The Epstein-Barr Virus Causes Epithelial–Mesenchymal Transition in Human Corneal Epithelial Cells Via Syk/Src and Akt/Erk Signaling Pathways

Abstract

Purpose.: Although Epstein-Barr virus (EBV)–associated keratitis is rare, it can cause acute corneal necrosis and neovascularization. We aimed to examine the signaling mechanism by which EBV causes epithelial–mesenchymal transition (EMT) in human corneal epithelial cells (HCECs) in vitro.

Methods.: The cellular response to EBV was assessed by real-time PCR, Western blot, migration assay, invasion assay, inhibitor assay, and ELISA assay.

Results.: A model of EBV-induced EMT was established in HCECs. The EBV induced morphologic changes in the cells; the loss of epithelial markers E-cadherin, ZO-1, and β-catenin; and an increase in the mesenchymal markers N-cadherin, Vimentin, Snail, and TCF8/Zeb1. The EBV infection also led to the nuclear translocation of Snail and TCF8/Zeb1; enhanced the secretion of IL-6, IL-8, VEGF, TGF-β1, TNF-α, and MCP-1; and upregulated the expression of MMP2 and MMP9. The EBV-infected HCECs exhibited increased migration and invasiveness compared to uninfected HCECs. We measured the involvement of Syk, Src, PI3K/Akt, and Erk signaling, but not Smad, in EMT by EBV-induced TGF-β1. We demonstrated that treatment with TGF-β1, TGF-β receptors, Syk, or Src inhibitor blocked TGF-β1, Syk, or Src signaling activation, and EMT development by EBV. Moreover, these inhibitors prevented PI3K/Akt and Erk activation.

Conclusions.: An EBV infection in HCECs can lead to a mesenchymal fibroblast-like morphology, and cause EMT through the activation of PI3K/Akt and Erk by TGF-β1–mediated Syk and Src signaling. This phenomenon may have implications for EBV-associated keratitis and molecular approaches to treatment.

Introduction

Homeostasis of the corneal epithelium is important in the maintenance of normal vision and epithelial tissue integrity. If homeostasis is disturbed by infection, injury, epithelial hypertrophy, or stem cell deficiency, visual impairment can result, which, in turn, can cause corneal fibrosis. Epithelial injury must heal quickly to prevent additional damage from infection. Corneal wound healing involves various cellular systems, including cell migration and proliferation. It generally is accepted that TGF-β1 regulates the diverse events surrounding wound healing. 

Epithelial–mesenchymal transition (EMT) is a process that induces cell motility, and increases survival during wound healing and embryogenesis. ^1,2 A key mechanism in fibrosis, chronic inflammation, and tumor metastasis, EMT is induced by various growth factors and external stressors, such as TGF-β1, PDGF, EGF, and hypoxia, ^2,3 and is regulated by signaling networks that include Wnt, Notch, PI3K/Akt, Ras/MAPK/Erk, FAK, Smad, and RhoB. ^2,4 The EMT is characterized by a transition from a polarized epithelial phenotype to a mesenchymal phenotype or highly motile fibroblastoid by promoting downregulation of epithelial markers, such as E-cadherin, β-catenin, and ZO-1, as well as upregulation of mesenchymal markers, such as fibronectin, Vimentin, and α-SMA. ^1,2  

Various transcription factors, including Snail, Slug, TCF8/Zeb1, and Twist, are known to be capable of regulating EMT in epithelial cells. ² Recent studies support the notion that Vimentin functions as a positive regulator of EMT. ⁵ The PI3K/Akt is known to have a crucial role in actin cytoskeleton remodeling and cell motility in many physiologic processes, including TGF-β1–induced EMT. ⁶ The PI3K/Akt and Erk have been shown to repress E-cadherin and increase Snail transcription. ^7,8 The TGF-β signals mainly through heteromeric complexes of type I and type II receptors that activate Smad signaling. ^9,10 However, TGF-β responses are not solely the result of the Smad signaling activation, and can, in turn, activate other signaling pathways, such as Erk, p38, MAPK, and JNK. ^11,12 Therefore, the downstream effects of TGF-β should not be considered a direct consequence of the Smad signaling pathway, but instead a result of multiple pathways. 

The Epstein-Barr virus (EBV), a virus infecting more than 90% of the world's population, is an ubiquitous human herpes virus 4 associated with various lymphoid and epithelial malignancies. ¹³ Several recent reports suggest that EBV infection may cause phenotypic changes and EMT in several types of epithelial cells. ^14–16 However, the relationship between EBV infection and EMT in primary human corneal epithelial cells (HCECS) remain poorly understood. In this study, we investigated EBV infection and EMT in HCECs. 

Materials and Methods

Cell Culture and Reagents

Primary HCECs were purchased from Invitrogen-Gibco (Carlsbad, CA). Cells were maintained in keratinocyte serum-free medium (KSFM) supplemented with bovine pituitary extract (Invitrogen-Gibco) and human recombinant epidermal growth factor (Invitrogen-Gibco) at 37°C in 5% CO₂. These cells were maintained for up to 12 passages. LY294002, PD98059, PP1, and Bay-61-3606 were purchased from Calbiochem (San Diego, CA). LY2109761 was obtained from Selleckchem (Houston, TX). Anti–TGF-β1 neutralizing antibody was obtained from R&D Systems (Minneapolis, MN). 

Preparation of EBV-Containing Culture Supernatant and Generation of EBV-Infected HCECs

Cell-free EBV-containing culture supernatant was prepared from the B95-8 cell line (EBV type I; ATCC, Manassas, VA). The virus titer of B95-8 EBV was >10² transforming units/mL. The transforming units were determined by clumping formation of peripheral mononuclear cells after infection using infectious culture supernatant. ¹⁷ After HCECs (passage 3) were completely attached at 6.25 × 10² cells/mL in a 75 cm² tissue culture flask, an equal volume of EBV-containing culture supernatant (8 mL, >10² transforming units/mL) was added. The cultures were incubated between 1 day and 4 weeks. The EBV-infected HCECs were maintained for up to 10 weeks of cultures. 

Quantitative Real-Time PCR

Total cellular RNA was isolated using an RNeasy Mini kit (Qiagen, Hilden, Germany) and transcribed into cDNA using oligo (dT) primers and reverse transcriptase 3 weeks after infection. Quantitative mRNA levels were measured by using an ECO real-time PCR system (Illumina, Inc., San Diego, CA) and SYBR Green Master Mix kit (Takara, Tokyo, Japan) with specific primer sets (see Table; Bioneer, Daejeon, Korea). 

Table

 

Specific Primers Sequences Used in RT-PCR

Target	Primers, 5′ → 3′
	Sense	Antisense
EBNA1	GAGCGGGGAGATAATGTACA	TAAAAGATGGCCGGACAAGG
EBNA2	AACCCTCTAAGACTCAAGGC	ACTTTCGTCTAAGTCTGCGG
LMP1	CACGACCTTGAGAGGGGCCCA	GCCAGATGGTGGCACCAAGTC
LMP2A	ATGACTCATCTCAACACATA	CATGTTAGGCAAATTGCAAA
IL-6	GTGTTGCCTGCTGCCTTCCCTG	CTCTAGGTATACCTCAAACTCCAA
IL-8	ATGACTTCCAAGCTGGCCGTGGCT	TCTCAGCCCTCTTCAAAAACTTCTC
TGF-β	GGACACCAACTATTGCTTCAG	TCCAGGCTCCAAATGTAGG
TNF-α	AACATCCAACCTTCCCAAACG	GACCCTAAGCCCCCAATTCTC
Hif-1α	TGATTGCATCTCCATCTCCTACC	GACTCAAAGCGACAGATAACACG
VEGF	AGGAGGGCAGAATCATCACG	CAAGGCCCACAGGGATTTTCT
Stat3	ACCTGCAGCAATACCATTGAC	AAGGTGAGGGACTCAAACTGC
MCP-1	AATGCCCCAGTCACCTGCTGTTAT	GCAATTTCCCCAAGTCTCTGTATC
MMP2	TGGCAAGTACGGCTTCTGTC	TGGCAAGTACGGCTTCTGTC
MMP9	TGCGCTACCACCTCGAACTT	GATGCCATTGACGTCGTCCT
β-actin	ATCCACGAAACTACCTTCAA	ATCCACACGGAGTACTTGC

Immunoblotting

Cells were harvested, lysed, and subjected to SDS-PAGE 4 weeks after infection as described previously. ¹⁷ The following primary antibodies were used: phospho-Stat3 (Tyr⁷⁰⁵), Stat3, Hif-1α_, MMP-2, MMP-9, phospho-PI3K p85 (Tyr⁴⁵⁸), PI3K p85, phospho-Akt (Ser⁴⁷³), Akt, phospho-ERK1/2 (Thr²⁰²/Tyr²⁰⁴), ERK1/2, E-cadherin, N-cadherin, β-catenin, Vimentin, Snail, TCF8/Zeb1, PARP, phospho-Lyn (Tyr⁵⁰⁷), Lyn, Fyn, phospho-Src (Tyr⁴¹⁶), Src, phospho-Syk (Tyr³²³), phospho-Syk (Tyr^525/526), Syk, phospho-Smad2/3 (Ser^465/467/Ser^423/425), Smad2/3, ▵p63, CNX43, and β-actin from Cell Signaling Technology (1:1000; Beverly, MA); phospho-Fyn (Thr¹²), EBNA-2, EBNA-3A, LMP-1, LMP-2A, Ang-1, and VEGF from Santa Cruz Biotechnology (1:200; Santa Cruz, CA); EBNA-1 from Thermo Scientific (1:50; Rockford, IL); and β-tubulin from BD Biosciences (1:500; San Diego, CA). 

ELISA Test

At 24 hours, the conditioned media were collected, and the quantity of IL-6, IL-8, and MCP-1 secreted by HCECs or EBV-infected HCECs was measured by the Single Analyte ELISArray Kit (Qiagen) according to the manufacturer's instructions. The VEGF, TNF-α, and active TGF-β1 were quantified by the Single cytokine ELISA assay Kit (R&D Systems). Total TGF-β1 was quantified by the LEGEND MAX Total TGF-β1 ELISA Kit (BioLegend, San Diego, CA). Data are expressed as the average of the number of biological replicates ± SD. 

Measurement of Snail and TCF8/Zeb1 Translocation

Cytosol and nuclear cellular fractions were prepared using a Nuclear/Cytosol Fractionation Kit (BioVision, Mountain View, CA) according to the manufacturer's instructions 4 weeks after infection. All fractions were stored at −80°C until use. 

Transendothelial Migration and Invasion Assay

The transendothelial migration of EBV-infected HCECs was detected using CytoSelect tumor transendothelial migration assay Kit (Cell Biolabs, Inc., San Diego, CA) according to the manufacturer's instructions 4 weeks after infection. The relative fluorescence units (RFU) of migrated cells were measured by a microplate reader. The invasion assay was determined using the CultureCoat 96-well Medium BME Cell Invasion Assay Kit (R&D Systems) according to the manufacturer's instructions. Invaded cells were stained with calcein AM and quantified by a microplate reader. 

Statistical Analysis

The PCR data, Western blot band, and ELISA data were analyzed statistically. Data were expressed as the mean ± SD. The statistical significance of differences was tested by Kruskal-Wallis testing using Prism (version 5). Differences were considered significant when P < 0.05. 

Results

Detection of EBV-Related Molecules in EBV-Infected HCECs

The EBV can infect B cells and epithelial cells. To infect HCECs with EBV, we first determined the presence of the well-known EBV receptors CD21, MHC class II, and integrin β1 by flow cytometry. The HCECs expressed only integrin β1 (data not shown). We established EBV-carrying HCECs showing type III-like latency. Typically, the EBV transformation process in B cells is established after 4 weeks. ¹³ We found that EBV-infected HCECs (EBV-HCECs) changed their shapes from cuboidal and ovoidal to spindle-like within the first week. Cells developed mesenchymal fibroblast-like spindle shapes and lost cell-to-cell contact 3 weeks after EBV infection (Fig. 1B), while uninfected cells preserved their typical epithelial pebble pattern (Fig. 1A). The EBV infection was confirmed by real-time PCR and Western blot to detect viral transcripts and proteins, which identified EBNA1, EBNA2, EBNA3A, and LMP1. No EBV-related gene transcripts or proteins were detected in uninfected HCECs (Figs. 1C, 1D). 

Figure 1

 

Analysis of the morphologic changes and EBV-related molecule expression of HCECs after EBV infection. (A, B) Morphology of EBV-infected versus uninfected HCECs. Nonexposed HCECs had a typical cobblestone-like monolayer appearance (A), while EBV infection (>4 weeks) induced a phenotypic transition from cuboidal clustered epithelial cells to elongated fibroblast-like spindle-shaped cells with a loss of cell-to-cell contact (B). Morphology was observed under an inverted phase-contrast microscope (Olympus, Tokyo, Japan). Photographs were taken at ×100 magnification by a digital camera. mRNA levels (C) and protein levels (D) of EBV-related gene expression in EBV-infected and uninfected HCECs were measured using real-time PCR and Western blot, respectively. As a positive control, we used EBV-transformed B cells. Data are representative of 3 independent experiments.

EBV Stimulates the Production of Pro-Inflammatory and Pro-Angiogenic Factors

We analyzed mRNA levels and protein levels of pro-inflammatory cytokines and angiogenic factors, and their levels of secretion into the culture supernatant 3 weeks after infection. We found that EBV-HCECs had significantly higher levels of expression of several pro-angiogenic and growth factors compared to uninfected cells: IL-6, IL-8, VEGF, MCP-1, and TNF-α statistically (Figs. 2A, 2B, 3C). In addition, EBV infection increased significantly the mRNA expression of TGF-β1, a well-known EMT inducer, and induced the secretion of active and total TGF-β1 in HCECs (Figs. 2B, 3A, 3B). The EBV-HCECs secreted the additional pro-inflammatory and pro-angiogenic factors related to neovascularization, Hif-1α and Stat3, and had increased expression levels of Hif-1α, p-Stat3, and Ang-1 compared to uninfected cells. The Stat3 target genes MMP2 and MMP9 also were upregulated in EBV-HCECs (Figs. 2A, 2B). 

Figure 2

 

The EBV promotes expression of pro-inflammatory and pro-angiogenic factors. Total RNA and protein were extracted from EBV-infected and uninfected HCECs as described in the Materials and Methods section. Real-time PCR (A, B) and Western Blot (C) analysis of Hif-1α, Stat3, VEGF, IL-6, IL-8, TNF-α, TGF-β, MCP-1, Ang-1, MMP2, and MMP9. *Statistically significant (P < 0.01) increase in expression levels in comparison to HCECs. Data are representative of 3 independent experiments.

Figure 3

 

Effect of EBV on pro-inflammatory cytokines. The EBV induced the secretion of IL-8, MCP-1, VEGF, IL-6, TNF-α, and active TGF-β1. The supernatants of EBV-infected and uninfected HCECs were assayed by ELISA. Cells were seeded into 6-well plates (1 × 10⁵/well) and incubated for 48 hours. *Statistically significant (P < 0.001) increase in cytokine levels in comparison to HCECs. Data are representative of 2 independent experiments and show the mean ± SD of duplicate determinations.

As reported previously, viral infections, such as EBV, elicit the production of cytokines and chemokines in immune cells and tumor cells. ^18,19 Therefore, we hypothesized that the differentiation of HCECs into fibroblast-like cells after EBV infection involved the virus's capability to elicit growth and angiogenic factors. Our results supported our hypothesis. 

EBV Induces Dramatic Phenotypic Changes and EMT Characteristics

Several characteristics observed in EBV-HCECs, namely their differentiation into fibroblast-like cells and secretion of TGF-β1, mirror events in EMT, leading us to hypothesize that EBV-HCECs were mesenchymal-like cells. To test this hypothesis, we measured the difference in expression of EMT-related markers between EBV-HCECs and uninfected cells. We found that uninfected HCECs expressed primarily the epithelial markers E-cadherin, ZO-1, and β-catenin, with low or absent expression of the mesenchymal markers N-cadherin, Vimentin, Snail, and TCF8/Zeb1 (Fig. 4A). In contrast, EBV-HCECs expressed high levels of N-cadherin, Vimentin, Snail, and TCF8/Zeb1, but no detectable E-cadherin, ZO-1, or β-catenin (Fig. 4A). We also observed that expression of the corneal epithelial cell marker ▵p63 was decreased, while expression of the corneal differentiation marker CNX43 was increased in EBV-HCECs compared to uninfected cells (Fig. 4A). 

Figure 4

 

The EBV increases mesenchymal cell markers and enhances cell migration and invasion. Total protein was extracted from HCECs and EBV-infected HCECs as described in Materials and Methods. (A) Western blot analysis of EMT markers (E-cadherin, β-catenin, ZO-1, N-cadherin, Vimentin, Snail, and TCF8/Zeb1), a corneal epithelial cell marker (▵p63), and a corneal differentiation marker (CNX43). (B) Nuclear extracts (left panel) or cytosolic extracts (right panel) were analyzed by Western blot using antibodies against Snail and TCF8/Zeb1. The nuclear marker PARP and cytosol marker β-tubulin were used to check the purity of each fraction. (C) The migratory capacity and invasiveness of HCECs were enhanced by EBV infection, as determined by transwell migration assay kit and BME cell invasion assay kit as described in Materials and Methods. *P < 0.001. Each value is the mean ± SD of 3 determinations. Data are representative of 3 independent experiments.

Regulation of the nuclear localization of transcription factors is a crucial event in response to external stimuli, because transcription factors cannot function until they translocate to the nucleus. ²⁰ We performed immuoblotting to determine that Snail and TCF8/Zeb1 translocated to the nucleus in EBV-HCECs (Fig. 4B). In contrast, uninfected cells expressed small amounts of cytoplasmic Snail and TCF8/Zeb1, but none was detected in the nucleus. These results suggested that EBV infection activated Snail and TCF8/Zeb1, and resulted in their nuclear translocation, potentially leading to diminished epithelial features, enhanced mesenchymal features, and initiation of EMT. 

Because cell motility is a major factor regulating cell invasion and metastasis, we examined the effects of EBV infection on cell motility using a migration and invasion assay, which showed a marked increased cell motility and invasiveness in EBV-HCECs compared to uninfected cells (Fig. 4C). 

PI3K/Akt and Erk Signaling Are Required for EBV-Induced EMT

To understand further the signaling mechanisms involved in EBV-induced EMT, we examined the activation of PI3K/Akt and Erk signaling by EBV infection. Our results showed that the phosphorylation of PI3K-p85, Akt, and Erk was increased markedly in EBV-HCECs compared to uninfected cells (Fig. 5A). To confirm the requirement for PI3K/Akt and Erk activation in EBV-induced EMT, we used the PI3K/Akt inhibitor LY294002 (LY) and the Erk inhibitor PD98059 (PD). Pretreatment with LY or PD changed the morphology of EBV-HCECs from fibroblast-like shapes to epithelial pebble-like shapes (Fig. 5B). In EBV-HCECs, the expression of Snail, Vimentin, TCF8/Zeb1, and N-cadherin was significantly inhibited by LY or PD treatment, whereas E-cadherin, ZO-1, and β-catenin were restored (Fig. 5C). In addition, LY or PD treatment suppressed phospho-Stat3, MMP2, MMP9, and VEGF (Fig. 5D), and blocked EBV-induced migration and invasion (Fig. 5E). These finding suggested that PI3K/Akt and Erk signaling have a role in the initiation and maintenance of EMT by EBV infection in HCECs. 

Figure 5

 

The EBV activates PI3K/Akt and Erk signaling, key of EMT. (A) EBV-infected and uninfected HCECs were harvested and subjected to Western blot analysis with the indicated antibodies. The EBV infection increased the activity of PI3K/Akt and Erk1/2. (B–E) The inhibition of PI3K/Akt or Erk1/2 signaling pathways suppressed EBV-mediated EMT. Supplementary Figure S1A shows data that confirm the inhibition of PI3K/Akt or Erk1/2 activity by LY294002 or PD98059, respectively. (B) Morphologic changes in EBV-infected cells from fibroblast-like, spindle-shaped cells to an epithelial cobblestone-like monolayer as seen by inverted phase-contrast microscope (Olympus). Photographs were taken at ×100 magnification by a digital camera. (C, D) Cells were subjected to Western blot analysis with the indicated antibodies. (E) The migratory capacity and invasiveness of EBV-infected HCECs were inhibited by blocking the PI3K/Akt or Erk1/2 pathway. *P < 0.001. Data are representative of 3 independent experiments.

Syk and Src Signaling Result in EBV-Induced PI3K/Akt and Erk Activation, and EMT

Our results showed that phosphorylation of Syk and Src, but not Fyn and Lyn, were remarkably increased in EBV-HCECs compared to uninfected cells (Fig. 6A). To confirm the requirement for Src and Syk activation in EBV-induced EMT development, we used Src inhibitor PP1 and Syk inhibitor Bay-61-3606. The PP1 or Bay-61-3606 treatment changed the morphology of EBV-HCECs from fibroblast-like shapes to epithelial pebble-like shapes (Fig. 6B), and inhibited the expression of TCF8/Zeb1, Snail, Vimentin, and N-cadherin, while restoring E-cadherin and β-catenin (Fig. 6C). The PP1 or Bay-61-3606 treatment suppressed phosphorylated PI3K/Akt and Erk (Fig. 6D). 

Figure 6

 

TheEBV-induced PI3K/Akt and Erk1/2 activation, and EMT development depended on Src and Syk. (A) The EBV-infected and uninfected HCECs were harvested and subjected to Western blot analysis with the indicated antibodies. EBV infection increased the activity of Src and Syk, but not Fyn and Lyn. (B–D) The inhibition of Src or Syk signaling pathways suppressed EBV-mediated EMT, and blocked the activation of PI3K/Akt and ERK1/2 signaling pathways. The EBV-infected HCECs were treated with 200 nM of the Src inhibitor PP1 or 200 nM of the Syk inhibitor BAY-61-3606 for 48 hours. Supplementary Figure S1B shows data that confirm the inhibition of Src or Syk activity by PP1 or BAY-61-3606, respectively. (B) Morphologic changes in EBV-infected cells from fibroblast-like, spindle-shaped cells to an epithelial cobblestone-like monolayer as seen by inverted phase-contrast microscope (Olympus). Photographs were taken at ×100 magnification by a digital camera. (C, D) Cells were subjected to Western blot analysis with the indicated antibodies. Data are representative of 3 independent experiments.

Recently, the kinases Src, Fyn, and Lyn have been reported to take part in regulating EMT. ^21–23 Previous reports have shown that EBV induces the activation of Akt, Erk, and Syk in B cells and epithelial cells, ²⁴ and that Akt and Erk are downstream signals from Syk and Src. ^25,26 These previous publications and our data indicated that Src and Syk signaling is upstream from PI3K/Akt and Erk, and may have a role in maintaining EMT in EBV-HCECs. 

Inhibition of TGF-β1 Signaling Reverses EMT and Induces Mesenchymal-to-Epithelial Differentiation

To test the functional relevance of the activation of TGF-β1 signaling, we investigated the effect of a dual TGF-β R1/R2 inhibitor (LY2109761) or anti–TGF-β1 neutralizing antibody on the differentiation of EBV-HCECs. We measured the expression of the TGF-β1 pathway component Smad2/3, which was not increased in EBV-HCECs (Fig. 7A). The LY2109761 or anti–TGF-β1 neutralizing antibody treatment changed the morphology of EBV-HCECs from fibroblast-like shapes to epithelial pebble-like shapes (Fig. 7B) and suppressed the expression of TCF8/Zeb1, Snail, Vimentin, and N-cadherin, while restoring E-cadherin and β-catenin (Fig. 7C). The LY2109761 or anti–TGF-β1 neutralizing antibody treatment inhibited the EBV-induced phosphorylation of Src and Syk (Fig. 7D). These results suggested that EBV activates Smad-independent pathways and that TGF-β1 has a role in initiating EMT. 

Figure 7

 

Inhibition of TGF-β1 signaling reverses EMT and induces mesenchymal-to-epithelial differentiation in HCECs. (A) The EBV-infected and uninfected HCECs were harvested and subjected to Western blot analysis with the indicated antibodies. The EBV infection did not affect the phosphorylation of Smad2/3. (B–D) The inhibition of TGF-β1 signaling pathways suppressed EBV-mediated EMT, and prevented the activation of Syk and Src signaling. The EBV-infected HCECs were treated with 100 nM of the dual TGF-β receptor I and II kinase inhibitor, LY2109761, for 48 hours. The EBV-infected HCECs were cultured with anti-TGF-β1 neutralizing antibody (5 μg/mL) or mouse IgG1 antibody (5 μg/mL) for 48 hours. (B) Morphologic changes in EBV-infected cells from fibroblast-like, spindle-shaped cells to an epithelial cobblestone-like monolayer as seen by inverted phase-contrast microscope (Olympus). Photographs were taken at ×100 magnification by a digital camera. (C, D) Cells were subjected to Western blot analysis with the indicated antibodies. Data are representative of 3 independent experiments.

Discussion

Corneal fibrosis with pannus formation is the third leading cause of blindness worldwide. Fibrosis is a common pathologic event observed in chronic diseases of the kidney, liver, and lung. It has been reported that the cellular origins of pathogenic fibroblasts include embryogenesis, tissue-specific epithelial cells, and circulating cells. ^1,4 Recent studies have demonstrated EMT in a rabbit limbal explant, ²⁶ as well as in the fibrotic process of human pterygium. ²⁷ Particularly, epithelial keratinization and peculiar subepithelial fibrosis can bring about visual impairment, such as ocular fibrosis with stem cell deficiency. ²⁸ Corneas with stem cell deficiencies are characterized by neovascularization and chronic inflammation. 

Usually, the fibrosis events in corneal subepithelium are initiated by inflammatory cytokine-mediated corneal stromal fibroblast activation. ^29,30 In particular, keratitis has been shown to involve the mediators, such as MCP-1, TGF-β1, TNF-α, and VCAM-1. This activation then may lead to transdifferentiation into α-SMA–expressing myofibroblasts. ³¹ Shrinkage of the stress fibers in myofibroblasts subsequently results in the development of fibrosis. ^30,31 Our results indicated that EBV infection caused enhanced secretion of TGF-β1, MCP-1, IL-6, IL-8, VEGF, and TNF-α (Fig. 3), and EBV infection increased the cell proliferation compared to normal HCEC cell (data not shown). The expression and secretion of TGF-β1 in HCECs has been controversial. The TGF-β1 mRNA was not expressed in HCEC in the one study ³² ; however, other studies showed that protein and mRNA of TGF-β1 were present. ^33,34 These studies investigated the expression and secretion of TGF-β1 using immunohistochemistry and RT-PCR analysis. We mainly performed the quantitative analysis using real time PCR, Western blotting, and ELISA. Further study is required to clarify the TGF-β1 secretion in uninfected HCECs. 

The EMT is believed to have an essential role in several developmental processes, including the initiation of fibrosis, embryogenesis, and tumor progression. ^1–3 The EMT-related signaling pathways are, indeed, critical to understanding corneal fibrosis. Unlike fibrotic sites in other organs, myofibroblasts in the lens result exclusively from EMT. ³⁵ In our study, we found that EBV induced EMT in primary HCECs, as evidenced by an increase in mesenchymal markers and a concomitant decrease in epithelial markers. Different EMT regulators show diverse expression and roles in a variety of cells. In human hepatocytes, Snail, Vimentin, Twist, and Slug are upregulated and associated with EMT progression. ³⁶ In gastric cancer, different patterns of expression of Snail, Zeb2, and Twist according to histologic subtype are reported. ³⁷ Twist and Slug are upregulated and involved in breast cancer progression, whereas Snail is downregulated. ³⁸ In prostate cancer, Snail and Slug cause a decrease in E-cadherin expression, and Twist results in further reduction. ³⁹ More recent studies show that TGF-β induces EMT via upregulation of Snail in retinal pigment epithelial cells ⁴⁰ and via upregulation of TCF8/Zeb1 in renal tubular epithelial cells. ⁴¹ In our study, Snail and TCF8/Zeb1 were clearly overexpressed in EBV-HCECs compared to uninfected HCECs (Fig. 4). These results suggested that Snail and TCF8/Zeb 1 may be involved in EMT processes. 

Next, we investigated a signaling cascade that controls inflammatory cytokines and induces EMT. A recent study demonstrated that inflammation-induced cell migration and invasion occur through PI3K/Akt-dependent Snail expression. ⁴² In addition, the activation of Erk positively regulates the expression of Snail and Slug. ⁴³ We observed that Akt/Erk activation is required for Snail and TCF8/Zeb1 transcription in EBV-HCECs. Moreover, the inhibition of PI3K/Akt and Erk signaling significantly suppressed the expression of Snail and TCF8/Zeb1, migration, and invasion, suggesting that PI3K/Akt and Erk signaling has an essential role for EBV-stabilized Snail and TCF8/Zeb1 expression in EBV-HCECs. 

Src kinase has been reported to have a role in EMT initiation, cytoskeleton rearrangement, cell motility, and expression of mesenchymal markers by many stimulators. ^44,45 Src suppression allows restoration of E-cadherin, and inhibits expression of Vimentin ⁴⁶ and other mesenchymal markers. ⁴⁵ Syk kinase is associated with cell migration and invasion in various epithelial cells. ⁴⁷ However, whether Syk and Src are involved in EBV-induced EMT in corneal epithelial cells has not yet been elucidated to our knowledge. Our data showed that EBV activated Src and Syk, and that their inhibition restored E-cadherin, and blocked the expression of Snail, TCF8/Zeb1, and Vimentin (Fig. 6), indicating that Src and Syk activation is essential for EBV-induced EMT. We found that EBV-induced EMT was regulated by the production of TGF-β1 followed by the activation of Syk/Src and Akt/Erk. These findings provided a detailed understanding of the TGF-β-Syk/Src-Akt/Erk signaling pathway linking EBV infection to EMT. Furthermore, we proposed Syk/Src and Akt/Erk as potential intervention sites to protect against the EBV-induced corneal injury associated with EMT. 

The TGF-β–Smad signaling is a major event in the development of EMT, fibrosis, and corneal scarring, which prompted us to investigate whether Smad is activated by EBV infection. Contrary to our hypothesis, EBV did not induce activation of Smad2/3. The Akt and Erk activation appear to be independent of Smad activation for TGF-β–mediated EMT. ¹¹ Erk is required for the loss of cell-to-cell contact and induction of cell migration capacity by TGF-β. Our data are consistent with a previous study ⁴⁸ showing that inhibition of TGF-βR activity prevents TGF-β1–induced PI3K/Akt activation (Fig. 7). 

Based on our research, we have identified a link between EBV infection and EMT, namely a novel causative mechanism for the growth potential of primary HCECs during viral infection. However, there must be differences between what we found in vitro and the assumption of what is going in vivo. In addition, EBV–induced keratitis and keratoconjunctivitis are rare clinically. Blood–ocular barrier and immunologic balance may contribute to the differences, and EMT-related mechanisms appear to occur in concert with ocular disease progression, including keratitis and oncogenesis. Further study is required to clarify the role of EBV infection in TGF-β1 secretion and EMT. In summary, our findings suggested that an EBV-induced TGF-β1-Syk/Src-Akt/Erk signaling pathway is critical to the malignant and invasive potential of HCECs by promoting EMT, and serves as an attractive therapeutic target for ocular disease (Fig. 8). This study provided further information to guide therapeutic approaches to block EBV-induced EMT and to treat corneal disease in general. 

Figure 8

 

Proposed model depicting the signaling pathway of EBV-induced EMT in human corneal epithelial cells. The EBV infection induces the secretion of active TGF-β1, and increases the activity of Src, Syk, PI3K/Akt, and Erk1/2. These signaling molecules change EMT markers, namely by increasing the expression of N-cadherin, Vimentin, Snail, TCF8/Zeb1, and CNX43, while decreasing the expression of E-cadherin, ZO-1, β-catenin, and ▵p63. The TGF-β1 can activate Src and Syk through a Smad2/3-independent pathway. The activated Src and Syk pathway then regulate the PI3K/Akt and Erk1/2 pathway.

Supplementary Materials

pdf S1 

Acknowledgments

Supported by Grant A120006 from the Korea Healthcare Technology R&D Project, Ministry of Health & Welfare Affairs, Republic of Korea, and Grant 0920040 from the National R&D Program for Cancer Control, Ministry for Health, Welfare and Family Affairs, Republic of Korea. 

Disclosure: G.B. Park, None; D. Kim, None; Y.S. Kim, None; S. Kim, None; H.-K. Lee, None; J.W. Yang, None; D.Y. Hur, None