Role Played by Receptors for Advanced Glycosylation End Products in Corneal Endothelial Cells after HSV-1 Infection

Senescence, sterile inflammation, and infection cause dysfunction of corneal endothelial cells, leading to visual morbidity that may require corneal transplantation. With increasing age, the extracellular matrix is modified by non-enzymatic glycation forming advanced glycation end products (AGEs). The modifications are primarily sensed by the receptors for the AGEs (RAGE) and are manifested as a type I interferon response. Interestingly, in our study, human corneal endothelial cells (HCEn) cells did not respond to the typical RAGE ligands, including the AGEs, high mobility group box 1 (HMGB1), and serum amyloid-A (SAA). Instead, HCEn cells responded exclusively to the CpG DNA, which is possessed by typical corneal pathogen, herpes simplex virus-1 (HSV-1). Upon HSV-1 infection, the surface expression of RAGE was increased, and endocytosed HSV-1 was associated with RAGE and CpG DNA receptor, TLR9. RAGE DNA transfection markedly increased interferon-β secretion by CpG DNA or HSV-1 infection. HSV-1 infection-induced interferon-β secretion was abolished by TLR9 inhibition and partially by RAGE inhibition. Global transcriptional response analysis confirmed that RAGE and TLR9 were both significantly involved in type I interferon responses. We conclude that RAGE is a sensor of HSV-1 infection and provokes a type I interferon response.


Introduction
Corneal opacity is one of the major causes of blindness in the world [1,2]. In 2005, 1.3 million people were affected by corneal blindness, which is defined by a best-corrected visual acuity (BCVA) of <20/400 [1,2]. Important causes of corneal blindness in developing countries are active keratitis and corneal scars [3]. In addition, bullous keratopathy, failed corneal grafts, and corneal dystrophy have become more prevalent as causes of corneal blindness [3].
The corneal endothelium is an essential component for maintaining corneal clarity, and it does this by governing the fluid exchange and dehydration of the cornea. A loss or damage of corneal endothelial cells leads to bullous keratopathy that then results in severe visual impairment.
Atherosclerosis, cardiovascular diseases, age-related macular degeneration, and Alzheimer's disease are prevalent in older individuals [4]. All of these pathologies have been shown to be associated with the formation of advanced glycation endproducts (AGEs) [5][6][7]. AGEs are formed in vivo as non-enzymatic heterogeneous compounds from the oxidization of glucose and lipids in the aging processes or in oxidative environments [5][6][7]. AGEs are resistant to proteasomal degradation and need to be removed to avoid unnecessary stress responses or tissue malfunctions [6,7].
In the eyes of elderly individuals, the AGEs are present in the cornea, especially in Descemet's membrane and corneal endothelial cells [8], and it has been suggested that they are important factors associated with the dysfunction of corneal endothelial cells.
The receptor for the AGEs (RAGE) is a pattern-recognition receptor that belongs to the immunoglobulin superfamily. AGE-bound RAGE is endocytosed for lysosomal degradation and systemic clearance from the body. RAGE is also involved in the regulation of complex homeostatic processes, and it is not limited to a role as a scavenger receptor of age-related oxidative products. RAGE also recognizes other molecules or damage-associated molecular patterns [9]. These ligands are released or are formed during infections, inflammation, aging, and cellular stress, and they induce pro-inflammatory responses. Thus, RAGE regulates various cellular processes, including inflammation, antimicrobial responses, apoptosis, cell adhesion, stress responses, and autophagy.
An important function of corneal endothelial cells is to maintain the clarity of the cornea. Corneal endothelial insufficiency is often caused by surgical intervention, Fuchs' endothelial corneal dystrophy (FECD), or viral infection during senescence. This leads to significant visual morbidity and bullous keratopathy [3,8,[10][11][12]. Importantly, AGE is known to accumulate in corneal endothelial cells of patients with FECD [8]. However, it has not been determined how corneal endothelial cells use RAGE to respond to various adverse conditions.
Herpes simplex virus type1 (HSV-1) is a common pathogen that infects the cornea, and more than one-half of elderly individuals are seropositive for HSV-1 [13]. In healthy subjects, a spontaneous shedding of HSV-1 is often observed in the tears [14]. HSV-1 reactivation causes a loss in corneal clarity as well as corneal endothelial cell loss in elderly individuals. In vitro studies have shown that corneal endothelial cells are highly permissive to HSV-1 infections [15,16].
To understand the role played by the RAGE in corneal endothelial cells, the responsiveness of human corneal endothelial cells (HCEn) to senescence-related molecules, inflammation, and viral pathogens was examined. A viral infection or inflammation often culminates in interferon production [15,16].
Thus, the purpose of this study was to determine the contribution of the RAGE ligands to the type I interferon responses of corneal endothelial cells, and to analyze the role that RAGE plays in maintaining the integrity of corneal endothelial cells.

Results
To examine the roles played by the RAGE in corneal endothelial cells, we tested a panel of RAGE ligands on their ability to stimulate HCEn cells to produce interferon-β ( Figure 1). The panel included AGE, high mobility group box 1 (HMGB1), CpG oligonucleotide, and serum amyloid-A (SAA). Our results showed that AGE, HMGB1, and SAA did not significantly stimulate interferon-β expression. However, the CpG oligonucleotide significantly stimulated the expression of interferon-β.
HSV-1 is the most common pathogen causing corneal endothelial inflammation [17], and the HSV-1 genome contains CpG DNA sites [18]. Therefore, we hypothesized that corneal endothelial cells use RAGE to recognize HSV-1 infections and initiate antiviral interferon responses. We first examined whether HSV-1 infection can transcriptionally induce the expression of the RAGE. To accomplish this, HCEn cells were infected with HSV-1, and the induction of the mRNA of RAGE was determined by real-time RT-PCR ( Figure 2A). The results showed that HSV-1 infection stimulated the induction of RAGE at 6 h post-infection (PI), and a 60-fold increase in the induction was observed at 24 h PI, compared to mock infections. Exposure to CpG DNA significantly induces interferon-β expression after 12 h (left). Advanced glycation end products (AGE), HMGB1, and serum amyloid A (SAA) (right) do not significantly induce interferon-β. DMEM with 10% FBS was used for the assay media. Left and right panels were obtained from different batches of cells of single origin. For * p < 0.00001; n = 6. Ns: not significant.
HSV-1 is the most common pathogen causing corneal endothelial inflammation [17], and the HSV-1 genome contains CpG DNA sites [18]. Therefore, we hypothesized that corneal endothelial cells use RAGE to recognize HSV-1 infections and initiate antiviral interferon responses. We first examined whether HSV-1 infection can transcriptionally induce the expression of the RAGE. To accomplish this, HCEn cells were infected with HSV-1, and the induction of the mRNA of RAGE was determined by real-time RT-PCR ( Figure  2A). The results showed that HSV-1 infection stimulated the induction of RAGE at 6 h post-infection (PI), and a 60-fold increase in the induction was observed at 24 h PI, compared to mock infections.
To determine whether a surface expression of RAGE occurred in HCEn cells, the cell surface was assessed by FACS analysis ( Figure 2B). Consistent with the RAGE mRNA induction, HSV infection induced the surface expression of RAGE, which peaked at 24 h PI.  Responsiveness of human corneal endothelial (HCEn) cells to the receptors of advanced glycation end products (RAGE) ligands assessed by the level of expression of interferon-β. Exposure to CpG DNA significantly induces interferon-β expression after 12 h (left). Advanced glycation end products (AGE), HMGB1, and serum amyloid A (SAA) (right) do not significantly induce interferon-β. DMEM with 10% FBS was used for the assay media. Left and right panels were obtained from different batches of cells of single origin. For * p < 0.00001; n = 6. Ns: not significant.
HSV-1 is the most common pathogen causing corneal endothelial inflammation [17], and the HSV-1 genome contains CpG DNA sites [18]. Therefore, we hypothesized that corneal endothelial cells use RAGE to recognize HSV-1 infections and initiate antiviral interferon responses. We first examined whether HSV-1 infection can transcriptionally induce the expression of the RAGE. To accomplish this, HCEn cells were infected with HSV-1, and the induction of the mRNA of RAGE was determined by real-time RT-PCR ( Figure  2A). The results showed that HSV-1 infection stimulated the induction of RAGE at 6 h post-infection (PI), and a 60-fold increase in the induction was observed at 24 h PI, compared to mock infections.
To determine whether a surface expression of RAGE occurred in HCEn cells, the cell surface was assessed by FACS analysis ( Figure 2B). Consistent with the RAGE mRNA induction, HSV infection induced the surface expression of RAGE, which peaked at 24 h PI. To determine whether a surface expression of RAGE occurred in HCEn cells, the cell surface was assessed by FACS analysis ( Figure 2B). Consistent with the RAGE mRNA induction, HSV infection induced the surface expression of RAGE, which peaked at 24 h PI.
The CpG DNA sites are mainly recognized by TLR9, and HSV is well known to be recognized by TLR9 [19]. Because this suggested roles for RAGE after HSV infection or CpG stimulation, we hypothesized that the RAGE interacts with TLR9. After HSV-1 infection, HSV-1 is endocytosed or fused with the plasma membrane of the infected HCEn cells. Immunohistochemical analyses showed that the GFP-expressing HSV-1 colocalized with the punctate expression of the cell surface RAGE at 1 h PI. GFP-expressing HSV-1 was also colocalized with TLR9 near the cell surface at 3 h PI, which partly overlapped with RAGE expression. The expression of TLR9, which was associated with GFP-HSV-1, was transitioned to the endoplasmic reticulum structure (Figure 3). 0.000), and 149 ± 7 (24 h, p = 0.000). n = 374/group. The CpG DNA sites are mainly recognized by TLR9, and HSV is well known to be recognized by TLR9 [19]. Because this suggested roles for RAGE after HSV infection or CpG stimulation, we hypothesized that the RAGE interacts with TLR9. After HSV-1 infection, HSV-1 is endocytosed or fused with the plasma membrane of the infected HCEn cells. Immunohistochemical analyses showed that the GFP-expressing HSV-1 colocalized with the punctate expression of the cell surface RAGE at 1 h PI. GFP-expressing HSV-1 was also colocalized with TLR9 near the cell surface at 3 h PI, which partly overlapped with RAGE expression. The expression of TLR9, which was associated with GFP-HSV-1, was transitioned to the endoplasmic reticulum structure ( Figure 3). Figure 3. Cellular localization of TLR9, RAGE, and HSV-1. HCEn cells were infected with GFP-HSV (at MOI 50, arrow (green)) and stained for cell surface RAGE (arrow (red): PE-labeled) and TLR9 (Alexa647-labeled, arrow (yellow)) without permeabilization. HSV-1 is colocalized with cell surface RAGE, which partly overlapped with the accumulated TLR9 expression (upper panel). Lower panel: GFP-HSV (arrow (green)) was associated with TLR9 (arrow (yellow)) near cell surface. TLR9 expression was transitioned to endoplasmic reticulum (ER, blue) surrounding nucleus (lower panel). Nuclear transition of GFP-HSV is also observed at 3 h PI. Bar indicates 10 µm.
To confirm whether HSV-1 is associated with RAGE, HCEn cells were infected with GFP-expressing HSV-1, and the cell lysates were assayed for the GFP pull-down assay Figure 3. Cellular localization of TLR9, RAGE, and HSV-1. HCEn cells were infected with GFP-HSV (at MOI 50, arrow (green)) and stained for cell surface RAGE (arrow (red): PE-labeled) and TLR9 (Alexa647-labeled, arrow (yellow)) without permeabilization. HSV-1 is colocalized with cell surface RAGE, which partly overlapped with the accumulated TLR9 expression (upper panel). Lower panel: GFP-HSV (arrow (green)) was associated with TLR9 (arrow (yellow)) near cell surface. TLR9 expression was transitioned to endoplasmic reticulum (ER, blue) surrounding nucleus (lower panel). Nuclear transition of GFP-HSV is also observed at 3 h PI. Bar indicates 10 µm.
To confirm whether HSV-1 is associated with RAGE, HCEn cells were infected with GFP-expressing HSV-1, and the cell lysates were assayed for the GFP pull-down assay ( Figure 4). After HSV-1 infection, GFP pulled down RAGE, confirming that HSV-1 is associated with RAGE. GFP was also associated with TLR9.  Figure 4). After HSV-1 infection, GFP pulled down RAGE, confirming that HSV-1 is associated with RAGE. GFP was also associated with TLR9. To determine the role played by RAGE in HSV-1 infections, we examined whether the RAGE could affect viral replication in cultured HCEn cells. HCEn cells were infected with HSV-1 at MOI 1 and were examined for viral titer by RAGE blockade. An inhibition by an anti-RAGE antibody did not significantly affect the viral replication, e.g., HSV-1 titer with control IgG was 4.4 × 10 7 plaque forming units (PFUs), and that for the anti-RAGE antibody was 4.3 × 10 7 PFU (NS; N = 4).
Next, we analyzed the global transcriptional responses of HCEn cells after HSV-1 infection. To accomplish this, we conducted a network analysis, using the highly induced molecules (more than a 4-fold increase) and pattern recognition receptors, TLR9 and RAGE ( Figure 5).
The top canonical pathway was interferon signaling (p = 1.1 × 10 −11 ), followed by the role of pattern recognition receptors in the recognition of viruses (p = 5.8 × 10 −8 ). Top upstream regulators were type 1 interferon, interferon lambda-1, interferon regulatory factor 7 (IRF7), and STAT1. Thus, the transcriptional network was highly enriched with a type 1 interferon response and pattern recognition receptors, including RAGE and TLR9.
The resultant networks indicated direct associations with RAGE and TLR9 ( Figure 5), and their significant contribution to type I interferon responses in recognition of the virus. To determine the role played by RAGE in HSV-1 infections, we examined whether the RAGE could affect viral replication in cultured HCEn cells. HCEn cells were infected with HSV-1 at MOI 1 and were examined for viral titer by RAGE blockade. An inhibition by an anti-RAGE antibody did not significantly affect the viral replication, e.g., HSV-1 titer with control IgG was 4.4 × 10 7 plaque forming units (PFUs), and that for the anti-RAGE antibody was 4.3 × 10 7 PFU (NS; N = 4).
Next, we analyzed the global transcriptional responses of HCEn cells after HSV-1 infection. To accomplish this, we conducted a network analysis, using the highly induced molecules (more than a 4-fold increase) and pattern recognition receptors, TLR9 and RAGE ( Figure 5).
The top canonical pathway was interferon signaling (p = 1.1 × 10 −11 ), followed by the role of pattern recognition receptors in the recognition of viruses (p = 5.8 × 10 −8 ). Top upstream regulators were type 1 interferon, interferon lambda-1, interferon regulatory factor 7 (IRF7), and STAT1. Thus, the transcriptional network was highly enriched with a type 1 interferon response and pattern recognition receptors, including RAGE and TLR9.
The resultant networks indicated direct associations with RAGE and TLR9 ( Figure 5), and their significant contribution to type I interferon responses in recognition of the virus. We then tested whether RAGE plays a role in the induction of interferon-β. For this, HCEn cells were transfected with RAGE plasmid to overexpress RAGE [20]. The overexpression of RAGE was confirmed by Western blot ( Figure 6A). RAGE-transfected HCEn cells were stimulated with CpG DNA or HSV-1 ( Figure 6B,C), and the supernatant was assayed for interferon-β secretion. The results showed that the induction of RAGE significantly increased the production of interferon-β by CpG DNA ( Figure 6B).
Next, RAGE-transfected HCEn cells were infected with HSV-1, and the supernatant was examined for interferon-β production ( Figure 6C). Again, the RAGE-transfected HCEn cells increased the induction of interferon-β significantly. We then tested whether RAGE plays a role in the induction of interferon-β. For this, HCEn cells were transfected with RAGE plasmid to overexpress RAGE [20]. The overexpression of RAGE was confirmed by Western blot ( Figure 6A). RAGE-transfected HCEn cells were stimulated with CpG DNA or HSV-1 ( Figure 6B,C), and the supernatant was assayed for interferon-β secretion. The results showed that the induction of RAGE significantly increased the production of interferon-β by CpG DNA ( Figure 6B).
Next, RAGE-transfected HCEn cells were infected with HSV-1, and the supernatant was examined for interferon-β production ( Figure 6C). Again, the RAGE-transfected HCEn cells increased the induction of interferon-β significantly. We then evaluated the contribution of RAGE to interferon-β induction by CpG DNA. HCEn cells were stimulated with CpG DNA and were examined for the induction of interferon-β. The RAGE blockade significantly reduced the expression of the mRNA of interferon-β ( Figure 7A) and the protein ( Figure 7B) induced by CpG DNA. We then evaluated the contribution of RAGE to interferon-β induction by CpG DNA. HCEn cells were stimulated with CpG DNA and were examined for the induction of interferon-β. The RAGE blockade significantly reduced the expression of the mRNA of interferon-β ( Figure 7A) and the protein ( Figure 7B) induced by CpG DNA.

Figure 7. Role of RAGE in interferon-β production by CpG oligonucleotide stimulation of corneal endothelial cells. (A) HCEn cells were stimulated by CpG oligonucleotides and were assessed for interferon-β mRNA induction at 12 h by real-time RT-PCR. HCEn cells stimulated by CpG oligonucleotides significantly induced interferon-β. Inhibition by anti-RAGE antibody significantly reduced the induction of interferon-β by CpG oligonucleotide. (B) HCEn cells were stimulated by
CpG and were assessed for interferon-β production at 24 h by ELISA. The interferon-β production by CpG oligonucleotides (50 µM) was significantly reduced by anti-RAGE antibody. n = 6, *: p = 0.01, **: p < 0.001. Ns: not significant.
Next, we evaluated the contribution of RAGE to the induction of interferon-β after HSV-1 infection. HCEn cells were inhibited for the expression of RAGE and were assessed for interferon-β induction after HSV-1 infection. HCEn cells were transfected with RAGE siRNA and were assessed for cell surface expression of RAGE ( Figure 8A). The transfection of siRNA abolished HSV-induced upregulation of the RAGE expression.
When the RAGE siRNA-transfected HCEn cells were infected with HSV-1, the induction of the mRNA of interferon-β was significantly reduced ( Figure 8A).
Next, we examined how TLR9 is involved in the RAGE-mediated interferon-β expression. The level of HSV-1 infection-induced interferon-β was significantly reduced after a RAGE blockade ( Figure 8B). Inhibition of TLR9 abolished the HSV infection-induced interferon-β secretion from HCEn cells. Collectively, RAGE is involved in signaling by HSV-1 infection of interferon-β induction, which requires TLR9.

Figure 7. Role of RAGE in interferon-β production by CpG oligonucleotide stimulation of corneal endothelial cells. (A) HCEn cells were stimulated by CpG oligonucleotides and were assessed for interferon-β mRNA induction at 12 h by real-time RT-PCR. HCEn cells stimulated by CpG oligonucleotides significantly induced interferon-β. Inhibition by anti-RAGE antibody significantly reduced the induction of interferon-β by CpG oligonucleotide. (B) HCEn cells were stimulated by
CpG and were assessed for interferon-β production at 24 h by ELISA. The interferon-β production by CpG oligonucleotides (50 µM) was significantly reduced by anti-RAGE antibody. n = 6, *: p = 0.01, **: p < 0.001. Ns: not significant.
Next, we evaluated the contribution of RAGE to the induction of interferon-β after HSV-1 infection. HCEn cells were inhibited for the expression of RAGE and were assessed for interferon-β induction after HSV-1 infection. HCEn cells were transfected with RAGE siRNA and were assessed for cell surface expression of RAGE ( Figure 8A). The transfection of siRNA abolished HSV-induced upregulation of the RAGE expression.
When the RAGE siRNA-transfected HCEn cells were infected with HSV-1, the induction of the mRNA of interferon-β was significantly reduced ( Figure 8A).
Next, we examined how TLR9 is involved in the RAGE-mediated interferon-β expression. The level of HSV-1 infection-induced interferon-β was significantly reduced after a RAGE blockade ( Figure 8B). Inhibition of TLR9 abolished the HSV infection-induced interferon-β secretion from HCEn cells. Collectively, RAGE is involved in signaling by HSV-1 infection of interferon-β induction, which requires TLR9.

Discussion
Apart from aging, viral endotheliitis is an important risk factor for corneal endothelial cell loss, which then leads to bullous keratopathy [21]. Both aging and viral infections, including HSV-1 and cytomegalovirus, are strongly associated with type I interferon responses manifested as inflammation. Alternatively, the natural course of senescence can cause cell damage as sterile inflammation.
We examined how RAGE contributed to endothelial cell defense and showed that an important aspect of RAGE was as a sensor of HSV-1 infection and the induction of type 1 interferon responses. Viral corneal endotheliitis is typically caused by infections by HSV-1, CMV, or varicella-zoster virus [10][11][12]22]. The viral infections are initially recognized by pattern recognition receptors as part of the innate immune system, including TLR-like receptors, NODlike receptors, and scavenger receptors [23]. The signaling by these receptors typically results in type I interferon responses or NF-κB activation. When HSV-1 was assessed for recognition by pattern recognition receptors of corneal endothelial cells, TLR9 played a pivotal role in inducing the interferon response [16]. However, TLR9 is located endosomally and requires signaling input from the cell surface receptors. We showed that the cell surface RAGE contributed to HSV recognition by corneal endothelial cells and route signaling to TLR9. AGE can be formed by the collagen or elastin of the extracellular matrix in the skin of elderly individuals [24]. This can also occur in the Descemet's membrane of diseased cornea or FECD [8], and the altered extracellular matrix environment can signal back to the endothelial cells through RAGE [25].
In the cornea of elderly FECD patients, AGE is known to accumulate abundantly in the stressed or thickened anterior portion of basement membrane (Descemet's membrane)

Discussion
Apart from aging, viral endotheliitis is an important risk factor for corneal endothelial cell loss, which then leads to bullous keratopathy [21]. Both aging and viral infections, including HSV-1 and cytomegalovirus, are strongly associated with type I interferon responses manifested as inflammation. Alternatively, the natural course of senescence can cause cell damage as sterile inflammation.
We examined how RAGE contributed to endothelial cell defense and showed that an important aspect of RAGE was as a sensor of HSV-1 infection and the induction of type 1 interferon responses. Viral corneal endotheliitis is typically caused by infections by HSV-1, CMV, or varicellazoster virus [10][11][12]22]. The viral infections are initially recognized by pattern recognition receptors as part of the innate immune system, including TLR-like receptors, NOD-like receptors, and scavenger receptors [23]. The signaling by these receptors typically results in type I interferon responses or NF-κB activation. When HSV-1 was assessed for recognition by pattern recognition receptors of corneal endothelial cells, TLR9 played a pivotal role in inducing the interferon response [16]. However, TLR9 is located endosomally and requires signaling input from the cell surface receptors. We showed that the cell surface RAGE contributed to HSV recognition by corneal endothelial cells and route signaling to TLR9. AGE can be formed by the collagen or elastin of the extracellular matrix in the skin of elderly individuals [24]. This can also occur in the Descemet's membrane of diseased cornea or FECD [8], and the altered extracellular matrix environment can signal back to the endothelial cells through RAGE [25].
In the cornea of elderly FECD patients, AGE is known to accumulate abundantly in the stressed or thickened anterior portion of basement membrane (Descemet's membrane) of corneal endothelial cells accompanied by endothelial cell loss [8]. This suggests that accumulated AGEs can cause significant cellular stress that contributes to endothelial cell loss. Accumulated AGEs need to be removed by scavenger receptors, including RAGE. However, eyes with FECD are impaired in the expression of RAGE. In normal elderly individuals, AGE is also observed in the cornea; however, its deposition in the corneal endothelium is limited to a scattered manner.
Infections by HSV-1 also require cell surface receptors; heparan sulfate on the cell membrane initially binds glycoproteins B and C of the HSV-1 envelope and this plays a role in the initial attachment [29]. Interestingly, heparan sulfate has been proposed as a co-receptor of RAGE binding to HMGB1 [30].
We observed an increase in the surface expression of RAGE in HCEn cells after HSV-1 infection, and this may allow their higher responsiveness. TLR9 is indispensable for sensing CpG DNA or HSV-1 infections even when RAGE is functioning. This is consistent with the findings that cell surface RAGE is located upstream of endosomal TLR9, and RAGE appears to help route the viral stimulus to endosomal TLR9. This can then decrease the threshold of cellular activation and allow responses to viral infections [31].
Another property of RAGE is its ability to be directly associated with released DNA or RNA from damaged cells on the cell surface [32]. This is in marked contrast to TLR9, which is located in the endosome and appears to not be readily activated.
The RAGE in corneal endothelial cells plays multiple roles, including that of scavengers and antiviral receptors. For example, IL-6 is a typical scavenger receptor-induced cytokine. Scavenger receptors type A, including COLEC12, MARCO, MSR1, SCARA3, and SCARA5, induce IL-6 ( Figure 5). We observed that AGE stimulation of corneal endothelial cells induced IL-6 ( Figure 5). However, AGE did not stimulate interferon-β ( Figure 1). The RAGE-induced-interferon-β appears restricted to TLR9-stimulating input. The presumed dual roles of RAGE may affect the endothelial physiology. In FECD patients, the expression of RAGE in the corneal endothelial cells is reduced [8]. This impairs its scavenger function as well as the antiviral responses when they are affected by a primary or reactivated HSV-1 infection.
The type 1 interferon response is an efficient antiviral response; however, chronic and intermittent responses may become damaging for corneal endothelial cells. One way for this is through collateral damage by type 1 interferon-activated cytotoxic effector CD8 + T cells [33]. Another way is the promotion of cell senescence due to a shortening of the telomere length by prolonged interferon stimulation [34].
To treat viral infection-related endothelial inflammation, topical antiviral drugs are suggested to prevent progressive endothelial cell loss. Failure to control viral activity can lead to the need for corneal endothelial cell transplantation. Currently, it remains unclear whether strategies to remove AGE or prevent RAGE signaling may have clinically beneficial effects. This requires future clinical trials.
Collectively, RAGE not only functions as a recognizer of age-related oxidized products or AGEs, but it also contributes to antiviral responses as the innate immune arm.
There are some limitations in this study. We reported that the cytokine responses of our HCEn cells were very similar to those of primary cultured corneal endothelial cells [33,35,36], and confirmed RAGE expression or strong induction of type 1 interferon by primary endothelial cells after HSV-1 infection or TLR9 stimulation (data not shown). We are aware that HCEn cells may not exactly mirror the physiological functions of primary endothelial cells. However, a subculturing of primary corneal endothelial cells is difficult, as only low-passaged cells replicate. This significantly affects the reproducibility of HSV-1 infections after plasmid transfection or manipulation.
The function of RAGE is presumably context-dependent, and a complete picture of the contributions of RAGE to diseases appears more complex than believed. Importantly, RAGE recognized CpG DNA in conjunction with TLR9 and induced antiviral interferon responses in corneal endothelial cells. This information may help develop therapy to treat viral corneal endotheliitis or senescence-related endothelial cell dysfunction.

Virus
HSV-1 viral stocks were prepared from confluent monolayers of Vero cells infected with the KOS strain or GFP-KOS strain (generously gifted from Kozaburo Hayashi and David Tscharke) of HSV-1 [42]. Purified viral stocks were prepared using a sucrose density gradient as described in detail [42]. The samples were aliquoted and stored at −80 • C until use.

ELISA
The level of interferon-β in the supernatant was measured with an ELISA kit (ANTI-GENIX America, Huntington Station, NY, USA).

Network Analysis of Corneal Endothelial Cell Transcriptome after HSV-1 Infection
HCEn cells were infected with HSV-1 (KOS strain) at MOI 1 and the mRNAs were extracted 12 h post-infection [15]. The transcriptome data obtained using whole human genome microarray (Agilent Technologies, Santa Clara, CA, USA) were analyzed [15]. To identify the canonical pathways and transcriptional networks that were most significant to the dataset, the data were analyzed by the Ingenuity Pathway Analysis (IPA; 2018 winter version, QIAGEN Inc. accessed March 2019)

Statistical Analyses
Data are presented as the means ± standard error of the means (SEMs). The significance of the differences was determined by t tests or ANOVA and post hoc tests.