The Necroptosis Pathway Is Upregulated in the Cornea in Mice With Ocular Graft-Versus-Host Disease

Abstract

Purpose: To identify molecular signatures specific for ocular graft-versus-host disease (GVHD) by proteomic analysis of corneas from mice with GVHD.

Methods: We identified differentially expressed proteins (DEPs) in corneal samples from GVHD model mice and syngeneic control mice 4 weeks after bone marrow transplantation. Data-independent acquisition analysis was performed on individual samples, and the roles of DEPs in biological pathways related to GVHD were evaluated via bioinformatics and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses.

Results: Three important signaling pathways were upregulated in the cornea in mice with GVHD: (1) the necroptosis pathway, (2) the mitogen-activated protein kinase (MAPK) pathway, and (3) as previously reported, the neutrophil extracellular trap (NET) pathway. In those signaling pathways, we identified new upregulated molecules, including (1) receptor-interacting protein kinase 1 (RIPK1), RIPK3, interferon regulatory factor 9, the interferon-induced double-stranded RNA-activated protein kinase lipoxygenase, and high mobility group box1 (HMGB1) which are damage-associated molecular patterns (DAMPs) in the necroptosis pathway; (2) the sequentially upregulated interleukin 1 (IL-1) receptor-associated kinase (IRAK), an evolutionarily conserved signaling intermediate in the Toll pathway (ECSIT), and p38, which is downstream of the IL-1 receptor and increased CDC42/Rac (Rac2), a Rho family GTPase in the MAPK pathway; and (3) the integrin components CR3 and macrophage-1 antigen (MAC-1), which are DAMPs, and the pyroptosis-related protein gasdermin D (GSDMD) in the NET pathway.

Conclusions: These novel molecules may help researchers elucidate the pathogenesis of GVHD and identify new therapeutic targets for corneal changes in patients with ocular GVHD.

Chronic ocular graft-versus-host disease (GVHD) is a long-term complication of hematopoietic stem cell transplantation (HSCT). More than 50% of patients develop ocular GVHD after HSCT, and quality of life and vision are negatively affected in 60% to 90% of patients with systemic GVHD.^1–4 Ocular GVHD is characterized by progressive and fibroinflammatory lesions that can lead to irreversible and dysfunctional damage in the eye.⁵ Early detection, prevention, and treatment of ocular GVHD and assessment of disease severity must be performed for patients after HSCT.⁶ Therefore, we need to explore the possible biological pathways involved in ocular GVHD.⁵ 

Corneal complications of ocular GVHD include corneal epitheliopathy, filamentary keratitis, ulcers, persistent epithelial defects, limbal stem cell deficiency, thinning, scarring, melting, perforation, conjunctivalization, and neovascularization; these complications are caused by various factors, including infection, inflammation, immunity, and neurotrophic responses.^7–12 Severe cases can result in significantly impaired vision or, potentially, blindness; however, the molecular pathology of ocular GVHD in the cornea is largely unknown. Exploring potential surrogate markers or therapeutic targets for ocular GVHD is crucial to address these unmet medical needs. 

Proteomics, an essential approach for identifying protein signatures, provides mechanistic insight into disease etiology at the molecular level.¹³ This approach has been used to analyze tear fluid in patients with ocular diseases, including dry eye disease.^13,14 However, detailed corneal analyses in patients with ocular GVHD have not been conducted. In this study, we aimed to identify the proteomic signature and biological pathways in the cornea via a chronic GVHD (cGVHD) mouse model. We identified three key signaling pathways upregulated in the corneas of mice with GVHD: (1) the necroptosis pathway, (2) the mitogen-activated protein kinase (MAPK) pathway, and (3) as previously reported, the neutrophil extracellular trap (NET) pathway. In addition, we identified novel molecules and their interactions within these pathways in the corneas of mice with GVHD. 

Methods

Study Approval

All of the experimental procedures were performed in accordance with the institutional guidelines of animal experimentation at Keio University School of Medicine and with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Our protocols were approved by the Institutional Animal Care Unit Committee, Keio University School of Medicine (A2022-178), and Yonsei University College of Medicine (IACUC approval no. 2019-0142). 

Mice

The B10.D2/nSnSlc (H2^d) and BALB/cCrSlc mice (H2^d) (7–9 weeks old) were purchased from Sankyo Laboratory, Inc. (Tokyo, Japan). 

Whole Bone Marrow Transplantation

Allogeneic bone marrow transplantation (BMT) was performed with 7- to 9-week-old male B10.D2/nSnSlc (H2^d) and female BALB/cCrSlc (H2^d) mice as donors and recipients, respectively, for the GVHD group. For the non-GVHD control group, syngeneic BMT was conducted between male and female BALB/cCrSlc mice¹⁵ (Fig. 1A). Details are provided in the Supplementary Materials. 

Figure 1.

 

Bone marrow transplantation (BMT) and methods for preparing samples for proteomic analysis. (A) Procedure for BMT (left). The arrows indicate key events in the experimental timeline (bottom). Diarrhea, one of the symptoms of GVHD, occurred 3 weeks after transplantation (middle). Corneal phenotype as determined via bright field or fluorescein staining at 4 weeks after transplantation is shown (right). The mice in the GVHD group developed eyelid edema and blepharitis (a, blue arrow), diffuse superficial keratitis (b, orange arrow) or corneal ulcers (c, red arrow), but the mice in the syngeneic control group did not (d, e). B10.D2 mouse are shown in black, and the BALB/c mouse are shown in white. (B) Procedure for sample preparation for DIA analysis (left). Procedure for sample preparation for DDA analysis (right). DB, data base.

Sampling Methods

Whole cornea samples from five cGVHD model mice and five syngeneic control mice were collected and stored at –80°C at 4 weeks post-BMT. Corneal tissues were separated from the whole eye and subjected to proteomic analysis via liquid chromatography–tandem-mass spectrometry (LC-MS/MS) (Fig. 1B). 

Protein Extraction From Corneal Tissues

Corneal tissues were individually cryopulverized with a cryoPREP device (CP02; Covaris, Woburn, MA, USA).¹⁶ In brief, each tissue sample was placed in a cryovial (430487; Covaris) on dry ice and subsequently transferred to a tissue bag (TT1; Covaris). The tissue bag was placed into liquid nitrogen for 30 seconds, and the tissue sample was pulverized. The tissue powder was then placed in a sonication tube (002109; Covaris) and mixed with lysis buffer (8-M urea in 0.1-M Tris-HCl, pH 8.5). Tissue lysis was performed by sonication using a focused ultrasonicator (S220; Covaris) at a setting of 2 W (intensity 5) for 5 seconds followed by 36 W (intensity 10) for 20 seconds and 0 W (intensity 0) for 10 seconds. The homogenate was centrifuged at 16,000g and 20°C for 10 minutes (5810 R; Eppendorf, Hamburg, Germany), and the supernatant was transferred to a new tube. The protein concentration was then measured via a bicinchoninic acid (BCA) protein assay (Pierce BCA Protein Assay Kit; Thermo Fisher Scientific, Waltham, MA, USA). 

In-Solution Digestion

Total protein in each sample (100 µg) was subsequently digested into peptides via an in-solution digestion method. In brief, the mixture was incubated for 30 minutes at room temperature for denaturation. Then, 10-mM dithiothreitol for reduction and 30-mM iodoacetamide for alkylation were used for protein denaturation. Trypsin was added at a 50:1 (w/w) protein-to-trypsin ratio, and the mixture was incubated at 37°C overnight. Trypsin activity was quenched with 0.4% trifluoroacetic acid, and the peptides were desalted on a C18 macro SpinColumn (Harvard Apparatus, Holliston, MA, USA). The resulting peptides were dried and stored at −80°C. 

Data-Independent Acquisition and Data-Dependent Acquisition Mass Spectrometric Analyses

For data-dependent acquisition (DDA) analysis (Fig. 1B right), peptides were resuspended in 0.1% formic acid in water and analyzed with a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) coupled to an EASY-nLC 1000 system (Thermo Fisher Scientific). Solvents A and B were 0.1% formic acid in water and 0.1% formic acid in acetonitrile, respectively. The peptides were loaded onto a trap column (75 µm × 2 cm, 3 µm, C18, 100 Å) and ionized via an EASY-Spray column (50 cm × 75 µm inner diameter) packed with 2 µm C18 particles at a voltage of 2.0 kV. A 180-minute gradient (increasing from 10% to 35% solvent B over 120 minutes, increasing from 35% to 50% solvent B over 10 minutes, holding at 95% solvent B for 10 minutes, and holding at 5% solvent B for 20 minutes) was used. The raw data were processed via Proteome Discover 2.4 software. A false discovery rate (FDR) cutoff of 1% was applied at the peptide–spectrum match and protein levels. 

For the data-independent acquisition (DIA) analysis (Fig. 1B left), the components of a retention time kit (iRT Kit; Biognosys, Schlieren, Switzerland) were used to spike samples at a ratio of 1:20 (v/v), and 2 µg of each peptide sample was analyzed with the Q Exactive Plus mass spectrometer coupled to the EASY-nLC 1000 ultra-high-performance liquid chromatography system over a mass range of 500 to 900 m/z with a resolution of 170,000 at 200 m/z. Twenty optimal acquisition windows covering a mass range of 500 to 900 m/z were used. The DIA data were analyzed with Biognosys Spectronaut Pulsar (version 11.0.15038.4.29119) via a search archive spectral library, and the default settings were used for targeted analysis. In brief, a dynamic window for the generation of extracted ion chromatograms and a nonlinear iRT calibration strategy were used. The mass calibration parameter was set to local mass calibration. Interference correction was enabled at the MS1 and MS2 levels, eliminating fragments or isotopes from quantification on the basis of the presence of interfering signals but retaining at least three fragments or isotopes for quantification. The FDR was set to 1% at the peptide precursor level and 1% at the protein level. 

Statistical Analysis of Proteomic Data

Statistical analysis between the syngeneic control and GVHD groups was performed via the Wilcoxon signed-rank test. Differentially expressed proteins (DEPs) between the two groups were defined on the basis of the following criteria: missing values in no more than 50% of the samples, fold change (FC) > 2 or < 0.5, and P < 0.05.¹⁴ 

Gene Ontology (GO) and Reactome analyses, including principal component analysis, heatmap construction, and volcano plot construction, were performed to clarify the molecular basis of biological processes (BPs) related to the cornea in the GVHD group compared with the syngeneic control group via Perseus software. GO biological processes (GOBPs) enriched in the DEPs were defined as those with P < 0.05. 

Finally, enrichment and network analyses were performed to identify the pathological network involved in ocular GVHD. The network model was built via Cytoscape¹⁷ and ClueGO¹⁸ software. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEPs was performed with the R package clusterProfiler¹⁹ and the Bioconductor package pathview.²⁰ 

Results

GVHD was Induced Using an Established Mouse Model of GVHD

Four weeks after allogeneic BMT, the mice in the GVHD group had developed the GVHD phenotype, including diarrhea, blepharitis, and corneal epithelitis, but this phenotype was not observed in the syngeneic control group (Fig. 1A). The onset of ocular and systemic GVHD is typically reported to occur 3 weeks after BMT.¹⁵ We analyzed the samples in this study at 1 week after the development of the ocular and systemic GVHD phenotype. The severity of ocular GVHD was evaluated according to a previously reported scoring system for ocular GVHD in mice.²¹ Ocular surface disease as indicated by corneal staining (Figs. 1Ab, 1Ac), corneal ulcers (Fig. 1Ac), ocular surface inflammation as indicated by bulbar redness (Figs. 1Aa, 1Ad; Supplementary Fig. S1A), and ocular discomfort as assessed by palpebral fissure width (Figs. 1Aa, 1Ad; Supplementary Fig. S1A) were classified as grade 2 severity on the basis of the scoring system. The mice in this ocular GVHD model exhibited dry eye, with significantly reduced tear production at 3 and 4 weeks after BMT.²² Regarding additional corneal complications in this GVHD model, corneal opacity and neovascularization (Supplementary Fig. S1B) were observed. The GVHD phenotype was successfully reproduced in this mouse model.^22–32 

Profiling of Detected Proteins

In total, 4027 proteins were identified in the corneas of mice with GVHD (n = 5) and syngeneic control mice (n = 5) through DIA analysis (Supplementary Table S1). GO cellular component (GOCC) functional enrichment analysis revealed that the identified proteins exhibited particular enrichment in extracellular components (extracellular exosome, extracellular organelle, and vesicle) and intracellular components (mainly cytoplasm, intracellular anatomical structure, and protein-containing complex) (Fig. 2A). Principal component analysis revealed significant differences between the GVHD and syngeneic control groups (Fig. 2B), suggesting the existence of GVHD-specific molecules. 

Figure 2.

 

GOCC analysis, principal component analysis, and volcano plots of proteomic data from corneas of mice in the syngeneic control group and GVHD group. (A) The asterisks (*) indicate an extracellular component. The y-axis shows the number of proteins in each CC, and the x-axis shows the names of the CCs identified in this study. (B) PC analysis showing GVHD samples (red dots) and syngeneic control samples (green dots). (C) Volcano plot. Significantly upregulated DEPs are shown in red, and significantly downregulated DEPs are shown in blue.

Identification of Up- and Downregulated Proteins

The volcano plot shows significant differences in corneal protein expression between the syngeneic control and GVHD groups (Fig. 2C). A total of 790 proteins were significantly differentially abundant between the syngeneic control and GVHD groups, with 540 proteins increased in abundance and 250 proteins decreased in abundance in the GVHD group compared to the syngeneic control group (Fig. 2C). All DEPs are listed in Supplementary Table S2. 

Clarification of the Molecular Basis of Chronic Ocular GVHD by GO Enrichment Analysis

DEPs from the volcano plot were further visualized in heatmaps (Fig. 3A). To elucidate the functional roles of these proteins in the cornea in ocular GVHD, enrichment analyses, including GOBP enrichment and Reactome analyses, were performed (Fig. 3B). Several pathways, including innate immune system, neutrophil degradation, intron-containing pre-mRNA, RNA splicing–major pathway, mRNA splicing, metabolism of RNA, and membrane trafficking, differed between the syngeneic control group and the GVHD group (Fig. 3B). The raw GOBP and Reactome analysis data are shown in Supplementary Table S3. The potential pathways were visualized via the Cytoscape ClueGO plug-in, which revealed that the enriched GOBPs (Supplementary Table S4) were highly activated in the corneas of mice with GVHD (Fig. 3C). 

Figure 3.

 

GO biological process analysis and protein‒protein interactions of DEPs in the corneas of control mice and mice with GVHD. (A) The heatmap shows upregulated (red) and downregulated (blue) DEPs in each sample. (B, C) The bar graph (B) and visualization (C) of cellular network changes in the corneas of mice in the GVHD and syngeneic control groups (ClueGO analysis). The GO terms are represented as nodes, and the node size represents the significance of enrichment of the term.

KEGG Pathway Analysis

KEGG pathway analysis of GVHD-affected corneas revealed three altered molecular signaling pathways (Figs. 4A, 5A, 6A). The log₂(FC) values of the proteins mapped to these signaling pathways are shown in Figures 4B, 5B, and 6B. 

Figure 4.

 

Exploration of the necroptosis signaling pathway in GVHD-affected corneas through analysis of the DEPs in mice. (A) KEGG pathway enrichment analysis of DEPs was performed via the R package clusterProfiler¹⁹ and the Bioconductor package pathview.²⁰ The upregulated proteins are shown in red, and the downregulated proteins are shown in green. (B) The bar graph shows the log₂(FC) values of the upregulated and downregulated proteins.

Figure 5.

 

MAPK signaling pathway in GVHD-affected corneas in mice. (A) KEGG pathway enrichment analysis of DEPs was performed via the R package clusterProfiler¹⁹ and the Bioconductor package pathview.²⁰ The upregulated proteins are shown in red, and the downregulated proteins are shown in green. (B) The bar graph shows the log₂(FC) values of the upregulated and downregulated proteins.

Figure 6.

 

The neutrophil extracellular trap signaling pathway in the cornea is activated in cGVHD. (A) KEGG pathway enrichment analysis of DEPs The upregulated proteins are shown in red, and the downregulated proteins are shown in green. (B) The bar graph shows the log₂(FC) values of the upregulated and downregulated proteins.

Upregulation of the Necroptosis Signaling Pathway

Necroptosis, a type of regulated necrosis that is also a newly discovered cell death pathway, is associated with the exacerbation of inflammatory diseases.³³ Molecules that form necrosomes, including receptor-interacting serine/threonine-protein kinase 1 (RIPK1; FC = 3.02), interferon (IFN) regulatory factor 9 (IRF9; FC = 3.26), and IFN-induced, double-stranded RNA-activated protein kinase R (PKR; FC = 3.08), were upregulated (Figs. 4A, 4B). Altered RIPK1 expression is associated with the following necroptosis inducers: tumor necrosis factor (TNF),³⁴ TNF-related apoptosis-inducing ligand (TRAIL),³⁵ FAS ligand,³⁶ and IFN³³ (Fig. 4). Downstream of RIPK1, arachidonate 12-lipoxygenase (ALOX12) and arachidonate 15-lipoxygenase (ALOX15) were also upregulated (FC = 5.34 and FC = 5.87, respectively). In contrast, the expression of the calpain-1 catalytic subunit was lower in the GVHD group than in the syngeneic control group (FC = 0.47) (Figs. 4A, 4B). The secretion of high mobility group protein B1 (HMGB1), a type of damage-associated molecular pattern (DAMP),³⁷ was increased (FC = 2.79) (Figs. 4A, 4B). These results suggest that necroptosis mediated by RIPK1 and RIPK3, along with the release of HMGB1, a DAMP, plays a role in GVHD-affected corneas. 

Upregulation of the MAPK Signaling Pathway

The MAPK pathway, a key inflammatory pathway, plays a role in cellular reprogramming processes, including apoptosis.³⁸ MAPK has been studied in corneal cells in relation to dry eye disease, limbal stem cell deficiency, and squamous metaplasia.^38–41 

In this study, MAPK14, also called p38α, was upregulated (FC = 19.94) (Figs. 5A, 5B). CDC42/Rac (Rac2; FC = 4.60), a small GTPase expressed in polymorphonuclear neutrophils, reportedly regulates the cytoskeleton, cell shape, adhesion, and migration and is also an essential component of the NADPH oxidase complex.⁴² Molecules downstream of interleukin 1 (IL-1) and the IL-1 receptor (IL-1R)—including myeloid differentiation primary response 88 (MyD88; FC = 3.43) and IL-1R-associated kinase 4 (IRAK4; FC = 5.07); evolutionarily conserved signaling intermediate in the Toll pathway, mitochondrial (ECSIT; FC = 6.21); protein phosphatase 1B (Ppm1b, also referred to as protein phosphatase 2 catalytic subunit beta (PP2CB); FC = 3.21); and serine/threonine protein phosphatase 5 (Ppp5c, also referred to as protein phosphate 5 (PP5); FC = 4.88)—were sequentially upregulated (Fig. 5A). These results suggest that the IL-1R–mediated MAPK signaling pathway was upregulated in GVHD-affected corneas. 

Upregulation of the NET Pathway

The contribution of NETs to pathological changes in ocular GVHD has been studied.^43–45 HMGB1 and MAPK14 are also involved in this pathway. The levels of integrin alpha-M components (macrophage-1 antigen [MAC-1] and complement receptor 3 [CR3]; FC = 3.17) and DAMP molecules expressed on the neutrophil surface were increased in the GVHD group. Changes in histone modifications and increases in the expression levels of NETosis-related molecules, including myeloperoxidase (MPO; FC = 5.55) and cathepsin G (CG; FC = 7.83), were also observed in the GVHD group, as previously reported^43–45 (Figs. 6A, 6B). Gasdermin D (GSDMD), which is involved in pyroptosis, downstream of the inflammasome, was upregulated in the GVHD group (FC = 6.20). Moreover, Rac2 (FC = 4.60), in the NET pathway is elevated in GVHD-affected corneas.⁴² Histone deacetylase 1 (HDAC1; FC = 0.33) was downregulated in the NET pathway in this study. Low-dose HDAC inhibitors have been reported to trigger NET formation.⁴⁶ Collectively, these results suggest that these molecules associated with NETs may contribute to the development of corneal pathology in GVHD. 

Taken together, our findings suggest that these three pathways are upregulated simultaneously at one time point (i.e., 4 weeks after BMT) in GVHD-affected corneas. 

Discussion

Through KEGG enrichment analysis, we revealed that the following key signaling pathways are upregulated in the corneas of mice with GVHD: necroptosis, MAPK, and NET signaling pathways. Within these pathways, we identified molecules that have not been reported in GVHD-affected corneas, including RIPK1, IRF, PKR, cylindromatosis (CYLD), ALOX, H2A histone family member X (H2AX), and HMGB1, which are involved in necroptosis; receptor tyrosine kinase (RTK), growth factor receptor-bound protein 2 (GRB2), IRAK, ECSIT, PP2CB, PP5, and Rac2, which are involved in MAPK signaling; and HMGB1, CG, MAC-1, CR3, H2A, H2B, Rac2, and GSDMD, which are involved in NET signaling. 

First, RIPK1 and RIPK3 in the necroptosis pathway were upregulated. Generally, necrosis and programmed cell death are the types of cell death that can occur. Necrosis is traditionally considered an unregulated form of cell death; however, regulated necrosis was recently discovered.⁴⁷ The discovery of different types of regulated necrosis, including necroptosis, ferroptosis, NETosis, and pyroptosis, revealed new mechanisms influencing inflammation and immunity.^33,48,49 Necroptosis is universally defined as a type of regulated cell death triggered by perturbations in extracellular or intracellular homeostasis; these activities are critically dependent on proteins, such as mixed lineage kinase domain-like pseudokinase (MLKL), RIPK3, and RIPK1.⁴⁹ RIPK1 is a promising therapeutic target for various human autoimmune and inflammatory diseases.⁵⁰ In addition, necroptosis contributes to chronic inflammation and liver fibrosis during aging.⁵¹ A previous study revealed that inhibiting necroptosis reversed intestinal GVHD via RIPK1 and RIPK3.⁵² In the cornea, regulated necrosis is associated with infection, dry eye, alkali burn injury, macular corneal dystrophy, and corneal endothelial keratopathy,⁵³ suggesting that this pathway plays a role in epithelial damage, including epithelitis and ulcers, in GVHD-affected corneas. CYLD and H2AX have been reported to constitute a novel related gene signature and regulatory network for overall survival prediction in patients with lung adenocarcinoma,⁵⁴ suggesting that they could also be markers of necroptosis in the context of ocular GVHD. 

The expression of IRF9, which is involved in the type I IFN response, was found to be elevated in the necroptosis pathway. IRF9, a transcription factor, is associated with inflammation, autoimmunity, and immune cell migration,⁵⁵ indicating its hub role in the corneal immune response, inflammation, and metabolism⁵⁶ in immune cells, vascular cells, and epithelial cells in GVHD-affected corneas and its potential relation to epithelial migration, neovascularization, and conjunctivalization. PKR, which is crucial for IFN-induced RIP1/RIP3-mediated necrosis,⁵⁷ is upregulated by RIPK1/RIPK3 in the IFN-related necroptosis pathway in corneas affected by GVHD. 

Lipoxygenases (LOXs) are enzymes that metabolize lipids, producing active eicosanoids from polyunsaturated fatty acids.⁵⁸ ALOX15 and its eicosanoid metabolites 12/15-hydroxyeicosatetraenoate (12/15-HETE) play important roles in both physiological processes and inflammatory diseases.^58–61 12/15-HETE activates peroxisome proliferator-activated receptor γ (PPARγ), and, at high concentrations under conditions of excessive inflammation, toxic reactive oxygen species result in increased production of 12/15-HETE compared with that under physiological conditions.^60,61 We found that in the necroptosis pathway, LOX may be involved in GVHD-affected corneas via reactive oxygen species, leading to inflammation in GVHD-affected corneas. On the other hand, ALOX12 might be excessively increased to prevent excessive corneal inflammation in ocular GVHD. Calpain performs dual functions, either resolving or exacerbating inflammation.⁶² In corneal GVHD, calpain may function as a negative regulator of inflammation. Further clarification of the associations between the downregulation and upregulation of calpain and LOX, respectively, is needed. 

In this study, the level of HMGB1 was found to be increased in the necroptosis pathway, probably due to the significant release of DAMPs, which may be a potent inflammatory trigger, by disintegrating necroptotic cells.³³ Thus, HMGB1, which is released during necroptosis and is associated with extracellular components, such as exosomes and vesicles (Fig. 2A), may contribute to excessive inflammation in GVHD-affected corneas. The reduced expression of calpain, a regulator of necroptosis (Fig. 4), implies that the inhibition of JNK-interacting protein-1 (JIP-1), a downstream regulator, is decreased. This decrease activates the pathway that promotes necroptosis.⁶³ 

Second, the MAPK signaling pathway exhibited increased activity in the cornea in the GVHD mouse model. In the central nervous system during GVHD, microglia show increased activation of transforming growth factor beta (TGF-β)-activated kinase 1 (TAK1) and nuclear factor-κB (NF-κB)/p38 MAPK signaling.⁶⁴ GVHD-specific activation of signal transducer and activator of transcription 3 (STAT3)/STAT1 occurs before the activation of NF-κB and MAPK, which is followed by the induction of STAT1- or STAT3-dependent inflammatory gene expression programs.⁶⁵ We previously reported p38 MAPK expression in lacrimal gland tissue in GVHD.²⁵ However, the sequential upregulation of the MAPK signaling pathway in GVHD-affected corneas has not been reported. In this study, downstream molecules of IL-1R, including MyD88, IRAK4, ECSIT, and p38, were sequentially upregulated in the MAPK signaling pathway in corneas from mice in the GVHD group. The downstream pathway of IL-1R (i.e., MyD88/IRAK4 pathway) is important for activating innate and adaptive immune responses in GVHD; this pathway has been reported to involve a broad range of adaptor proteins of the IL-1R and Toll-like receptor (TLR) superfamily,⁶⁶ suggesting that these molecules are potential targets. Regarding dry eye disease, including ocular GVHD, the levels of IL-1R and its downstream molecule MyD88 were found to be increased in tear washes from patients,⁴⁴ supporting the results of our study. The expression of ECSIT, a key adaptor protein essential for NF-κB signaling and the Toll/IL-1 pathway,⁶⁷ was increased, suggesting that ECSIT is involved in corneal inflammatory pathogenesis in GVHD. These findings suggest that IL-1R signaling is sequentially activated, which in turn activates MAPK signaling and both the innate and adaptive immune responses in GVHD-affected corneas. 

Rac2 has been reported to play a role as a key molecule for macrophage polarization in the profibrotic phenotype and progression of pulmonary fibrosis,⁶⁸ suggesting that it could be related to corneal opacity in GVHD-affected corneas. The RTK family consists of platelet-derived growth factor receptor (PDGFR), vascular endothelial growth factor receptor (VEGFR), and epidermal growth factor receptor (EGFR).⁶⁹ Thus, VEGFR may be related to corneal neovascularization in GVHD-affected corneas. The expression of GRB2 was significantly enhanced in the skin of systemic sclerosis patients,⁷⁰ whose mechanisms leading to inflammation and fibrosis are similar to those in cGVHD,¹⁵ suggesting similar roles of increased GRB2 expression in inflammation and opacity in GVHD-affected corneas. The Crk adaptor proteins have been reported to be potential target molecules for the tissue-selective disruption of integrin-dependent inflammatory responses in systemic GVHD. Crk proteins could promote the migration of T cells to target organs in patients with GVHD.⁷¹ These findings suggest that the increased expression of Crk II found in this study may have caused T-cell migration and resulted in inflammation and corneal dysfunction in GVHD-affected corneas. β-Arrestins have been implicated in the regulation of several basic cellular functions, including cell-cycle regulation, cell migration, and apoptotic signaling, and perform more aggressive functions in invasion and neovascularization in cancers.⁷² If similar effects are present in GVHD-affected corneas, however, these effects could affect limbal stem cells and may lead to dysregulated epithelial cell migration, abnormal wound healing, and neovascularization. Ppm1b, also referred to as PP2CB, is involved in the negative regulation of the MAPK signaling pathway⁷³ and functions as a key regulator of the p38 MAPK-dependent aging pathway in human fibroblasts, which is involved in the regulation of cellular senescence.⁷⁴ We previously reported that cellular senescence could occur in GVHD-affected corneas, similar to lacrimal glands,³¹ and could contribute to the abnormal regeneration of corneal epithelia, as evidenced by the upregulation of H2AX in the necroptosis pathway also in the present study. The expression of PP5 is reportedly elevated in corneal epithelial hyperplasia and squamous epithelial neoplasia in mice.⁷⁵ Upregulation of PP5 may affect corneal squamous epithelial metaplasia in GVHD-affected corneas. 

Third, the abundance of NETs was reported to be elevated in tear washes from patients with ocular GVHD,^43–45 and NETs have been implicated in inflammation in immune-mediated dry eye diseases.⁴⁴ Consistent with previous reports,^43–45 we confirmed that the NET signaling pathway and several associated molecules were upregulated in corneal proteomic data from a GVHD mouse model. HMGB1, a DAMP, is upregulated and linked to p38 upregulation via TLR2 or TLR4.⁷⁶ Moreover, Rac2 was upregulated in the NET pathway as well as the MAPK pathway in GVHD-affected corneas in the present study, suggesting that NETs and fibrosis-associated epithelial–mesenchymal transition are linked.⁴² GSDMD, which is involved in pyroptosis and regulates necrosis, which is similar to necroptosis,⁷⁷ was upregulated in the NET pathway in the corneas of mice with GVHD. Pyroptosis depends on gasdermin-mediated formation of plasma membrane pores due to activation of inflammatory caspases.⁴⁹ The expression of integrin components, including the DAMPs MAC-1 and CR3, has been found to be upregulated.⁷⁸ CG is a neutral serine protease known for degrading proteoglycans and collagens of articular cartilage and enhancing elastase activity,⁷⁹ suggesting that this molecule contributes to the elevation of elastase in patients’ eye washes with ocular GVHD.⁴³ The expression of several molecules, including MPO, was previously reported to increase during NETosis. These observations validate our findings. 

One limitation of our study was the small size of the syngeneic control and GVHD groups. The DDA method was used to construct a library for DIA analysis, but subsequent detection was limited to only those molecules in the library; thus, fewer proteins may have been identified because of the smaller sample size. Second, our results are limited to a single post-BMT period in one animal model of GVHD. A better approach would be to identify molecules with rapid changes in expression in the cornea before and after HSCT in GVHD. Finally, further experiments are needed to determine whether this corneal pathophysiology results from the upregulation of the three identified signaling pathways or whether they are only partial etiologies for corneal GVHD. This possibility could be explored by analyzing earlier time periods. 

Ocular GVHD is characterized by chronic inflammation with the activation of both innate and adaptive immune responses and pathogenic fibrosis on the ocular surface.⁸⁰ The typical corneal findings of ocular GVHD include corneal epithelitis, corneal erosion, corneal ulcers, corneal opacity, limbal stem cell deficiency, corneal epithelial squamous metaplasia, neovascularization, conjunctivalization, and corneal perforation. Interestingly, we confirmed the key pathway molecules in mice with GVHD through proteomic analysis. The molecules identified in this study can cause each corneal finding in GVHD in mice. 

RIPK1 has a deteriorative effect on corneal neovascularization and inflammation in a mouse model of alkali burn. Disruption of apoptosis may trigger necroptosis via the infiltration of macrophages expressing RIPK1.^53,78 In GVHD-affected corneas, macrophages expressing RIPK1 may cause further damage to corneal epithelia, leading to necroptosis affected by GVHD. Necroptosis promotes inflammation through the release of inflammatory cytokines and the cellular contents as DAMPs.⁵⁰ Therefore, RIPK1 expression in GVHD-affected corneas may induce epithelial damage via DAMPs and inflammatory cytokines, resulting in corneal erosion, corneal ulcers, and corneal neovascularization. 

It has been reported that the activation of p38 MAPK inhibits the differentiation of limbal stem cells to corneal epithelial cells.³⁸ Activation of the p38 MAPK signaling pathway is reportedly involved in the pathological process of corneal limbal stem cell deficiency under severe dry eye conditions in mice.³⁸ Therefore, this pathway may be linked to abnormal corneal epithelial differentiation, squamous epithelial metaplasia, corneal ulcers, corneal neovascularization, and corneal perforation as severe complications of corneal GVHD. 

In the NET pathway, we identified GSDMD, which is related to pyroptosis. Several studies have suggested that dry eye disease stimulates keratitis through the canonical pyroptosis pathway.⁸¹ Desiccation stress caused by severe dry eye disease affected by GVHD activates corneal epithelial pyroptosis through inflammasome-mediated and GSDMD-dependent signaling pathways, resulting in dry eye disease progression, corneal ulcers and corneal perforation.⁸¹ Pyroptosis is a prominent consequence of inflammasome activation, which is characterized by GSDMD-driven cell lysis. The released DAMPs are recognized by IFN receptors, which in turn activate downstream MyD88, type I IFN, and MAPK-dependent inflammation, activating innate and adaptive immunity. Neutrophils are reported to be abundant on the ocular surface,^43,44,82 leading to NET formation via DAMPs.⁸³ In addition, NETs reportedly cause filamentary keratitis,⁸⁴ which leads to severe pain in patients with ocular GVHD.⁴³ It is important to determine the detailed mechanisms underlying this disease and develop new therapies for patients with ocular GVHD. 

In conclusion, we identified three upregulated signaling pathways and several new proteins in the corneas of mice with GVHD. The identification of these new proteins and signaling pathways in GVHD-affected corneas can facilitate elucidation of the pathogenesis of this disease and lead to the development of new treatments to prevent irreversible changes at the earliest stages. 

Acknowledgments

Supported by grants from the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education (RS-2024-00344204 to HLK) and the Japanese Ministry of Education, Science, Sports, Culture and Technology (18K09421, 24K12769 to YO). The funders had no role in this study. 

Disclosure: K. Asai, Keio University (P); H.K. Lee, None; S. Sato, Keio University (P); E. Shimizu, Keio University (P); J. Jung, None; T. Okazaki, Keio University (P); M. Ogawa, None; S. Shimmura, None; K. Tsubota, Keio University (P); Y. Ogawa, Keio University (P); K. Negishi, Keio University (P); M. Hirayama, Keio University (P)