Emodin Suppression of Ocular Surface Inflammatory Reaction

Abstract

purpose. To determine whether a Chinese herbal medicine component, emodin, suppresses inflammatory/fibrogenic reaction in cultured subconjunctival fibroblasts and reduces injury-induced increases in ocular surface inflammation in mice.

methods. Effects of emodin were measured in human subconjunctival fibroblasts on proliferation and migration with colorimetry and scratch wound assay, respectively. Neovascularization was evaluated using an endothelial cell-fibroblast coculture model. Proinflammatory mediator and extracellular matrix component gene and protein expression was characterized with real-time reverse transcription-polymerase chain reaction, enzyme immunoassay, and immunocytochemistry, respectively. Western blotting and immunohistochemistry evaluated the activation of nuclear factor-κB (NF-κB) and c-Jun N-terminal kinase (JNK). In a mouse corneal alkali-burn model, the effects of emodin on ocular surface inflammation and fibrosis were evaluated.

results. Emodin suppressed tumor necrosis factor α (TNF-α)-induced fibroblast migration and fibronectin deposition in vitro. VEGF induced neovascularization but did not affect cell proliferation and collagen type 1 production. Monocyte/macrophage-chemoattractant protein-1 gene and protein expression declined. Emodin inhibited TNF-α-induced NF-κB p65 and JNK activation but did not affect transforming growth factor β1-induced Smad2/3 signaling. In vivo, emodin inhibited proinflammatory and fibrogenic reactions.

conclusions. Emodin suppressed in vitro TNF-α-induced stimulation of proinflammatory reaction. In a mouse ocular alkali burn model, this herbal component lessened inflammation and scarring. Additional studies are warranted to evaluate the therapeutic potential of emodin in lessening ocular tissue inflammation and resultant fibrosis after injury.

Injury-induced fibroblast phenotype transformation into myofibroblasts is responsible for systemic or local fibrogenic disorders in various tissues. Some of the resultant changes include increases in expression of the profibrogenic cytokines transforming growth factor β (TGF-β), tumor necrosis factor α (TNF-α), and extracellular matrix components. ^{1 2} Proinflammatory cytokines expressed by fibroblasts, such as monocyte/macrophage-chemoattractant protein-1 (MCP-1), act as chemoattractants to induce inflammatory cell infiltration. In ocular tissues, inflammatory fibrogenic diseases include vernal or atopic conjunctivitis and Stevens-Johnson syndrome. ^{3 4} These conditions arise because of subconjunctival fibroblast activation resulting in inflammation, ocular surface fibrogenesis, and loss of ocular surface integrity. Ultimately, these pathologic changes can produce visual impairment. 

A Chinese herbal medicine, Inchin-ko-to, has therapeutic effects on liver fibrosis and cholestatic liver diseases. ^{5 6 7} This herbal medicine contains three components, Inchin-ko (Artemisiae Capillari Spica), San-shishi (Gardeniae Fructus), and Dai-ou (Rhei Rhizoma). Inchin-ko contains an iridoid-glyco compound, geniposide, which is converted to genipin by bacterial enzymes in the digestive organs and then is absorbed into the blood. ⁸ In the liver, genipin has an antifibrotic effect that is explained by its suppression of fibrogenic effects in hepatic stellate cells. Dai-ou, on the other hand, contains a bioactive constituent, emodin (C₁₅H₁₀O₅; 1,3,8-Tri-hydroxy-6-methyl-anthra-quinone; MW 270.24). Emodin also suppresses myofibroblast generation and collagen production in hepatic stellate cells but has variable effects on cell proliferation in other tissues. ^{9 10 11 12 13 14 15} Although genipin suppresses fibrotic behaviors of a lens epithelial cell line, there are no reports regarding whether emodin has any antifibrogenic or anti-inflammatory effects on ocular tissues in vitro or in vivo. ¹⁶  

We first examined in mouse subconjunctival fibroblasts the effects of emodin on fibrogenic mediators inducing proliferation and migration. In a mouse ocular surface (including cornea and conjunctiva) alkali-burn model, we determined the possible therapeutic effect of systemic emodin administration on reducing ocular surface inflammation and fibrosis (vernal conjunctivitis, Stevens-Johnson syndrome occurs, or excess fibrosis) in the filtering bleb after glaucoma surgery. 

Materials and Methods

Primary Cell Culture

Human subconjunctival fibroblasts were cultured as reported. ¹⁷ In brief, subconjunctival tissue was obtained during strabismus surgery with informed consent from the patients’ parents. The cells were cultured for 2 or 3 passages in Eagle minimum essential medium (MEM; Gibco, Grand Island, NY) supplemented with antibiotics, an antimycotic, and 10% fetal calf serum (MEM-10) before the subsequent experiments. An ethanol-containing emodin stock solution was diluted to obtain final concentrations of 1, 2.5, and 5 μg/mL in serum-free culture medium. The final ethanol concentration was 0.1% in each experiment. Emodin effects were determined in a serum-free condition with or without the proinflammatory cytokine, recombinant human TNF-α, at 10 ng/mL (R&D Systems, Minneapolis, MN). 

Evaluation of Cytotoxicity of Emodin

A commercial kit probed for a cytotoxic effect of emodin by measuring the release of nuclear matrix protein (NMP) 41/7 ¹⁸ (Cell Death Detection [Nuclear Matrix Protein] ELISA Kit; Oncogene Research Products, San Diego, CA). The cells were treated with emodin (natural, extracted type, 0–5.0 μg/mL; Sigma, St. Louis, MO) for 48 hours in MEM-10. Culture medium was harvested and processed for the assay according to the manufacturer’s protocol. In brief, 100 μL medium or NMP41/7 standards (0–1021 U/mL) were added to each well of a 96-well plastic plate and left for 90 minutes at room temperature. After washing three times, anti-NMP41/7 antibody reacted for 60 minutes. After another wash, peroxidase-conjugated secondary reagent was added, and then, after 60 minutes, a substrate included in the kit was added to initiate a color reaction. After stopping the reaction, optical densities at 450/570 nm were determined. 

Cell Proliferation

Subconjunctival fibroblasts (1.5 × 10³/100 μL/well) were seeded into 96-well culture plates and incubated in MEM-10 for 10 to 24 hours before reaching confluence. They were then incubated with different concentrations of emodin for another 24 hours in the presence or absence of 10 ng/mL TNF-α in serum-free medium. Cell proliferation was assayed by using Alamar blue (Trek Diagnostic Systems, West Sussex, UK) according to the manufacturer’s protocol. ¹⁹ After a wash with phosphate-buffered saline (PBS), 40 μL Alamar blue was diluted in culture medium (1:2). Three hours later, the optical absorbance was measured at 570 nm. 

Cell Migration

Cell migration rates were examined by a described scratch wound assay. ²⁰ In brief, the extent of closure of a linear defect produced in a fibroblast monolayer was determined in the presence or absence of 5 μg/mL emodin in a serum-free condition. The remaining distances between migrating cells at the defect boundaries were measured at three different points. 

Immunocytochemistry

Cells (2.0 × 10³/300 μL/well) were grown to subconfluence in 10% serum-plus medium in the wells of 8-well chamber slides (Laboratory-Tec; Nunc, Rochester, NY). They were incubated in the presence or absence of different concentrations of emodin in serum-free medium supplemented with TGF-β1 (1 ng/mL) for another 24 hours. The cells were then fixed with paraformaldehyde and processed for immunohistochemistry to detect collagen I and fibronectin expression as reported. ²¹ Antibodies used were goat polyclonal anti–collagen I antibody (1:100 dilution in PBS; Southern Biotechnology, Birmingham, AL) and goat polyclonal anti–fibronectin antibody (1:100 dilution in PBS; Santa Cruz Biotechnology, Santa Cruz, CA). After washing, the cells were first exposed to FITC-conjugated secondary antibodies (1:100 dilution in PBS; ICN Biomedicals, Aurora, OH) and then DAPI stained. Their staining patterns were observed with fluorescence microscopy. 

Real-Time Reverse Transcription-Polymerase Chain Reaction MCP-1 mRNA Expression Detection

Confluent fibroblast cultures were further incubated for 24 hours in the presence or absence of different emodin concentrations in serum-free medium in the presence or absence of 10 ng/mL recombinant human TNF-α. Total RNA was extracted (Mammalian Total RNA Miniprep Kit; Sigma, St. Louis, MO). MCP-1 mRNA expression was examined with reported primers and TaqMan probes (Table 1) . ²² Three independent experiments were conducted. 

Collagen Type 1, Fibronectin, and MCP-1 ELISA Determination

Confluent fibroblast cultures were further incubated for 48 hours in the presence or absence of different emodin concentrations in serum-free medium in the presence or absence of 10 ng/mL recombinant human TNF-α. Culture medium was harvested and a cell layer was sonicated in 500 μL PBS. The samples were stored at −80°C before analysis. Concentration of human procollagen type 1 C-peptide (PIP), human fibronectin, or human MCP-1/CCL2 was determined by using a PIP-EIA Kit (Takara, Tokyo, Japan) according to manufacturer’s protocols. Fibronectin detection used either an EIA kit (Takara) or an ELISA kit purchased from R&D Systems. ^{23 24}  

Effects of Emodin on the Formation of Neovascularization in an In Vitro Coculture Model

A commercial kit containing an in vitro coculture system of human vascular endothelial cells (HUVECs) and fibroblasts was used (NV kit; Kurabo, Tokyo, Japan) according to the manufacturer’s protocol. This system cocultures vascular endothelial cells on a fibroblast feeder layer and is used to evaluate new vessel formation based on increases in tubelike tissue formation. 

The effect of adding emodin (0–2.0 μg/mL) on VEGF-A (10 ng/mL; Kurabo)-stimulated vessel-like tube formation was determined in a serum-free condition according to the protocol provided by the manufacturer. Tubelike tissue was detected by immunostaining for CD31, an endothelial cell marker. Color development was performed by diaminobenzidine (DAB) color reaction, as reported. ²⁰ Five wells were prepared for each culture condition, and the length, number of branch points, and mean value were determined in three different 300-μm² regions. 

Emodin-Induced Changes in NF-κB, JNK, and TGF-β1–Mediated Signaling

Effects of emodin on TNF-α–induced signaling in subconjunctival fibroblasts were assayed by using immunocytochemistry and Western blotting. The cells (4.5 × 10³/300 μL/well) were grown to subconfluence in 8-well chamber slides (Laboratory-Tec; Nunc) in MEM-10 and were further incubated for 24 hours in a serum-free condition. They were then incubated in the presence or absence of emodin in serum-free medium for 6 hours. Finally, the cells were treated with recombinant human TNF-α (10 ng/mL; R&D Systems) for 0.5, 1, or 2 hours and were fixed with 4% paraformaldehyde for 24 hours, followed by processing for immunocytochemistry. ²⁵ Cell nuclei were labeled with DAPI dye. 

To further characterize TNF-α signaling events, the effects of emodin on these responses were characterized by Western blotting. Cells were grown to subconfluence in 60-mm culture dishes in MEM-10 and were further incubated for 24 hours in a serum-free condition. They were then incubated in the presence or absence of emodin in serum-free medium for 6 hours, after which they were treated with recombinant human TNF-α (10 ng/mL; R&D Systems) for specific intervals and were harvested in buffer (100 μL/dish; Mammalian Cell Lysis; Sigma) and processed for SDS-PAGE and Western blotting. ²⁶  

Antibodies used were rabbit polyclonal antibodies against p65RelA of NF-κB, phosphorylated p65 RelA, c-Jun N-terminal kinase (JNK) and phospho-JNK (Cell Signaling Technology, Beverly, MA). 

TGF-β1/Smad2/3 signaling was examined with the same methods as for TNF-α–mediated control. ^{22 27 28} TGF-β1 at 1 ng/mL (R&D Systems) was used, and signal activation was determined by using goat polyclonal anti–Smad2 antibody (1:1000 dilution in PBS; Santa Cruz Biotechnology), rabbit polyclonal anti–phospho-Smad2 (Ser^465/467) antibody (1:2000 in PBS; Chemicon, Temecula, CA), goat polyclonal anti–Smad3 antibody (1:1000 dilution in PBS; Santa Cruz Biotechnology), rabbit polyclonal anti–phospho-Smad3 (Ser^423/425) antibody (1:1000 in PBS; Biosource, Camarillo, CA). 

Emodin Effects on Ocular Surface Fibrosis in Mice

Experiments were approved by the Animal Care and Use Committee of Wakayama Medical University and were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

A mouse corneal burn model was created by topical application of 0.5 n NaOH (3 μL). Alkali was applied to an eye of adult C57BL/6 mice (n = 58) under general and topical anesthesia 24 hours after intraperitoneal injection of emodin (40 mg/kg) or its vehicle. Then these different groups of mice received injections (40 mg/kg daily) for as long as 15 days. The mice were killed at days 5, 10, and 15, and their eyes were processed for fixation in 4% paraformaldehyde or total RNA extraction, as previously reported. ^{21 22 23} The number of eyes used for histology were 4, 4, and 3 at each time point for all conditions. For RNA extraction, six were used at each time point. Each RNA sample contained two corneas. Immunohistochemistry was conducted by using an antibody against αSMA or F4/80 macrophage antigen, as reported. ^{25 26} Real-time RT-PCR was performed for collagen Iα2, TGF-β1, and MCP-1, as described (Table 1) . 

Statistical Analysis

Results are expressed as means ± SD. Student’s t-test was used to compare two groups of animals, whereas ANOVA was used in multiple group comparisons. P < 0.05 was considered significant. 

Results

Cell Viability Insensitive to Emodin

The possible cytotoxic effects of emodin were evaluated based on measurements of NMP41/7 release into the medium. In the presence or absence of recombinant TNF-α, its levels were not changed at 48 hours (range, 15–17 U/mL) by exposure to emodin at concentrations up to 5.0 μg/mL (Fig. 1a) . Therefore, emodin is nontoxic to mouse subconjunctival fibroblasts. 

Selective Retardation of Wound Closure by Emodin

Another approach to assess for emodin cytotoxicity was to determine whether it inhibited mouse subconjunctival fibroblast proliferation. The results in Figure 1balso show that emodin failed to suppress proliferation over the same concentration range used to evaluate cell viability, irrespective of the presence or absence of TNF-α. However, cell migration was markedly suppressed, as indicated by the results shown in Figure 1c . Under control conditions, the distance separating the encroaching cells was very small 24 hours after wounding, whereas with 5 μg/mL emodin only partial wound closure occurred (Fig. 1c) . 

Emodin Suppression of Expression of Fibrogenic Components in Subconjunctival Fibroblast

The effects of emodin on conjunctival fibroblast phenotype were evaluated based on ELISA measurements of type 1 collagen and fibronectin in the culture medium. 

Emodin did not alter the total tissue cellular levels of type 1 collagen (Fig. 2a)or fibronectin (Fig. 2b) , but it selectively reduced the fibronectin content in the extracellular space (Fig. 2c) . In culture with TGF-β1 (1 ng/mL) and TNF-α (10 ng/mL), collagen type 1 extracellular deposition was not affected by emodin (Figs. 2dA 2dB) . On the other hand, emodin dramatically reduced extracellular deposition of fibronectin (Figs. 2dC 2dD) . 

Emodin Suppression TNF-α–Induced MCP-1 Expression

MCP-1 is a chemoattractant expressed by various cell types, including fibroblasts. Its upregulation leads to recruitment of macrophages, which in turn induces fibrotic reactions through their expression of TGF-β. As TNF-α is a proinflammatory cytokine expressed by corneal epithelial cells and macrophages, changes in its level, along with those of TGF-β, serve as indicators of pathophysiological changes resulting from injury. To determine whether emodin suppresses the chemoattractive effects of fibroblasts on macrophages, real-time RT-PCR was used to probe for emodin-induced changes in MCP-1 mRNA expression in the presence and absence of TNF-α. Emodin over a concentration range from 1 to 5 μg/mL suppressed the dose-dependent increases by TNF-α on MCP-1 mRNA expression and MCP-1 protein concentration (Figs. 3a 3b , respectively). These results indicate that emodin suppresses TNF-α–induced pro-inflammatory reactions in fibroblasts. 

Effects of Emodin on the Formation of Neovascularization in an In Vitro Coculture Model

Exogenous VEGF-A induced vessel-like tube formation in HUVECs based on increases in CD31 staining. Furthermore, both 1 μg/mL and 2 μg/mL emodin markedly suppressed tube elongation and branching (Fig. 4) . At the higher emodin concentration, both responses to VEGF were completely obviated, suggesting that this herbal component is an effective inhibitor of neovascularization. 

Emodin Inhibits TNF-α–Induced JNK Pathway Activation

Because emodin inhibits inflammatory responses, as described, its effects were determined on cell signaling activated by TNF-α, one of the major proinflammatory cytokines. We examined whether emodin alters recombinant TNF-α–induced activation of the NF-κB and JNK signaling branches known to mediate responses to this proinflammatory cytokine. ²⁹ Figure 5shows the time-dependent changes in phosphorylation status of the NF-κB subunit, p65 RelA, and its nuclear localization induced by TNF-α. Figure 5ashows that exposure to 10 ng/mL TNF-α induced activation of the NF-κB subunit, p65 RelA, at 1 hour, whereas in the presence of either 2.5 or 5 μg/mL emodin, p65 RelA phosphorylation was reduced at 1 hour. Equivalence of protein loading was documented by the nearly identical intensity of the total RelA bands shown in the bottom portion of Figure 5a . Figure 5bshows the changes in phospho-RelA localization induced by 10 ng/mL TNF-α addition. Irrespective of the presence or absence of emodin, no nuclear localization was detectable of this NF-κB subunit. In the absence of emodin, nuclear translocation was first detectable after 30 minutes, which increased to reach a maximum after 1 hour and was followed by a decline 1 hour later. With 1 μg/mL or 2.5 μg/mL emodin in the medium, phospho-RelA translocation was not inhibited at 30 minutes, whereas 5 μg/mL emodin blocked this response. On the other hand, after either 1 or 2 hours, emodin at all tested concentrations fully blocked phospho-RelA localization. Figure 6shows the time-dependent changes of JNK phosphorylation status and its nuclear localization induced by 10 ng/mL TNF-α in the presence and absence of emodin. Figure 6ashows that, in the absence of emodin, phospho-JNK formation reached a maximum value after 1 hour, followed by its disappearance after 2 hours. Such activation by TNF-α was fully blocked by all emodin concentrations. Equivalence of protein loading is documented by invariant levels of total JNK. The correspondence between these changes in JNK activation and phospho-JNK nuclear translocation is provided in Figure 6b . In the absence of emodin, TNF-α (10 ng/mL) induced nuclear translocation of phospho-JNK within 30 minutes, which remained evident for the next 30 minutes At 2 hours, such localization was no longer detectable. On the other hand, with TNF-α and 1 μg /mL emodin together, immunoreactivity at 30 minutes decreased and was more transient because it was no longer detectable at 1 hour. 

Injury-induced increases in TGF-β1 mediated fibroblast transformation through Smad2/3 signaling stimulation. Because emodin suppresses injury-induced pathophysiology associated with TGF-β1–mediated myofibroblast transformation, we sought to determine whether emodin suppressed TGF-β1–induced signaling. At emodin concentrations up to 5 μg/mL, neither the TGF-β1–induced increase in Smad2/3 expression levels nor the nuclear immunoreactivity of C-terminal phospho-Smad2/3 was affected (data not shown). Therefore, emodin did not inhibit TGF-β1–induced signaling. 

Systemic Emodin Suppresses Ocular Surface Fibrosis in Mice

Given that emodin suppresses TNF-α–induced subconjunctival fibroblast responses, we hypothesized that emodin reduces mouse corneal scarring in an alkali burn model. An alkali burn of the ocular surface resulted in a corneal epithelial defect in the early phase (10 days) and then stromal scarring (opacification) and neovascularization in the later phase (15 days; Fig. 7aA 7aC ). However, the resultant stromal opacification was much lower in mice that received daily systemic intraperitoneal injections of emodin (Figs. 7aB 7aD)than in control animals. The injected volume of either emodin or its vehicle was the same as that used for treating malignant tumor in mice. ³⁰  

To determine whether there is an association between alkali-induced injury and changes in gene expression induced by TNF-α in vitro, real-time RT-PCR was performed with primer pairs for the detection of collagen Iα2, TGF-β1, and MCP-1 RNA in isolated ocular tissue samples. Figure 7bshows that in the nontreated controls, collagen Iα2 expression levels peaked at day 10. TGF-β1 expression was slightly higher at 5 days than at 10 and 15 days. On the other hand, MCP-1 levels decreased markedly from day 5 until day 15 (Fig. 7b) . Expression levels of the three genes were lowest in emodin-treated mice. 

Histologic examination and immunohistochemistry were performed to determine whether alkali-induced ocular surface changes in emodin-treated and control mice were associated with structural changes, myofibroblast generation, and macrophage infiltration during healing. The markers used to monitor fibroblast transdifferentiation and magnitude of macrophage infiltration were intensity of αSMA and F4/80 staining, respectively. There was a good association in the two groups between immunohistochemical changes and alterations in ocular surface integrity because less staining correlated well with improved tissue appearance. Hematoxylin and eosin staining showed that there was more cellularity and thickening in the bulbar conjunctiva region in control animals than in emodin-treated mice (Fig. 8a) . Overall, systemic emodin treatment suppressed such histologic changes during the healing interval (Fig. 8a) . Furthermore, there were fewer myofibroblasts and macrophages in the subconjunctival connective tissue of emodin-treated mice than in control mice at days 10 and 15 (Figs. 8b 8c) . These changes were in marked contrast with the more pronounced changes seen in control corneas (Fig. 9) . The control group exhibited more cellularity with stromal thickening, probably because of edema, than the emodin-treated group cornea at days 10 and 15 (Fig. 9a) . There were fewer myofibroblasts and macrophages in the healing stroma of the emodin-treated group mice than in control animals at these time points (Figs. 9b 9c) . Therefore, systemic emodin administration could have therapeutic potential in lessening corneal opacification caused by fibrosis and neovascularization during healing. 

Discussion

Emodin suppressed proinflammatory and profibrogenic activities in vitro in subconjunctival fibroblasts and in vivo in an alkali-burned mouse ocular surface. The protective effects of emodin are attributable to the inhibition of fibroblast migration, declining MCP-1 chemoattractant gene and protein expression, and the suppression of TNF-α–induced NF-κB and JNK branch signaling. Another attribute of the in vitro studies observed in vivo was that neovascularization was suppressed by emodin in proinflammatory and profibrogenic activities. These effects of emodin account for decreased alkali-induced corneal and conjunctival scarring. Taken together, the findings suggest that this herbal component may have therapeutic value in reducing the losses in corneal structural integrity and transparency that occur during alkali injury-induced healing. 

The mechanism accounting for the suppressive effect of emodin on cell migration was determined by evaluating its effect on TNF-α–induced JNK activation because this pathway can mediate the control of cell migration. ³¹ Given that emodin suppressed such JNK activation and inhibited cell migration, emodin-induced inhibition of JNK activation could account for its suppression of cell migration. Another contributing factor that could explain the decline in cell migration is that emodin also reduced fibronectin expression and its cell layer deposition. On the other hand, emodin was not toxic, antimitogenic, or inhibitory to in vitro type 1 collagen production. These negative effects suggest that emodin induces its anti-inflammatory fibrogenic effects through selective alteration of cell signaling pathways. Even though a previous study showed that emodin suppresses collagen production, such inhibition might be cell type dependent. ^{10 12}  

Tissue scarring during the healing process is a result of myofibroblast-mediated contraction of connective tissue. TGF-β levels rise during this response to injury, inducing fibroblast-to-myofibroblast transdifferentiation, which is characterized by the appearance of αSMA expression. ^{32 33} This response is a consequence of TGF-β stimulation of Smad2 signaling, leading to increases in αSMA gene promoter and extracellular scaffold αSMA activity. ^{33 34} In addition, the appearance of αSMA contractile cytoskeletal fibers, or the generation of myofibroblasts, depends on extracellular deposition of fibronectin, especially ED-A type fibronectin. ^{35 36} The present in vivo experiment showed that emodin suppressed myofibroblast generation in healing ocular tissue (corneal stroma and subconjunctival tissue). Because emodin did not directly inhibit the Smad signal in fibroblasts, the in vivo phenomenon might have been caused by the secondary reduction of TGF-β1 in tissue with less inflammation with emodin. However, our unpublished data showed that emodin suppressed aSMA-cytoskeletal fiber formation without altering the protein expression level of aSMA, as detected by Western blotting. Although the mechanism of this phenomenon is to be uncovered, a similar phenomenon was reported. ^{37 38} The present study leaves unexplained why there were decreases in fibronectin deposition within the fibroblasts, but the fibronectin medium content remained unchanged. Its deposition may be mediated by NF-κB or JNK signaling, but the upstream effectors of their activation are unclear. Even though tissue fibronectin deposition might be favorable for epithelial healing, emodin inhibition of fibronectin deposition instead improved the outcome of the healing process in our mouse corneal alkali burn model, suggesting the therapeutic effects of emodin are essentially of stromal rather than epithelial origin. However, its effects in the stroma are not attributable to a decline in the fibroblast population because in vitro emodin did not inhibit TNF-α–induced stimulation of fibroblast proliferation. The effects of emodin on cell proliferation are cell type specific because it suppresses neoplastic cell growth but has no effect on this process in mesenchymal cell types (hepatic stellate cells). This negative effect is consistent with our observation that emodin did not inhibit fibroblastic cell proliferation because they are also derived from mesenchyme. 

TNF-α and TGF-β are major proinflammatory/profibrogenic cytokines expressed by ocular surface tissues. We used them to mimic the inflammatory responses by subconjunctival cells after injury that resulted from their cognate receptor activation. Such responses in turn resulted from invasive infiltration of inflammatory cells mediated by such chemoattractants as MCP-1. The reduction in TNF-α–induced increases in MCP-1 gene and protein expression could lead to declines in invasive monocytes/macrophages from surrounding tissue, in turn decreasing fibroblast activation and transformation by proinflammatory and profibrogenic cytokines (e.g., TNF-α and TGF-β). Consistent with emodin inhibition of proinflammatory responses, this herbal component also suppressed TNF-α–induced NF-κB p65 and JNK activation. On the other hand, emodin did not affect the TGF-β1/Smad pathway signaling, as determined by its failure to suppress cytokine-mediated stimulation of C-terminal phosphorylation of Smad2/3 (data not shown). However, another effect by emodin on TGF-β signalιng might still have been possible because we failed to examine whether emodin affected Smad2/3 signal by modulating the phosphorylation of middle linker regions of Smad2/3 molecules. ^{39 40 41}  

Another possible factor contributing to the improved outcome of the healing response to ocular surface injury is that emodin therapy suppressed neovascularization. Neovascularization is also an important component in the development of unfavorable scar tissue formation on the ocular surface (i.e., the cornea). This effect is consistent with our in vitro observation in which emodin suppressed cells TNF-α–induced vessel-like tube formation in cultured endothelial. Therefore, emodin appears to be an effective antiangiogenic agent of potential use in a clinical setting. 

Tissue fibrosis is thought to be caused by cytokine acceleration of extracellular matrix deposition. Although emodin did not affect type 1 collagen expression by fibroblasts, it suppressed the expression of other components associated with extracellular matrix formation; namely, it inhibited fibronectin deposition, myofibroblast formation, and MCP-1 expression. These latter effects suggest that emodin might have a therapeutic effect on the ocular surface fibrogenic disorders vernal or atopic conjunctivitis or Stevens-Johnson syndrome. To explore this hypothesis, we tested whether systemic administration of emodin might have a therapeutic effect on a mouse alkali burn model of mice fibrogenic diseases. This is a relevant model because during the healing response to injury, corneal or conjunctival opacification or scarring develops. Such pathophysiology is also observed in human ocular surface scarring diseases, in the later phase of healing. Systemic intraperitoneal administration of emodin effectively suppressed fibrogenic reaction in cornea and conjunctiva, as revealed by reduced macrophage invasion, inhibited myofibroblast generation, and decreased mRNA expression levels of collagen Iα2, TGF-β1, and MCP-1. Reductions of MCP-1 expression might account for declines in macrophage infiltration, resulting in declines in TGF-β1 levels within the injured site. Although emodin did not affect type 1 collagen production by cultured fibroblasts, systemic emodin did suppress the expression of collagen Iα2 mRNA in a healing cornea. This phenomenon could be explained by the reduced expression of TGF-β1 mRNA in the healing tissue. 

In conclusion, our current understanding of the mechanism, emodin suppressed TNF-α–induced increases in proinflammatory mediators and profibrogenic cytokine levels in fibroblasts presumably by inhibiting JNK and NF-κB signaling. Even though our study provides evidence suggesting that emodin has potential beneficial value in reducing in vivo ocular surface scarring during healing, such symptom lessening occurred 24 hours after the systemic administration of emodin. It is not yet possible to extrapolate with certainty whether emodin is of use in the clinical setting because expression of the ocular surface abnormalities associated with such diseases as vernal conjunctivitis and Stevens-Johnson syndrome occurs over a much longer time. Before emodin can be considered for use in such disturbances, it is first necessary to evaluate whether this herbal component produces adverse effects after repeated systemic administration over long periods of time. The herbal medicine Inchin-ko-to may be used for the treatment of ocular fibrotic diseases. 

 Supported by Grants C15591871 (SS) and C16590150 (KI) from the Ministry of Education, Science, Sports and Culture of Japan, the research fund of the Institute of Kampo Medicine, and the research fund of Japan Kampo Medicine Manufacturers Association (OY).

 Submitted for publication April 3, 2007; revised June 27, 2007; accepted September 18, 2007.

 Disclosure: A. Kitano, None; S. Saika, None; O. Yamanaka, None; K. Ikeda, None; Y. Okada, None; K. Shirai, None; P.S. Reinach, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Primer Designs to Detect Gene Expression

Transcript	Sequence
hMCP-1	F: 5′-tct ctg ccg ccc ttc tgt-3′
	R: 5′-gca tct ggc tga gcg agc-3′
	P: 5′-ctg ctc ata gca gcc acc ttc att ccc-3′
mCol 1a2	F: 5′-aag ggt ccc tct gga gaa cc-3′
	R: 5′-tct aga gcc agg gag acc ca-3′
	P: 5′-cag ggt ctt ctt ggt gct ccc ggt at-3′
mTGFβ-1	F: 5′-gca aca tgt gga act cta cca gaa-3′
	R: 5′-gac gtc aaa aga cag cca ctc-3′
	P: 5′-acc ttg gta acc ggc tgc tga ccc-3′
mMCP-1	F: 5′-tgg ctc agc cag atg cag t-3′
	R: 5′-cca gcc tac tca ttg gga tca-3′
	P: 5′-ccc cac tca cct gct gct act cat tca-3′

 F, forward primer, R, reverse primer, P, TaqMan probe.

Figure 1.

 

Adding emodin up to 5 μg/mL neither induces cell death in the 10% serum-plus medium (a) nor inhibits cell proliferation in the serum-free medium (b) of cultured human subconjunctival fibroblast (SCF) in the presence or absence of exogenous TNF-α (10 ng/mL). (c) Emodin suppressed migration of human SCFs toward the defect produced in the monolayer sheet in the serum-free medium. A linear defect of the same width was made in control (cA) or emodin-plus (cB) culture. At 24 hours, cells that had migrated from each defect edge reached the center of the linear defect in the control culture (cC) but not in an emodin-containing culture (cD). Scale bar, 200 μm. (d) Distance separating the migrating cells from each defect edge at 24 hours.

Figure 2.

 

Effect of emodin on production of collagen type 1 and cellular fibronectin. Levels of collagen type 1 (a) and cellular fibronectin (b) in the serum-free culture medium are not affected by adding emodin up to 10 μg/mL in the presence or absence of TNF-α. TNF-α significantly increased cell layer–associated fibronectin expression with P < 0.01. Cellular fibronectin levels decreased in the cell layer more markedly at emodin concentrations in the presence of TNF-α than in its absence (c). (d) Immunohistochemistry shows that deposition of collagen type 1 is similar in human SCF with (dB) and without (dA) emodin at 5 μg/mL in the serum-free condition, whereas that of cellular fibronectin is reduced by emodin (dD) compared with control, TNF-α -plus, emodin-minus culture (dC). *P < 0.05 and **P < 0.01 compared with control in each culture condition. Scale bar, 100 μm.

Figure 3.

 

Effects of emodin on mRNA and protein expression levels of MCP-1. At the mRNA level, MCP-1 expression is suppressed in the presence and the absence of exogenous TNF-α in the serum-free condition. TNF-α significantly increased protein expression of MCP-1 with P < 0.01. MCP-1 protein production was reduced by emodin in the presence of TNF-α but not in its absence in the serum-free condition. mRNA expression data represent typical results of 1 of 3 independent experiments. **P < 0.01 compared with control.

Figure 4.

 

Effects of emodin on in vitro vessel-like tube formation by HUVECs cocultured on fibroblast feeder cells. (a) Immunocytochemical detection of CD31 in vessel-like tube structure newly formed by HUVECs in the presence of exogenous VEGF-A. In the absence of emodin, CD31-labeled structures are readily observed; adding emodin suppressed their formation in the serum-free condition. (b, c) Total length and number of branching points of the tube structure in 300-μm² region in each culture condition. Emodin at 2 μg/mL abolished the formation of the tube structure.

Figure 5.

 

Effect of emodin on nuclear factor-κB signaling on exposure to exogenous TNF-α. (a) Western blotting showed that adding TNF-α initially activated RelA within 1 hour and then deactivated it after 2 hours Adding emodin at 2.5 and 5 μg/mL reduced the expression level of phospho-RelA 1 hour after the addition of TNF-α. The total RelA level was unchanged with emodin. (b) Immunocytochemistry also showed that adding TNF-α induced the nuclear accumulation of phospho-RelA (arrowheads) at 0.5 to 1 hour that decreased at 2 hours. Nuclear phospho-RelA localization was reduced with emodin (1–5 μg/mL), and the duration of this response was shortened. NF-κB signaling was examined in the serum-free condition. (bA–bD, bE–bH, bI–bL, bM–bP) Cells at 0, 0.5, 1, and 2 hours after TNF-α was added. Cells treated without or with (bA, bE, bI, bM [1 μg/mL]; bB, bF, bJ, bN [2.5 μg/mL]; bC, bG, bK, bO or bD, bH, bL, bP [5 μg/mL]) emodin. Scale bar, 50 μm.

Figure 6.

 

Effect of emodin on JNK signaling on exposure to exogenous TNF-α. (a) Western blotting showed that adding TNF-α activated JNK within 1 hour and then deactivated it after 2 hours. Adding emodin at 1, 2.5, and 5 μg/mL reduced the expression level of phospho-JNK 1 hour after TNF-α was added. Total JNK level was unchanged with emodin. (b) Immunocytochemistry also showed that adding TNF-α induced the nuclear accumulation of phospho-JNK (arrowheads) from 0.5 to 1 hour but within 2 hours. Such nuclear phospho-JNK localization was reduced with emodin. JNK signaling was examined in the serum-free condition. (bA–bD, bE–bH, bI–bL, bM–bP) Cells at 0, 0.5, 1, and 2 hours after the addition of TNF-α. Cells treated without (bA, bE, bI, bM) or with (bB, bF, bJ, bN) 1 μg/mL, (bC, bG, bK, bO) 2.5 μg/mL, or (bD, bH, bL, bP) 5 μg/mL emodin. Scale bar, 50 μm.

Figure 7.

 

Healing of an alkali-burned cornea in a control mouse and in a mouse treated intraperitoneally with systemic emodin. (a) Stromal ulceration or opacification and neovascularization in healing corneas. At day 10, ulceration was observed in the control central cornea (aA). Eyelids were not completely open in a control eye, probably because of inflammation and scarring, whereas an emodin-treated eye was resurfaced but had stromal opacification (aB). At day 15, the cornea of a control mouse was prominently opaque in association with hemorrhage in the anterior chamber (aC). In marked contrast, minor opacification was detected in an emodin-treated eye (aD). Iris tissue is seen through the transparent cornea. (b) Data from real-time RT-PCR of total RNA extracted from eye tissue samples. Data are shown as time-fold increases over basal expression levels. Expression of collagen Iα2 peaked at day 10, whereas that of TGF-β1 and MCP-1 decreased until day 15. Overall systemic emodin treatment suppressed expression of these components during the healing interval. Data represent mean ± SD. *P < 0.05; **P < 0.01.

Figure 8.

 

Histology and immunohistochemistry of bulbar conjunctiva of eyes in control mice and those treated with systemic administration of emodin. (a) Hematoxylin and eosin–stained histology of healing conjunctiva at each interval. At day 10, more cells were observed in subconjunctival connective tissue (asterisks) in a control eye (aA) than in the tissue of an emodin-treated mouse (aB). At day 15, subconjunctival tissue of a control eye was thicker (aC), probably because of edema, compared with a treated specimen (aD). (b, c) Distribution of myofibroblasts and macrophages as detected by immunohistochemical staining intensity of αSMA and F4/80 antigen. (b) Myofibroblasts were observed in healing subconjunctival tissue of control (bA) and emodin-treated (bB) mice, as detected by rhodamine immunoreactivity. At day 15, choroid was also thickened and myofibroblasts were abundant (bC). At this time, almost no myofibroblasts were seen in subconjunctival space of an emodin-treated eye, whereas they were still observed in a control eye (bD). (bC) Abundant αSMA-positive cells were observed in choroid (star). (c) Macrophages were detected based on green fluorescence staining. At day 10, macrophages were observed in subconjunctival tissue (cA), but there were almost no cells in the tissue of day 10 treated mice (cB) and day 15 mice of both groups (cC, cD). Ch, choroid; E, epithelium. Scale bar: (a–c) 100 μm.

Figure 9.

 

Histology and immunohistochemistry of cornea of eyes in control mice and those treated with systemic administration of emodin. (a) Hematoxylin and eosin staining histology of healing cornea at each interval. At day 10, more cells were observed in corneal stroma (asterisk) in a control eye (aA) than in the tissue in an emodin-treated mouse (aB). At day 15, corneal stroma of a control eye was thicker (aC), probably because of edema, than in a treated specimen (aD). (b, c) Distributions of myofibroblasts and macrophages as detected by immunohistochemical staining of αSMA and F4/80 antigen. (b) Myofibroblasts were observed in healing corneal stroma of control (bA) and emodin-treated mice (bB), as detected by rhodamine immunoreactivity. At days 10 and 15, more myofibroblasts were seen in the corneal stroma of an emodin-treated eye (bB, bD) and were still observed in a control eye (bA, bC). (c) Macrophages were detected as green fluorescence staining. At days 10 (cA, cB) and 15 (cC, cD), fewer macrophages were observed in the corneal stroma of an emodin-treated eye (cB, cD) than in a control eye (cA, cC). E, epithelium. Scale bar: (a–c) 100 μm.