand, IL-6 acts as a pro-inflammatory cytokine. It is known to be induced in the alkali-burned cornea with a peak production 7 days after injury [24] and leads to infiltration by neutrophils, which may cause corneal stromal melting [25]. On the other hand, IL-6 is an important immunoregulatory cytokine which controls the development of APC [26, 27]. In fact, one study demonstrated that MSC inhibited differentiation of monocytes to DC [15]. In addition, a recent study showed that MSC secreted higher level of IL-6, interfering with maturation of DC [23]. We also found that MSC, upon coculture with damaged corneal epithelial cells, secreted a remarkably high level of IL-6 (around 1,500 pg/ml), whereas MSC or damaged epithelial cells alone produced only a minimal amount of IL-6 (less than 50 pg/ml) (supplemental online Fig. 1). Taken together with our in vivo finding, it is plausibly supposed that the exogenous high concentration of IL-6 might paradoxically suppress DC not to secrete the endogenous IL-6 as a biofeedback. Subsequently, we can speculate that MSC might interfere with maturation of DC by the hypothesis mentioned above and those immature DC might produce IL-10, which would reduce the corneal inflammation.

As far as neovascularization is concerned, we found rapid regression of new vessels in the MSC- or MSC-CM II-group, which is consistent with the previous studies using the cornea [4, 5]. This is in contrast to many studies that reported strong angiogenic activities of MSC in various organs [28] other than the cornea. To determine the anti-angiogenic mechanism of MSC in the cornea, we analyzed the expression of molecules known to be involved in corneal angiogenesis. As a result, we observed that TSP-1, which is a powerful anti-angiogenic factor [29], was upregulated at high levels in response to an injury in MSC or MSC-CM II group, whereas MMP-2, which is an inflammation-related pro-angiogenic factor [29], was significantly downregulated upon MSC treatment. Ma Y et al. also observed the reduced expression of MMP-2 [4]. In contrast with an ischemia or tumor model where MSC secrete VEGF [30, 31], the level of VEGF was not different between the control and MSC-group. This might be due to high-level of TSP-1 which has a modulatory effect on VEGF. TSP-1 is known to reduce angiogenesis by promoting apoptosis in the endothelial cells [32], inhibiting VEGF [33], or down-regulate MMPs [34]. Moreover, our unpublished data showed that cultured MSC secreted remarkably high-level of TSP-1. Taken together, it can be supposed that TSP-1, produced by MSC, might reduce the level of VEGF and MMP2 to restore anti-angiogenic privilege of the cornea.

In addition, we also studied a MSC-CM-treated group to evaluate the mechanism of MSC in paracrine manners in vivo. Previous in vitro reports have shown that both inhibition via cell-to-cell contact [18, 35] as well as the activity of soluble factors such as TGF-β1 [19] and IL-6 [23, 36] are involved in the immunoregulatory mechanism of MSC. Interestingly, our findings show that MSC-CM had anti-inflammatory properties only when it was applied more than two times. Although three-time application of MSC-CM was not as effective as transplantation of MSC themselves, it was more effective than the vehicle-treated control or one-time application of MSC-CM. These results suggest the following two explanations. One is that cell-to-cell contact may exert additive effects on anti-inflammation by MSC in addition to their action by soluble factors. Another is that the continuous secretion of cytokines from MSC throughout 3 weeks might lead to better outcomes compared with the three-time intermittent administration of cytokines. In other words, MSC might act as a cellular source for supplying soluble factors. Notably, another immunosuppressive cytokine TGF-β1 was increased only in the MSC-treated corneas. This may explain why the MSC transplantation was more effective in controlling inflammation than the MSC-CM application in this study.

Unlike the previous studies, which used i.v. administration or amniotic membrane as a carrier, we applied MSC topically. This approach could be enabled by the MSC's property to adhere rapidly on the surface. Moreover, we sutured the eyelids of rats and maintained as they were to keep rats from blinking, thereby to prevent the mechanical shedding of MSC from the corneal surface. As it turned out, re-epithelialization happened over the MSC layers during three weeks, as confirmed by the existence of PKH26-labeled MSC beneath the overlying epithelium. It is conceivable that topical application is much easier and less invasive than aforementioned methods. Moreover, we found that our method achieved the outcomes as comparable as the methods used in previous studies. This finding would highlight the potential therapeutic usage of our method in human, considering that suturing lids in rats could be substituted by the insertion of the contact lens in human.

Our study had limitations as follows. First, to confirm IL-6, IL-10, TGF-beta, or TSP-1 as mediators for the MSC action, we have to either block or directly add them to see the opposite or the same effects. However, there are currently no proven-to-be effective blocking antibodies against them that can be used in vivo as well as are capable of exerting effects in the cornea when applied either systemically or topically. Moreover, it is highly likely that MSC would act through more complex mechanisms of the cells and cytokines in addition to IL-10, TGF-beta, and TSP -1, making them just a part of the whole immunomodulatory mechanism. Either way, further study utilizing the knockout animal would be required to clarify the roles of IL-6, IL-10, TGF-beta, or TSP-1 in the action of MSC. Second, we could not tell the fate of the transplanted MSC. We do not know whether the transplanted MSC differentiated into functional cells that replaced the damaged cells in the tissue or remained as they were. Instead, we found MSC failed to transdifferentiate into corneal epithelial cells in one of our initial experiments (unpublished data) which also corresponds to a previous report [4]. The mechanism by which MSC display favorable activity in the present study could be that MSC facilitated the function of remnant corneal epithelial cells by restoring the pre-existing limbal stem cell niche in the cornea; or, they might differentiate into keratocytes (corneal fibroblasts) to support the favorable milieu for the restoration of limbal stem niche. We cannot check the differentiation of MSC into keratocytes, because keratocytes do not have their specific markers and they do share many markers with MSC. However, there are some suggestions that MSC might differentiate into keratocytes [5]. Further study would be needed to fully clarify the fate of MSC in the cornea.

Acknowledgements

We thank Dr. Curie Ahn of the Department of Internal Medicine, Seoul National University, Seoul, Korea, and Dr. Ki-Ho Park, of the Quantitative Real-Time Polymerase Chain Reaction Lab, Seoul National University Hospital Clinical Research Institute, Seoul, Korea. This study was supported by the Seoul National University Hospital research fund (04-2006-012 and 04-2007-071).

Conclusion

Taken together, whereas a definitive explanation for the mechanism requires further investigation, the results of our study demonstrate the beneficial effects of topical MSC application on restoration of the stem cell niche in the damaged cornea with its anti-inflammatory and anti-angiogenic activity in a chemical burn model of the rat cornea. The paracrine effects, involving IL-10, TGF-β, IL-6, and TSP-1, may be one of the mechanisms by which MSC display such activities.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.