The Role of SLIT3–ROBO4 Signaling in Endoplasmic Reticulum Stress–Induced Delayed Corneal Epithelial and Nerve Regeneration

Abstract

Purpose: In the present study, we aim to elucidate the underlying molecular mechanism of endoplasmic reticulum (ER) stress induced delayed corneal epithelial wound healing and nerve regeneration.

Methods: Human limbal epithelial cells (HLECs) were treated with thapsigargin to induce excessive ER stress and then RNA sequencing was performed. Immunofluorescence, qPCR, Western blot, and ELISA were used to detect the expression changes of SLIT3 and its receptors ROBO1-4. The role of recombinant SLIT3 protein in corneal epithelial proliferation and migration were assessed by CCK8 and cell scratch assay, respectively. Thapsigargin, exogenous SLIT3 protein, SLIT3-specific siRNA, and ROBO4-specific siRNA was injected subconjunctivally to evaluate the effects of different intervention on corneal epithelial and nerve regeneration. In addition, Ki67 staining was performed to evaluate the proliferation ability of epithelial cells.

Results: Thapsigargin suppressed normal corneal epithelial and nerve regeneration significantly. RNA sequencing genes related to development and regeneration revealed that thapsigargin induced ER stress significantly upregulated the expression of SLIT3 and ROBO4 in corneal epithelial cells. Exogenous SLIT3 inhibited normal corneal epithelial injury repair and nerve regeneration, and significantly suppressed the proliferation and migration ability of cultured mouse corneal epithelial cells. SLIT3 siRNA inhibited ROBO4 expression and promoted epithelial wound healing under thapsigargin treatment. ROBO4 siRNA significantly attenuated the delayed corneal epithelial injury repair and nerve regeneration induced by SLIT3 treatment or thapsigargin treatment.

Conclusions: ER stress inhibits corneal epithelial injury repair and nerve regeneration may be related with the upregulation of SLIT3-ROBO4 pathway.

The intactness of the corneal epithelium and abundant sensory innervation are vital for maintaining normal physiological function of the cornea. Trauma, infection, dry eye syndrome (DES), and keratitis can cause damage to the corneal epithelium and nerves, and lead to visual impairment.^1,2 The regeneration of the corneal epithelium and nerves in response to injury is essential for restoring normal corneal function. 

Recently, growing evidence has demonstrated the impact of endoplasmic reticulum (ER) stress in the pathogenesis of corneal epithelial – and neural defect – related diseases. ER stress is usually indicated by the level of ER stress markers, such as GRP78, XBP1, ATF6, and CHOP.^3–6 XBP1 serves as a crucial marker within the IRE1/α-XBP1 pathway of the UPR response and is essential for cell fate determination in response to ER stress.⁷ CHOP maintains low levels of expression under non stress conditions; however, in ER stress response, its expression is significantly increased through IRE1-, PERK-, and ATF6-dependent transcriptional induction.⁸ Inhibiting ER stress mitigates DES^9–11 and protects against alkaline burn induced corneal injuries.^12,13 One of our previous studies confirmed excessive ER stress (marked by a higher level of XBP1 and CHOP expression) in the process of corneal wound healing in diabetes; specifically, inhibiting ER stress with 4-phenylbutyric acid (4-PBA) significantly accelerates corneal epithelial and nerve regeneration.¹⁴ Although the primary downstream mechanisms are considered to be excessive inflammation and programmed cell death triggered by ER stress, the underlying molecular mechanism of ER stress affecting corneal epithelial and nerve repair have not been elucidated. 

Axon guidance molecules (AGMs), mainly including the slit, netrin, ephrin, and semaphorin families, are known to regulate axonal growth during development^15–18 and have been shown to play a crucial role in corneal wound healing.^19,20 Victor H. Guaiquil et al. found that following corneal epithelial injury, many AGMs, such as semaphorins, ephrins, and netrins, decreased and then quickly recovered; moreover, corneal epithelial wound healing could be promoted by semaphorin3A (Sema3A), ephrinB2 and netrin-4.²¹ Semaphorin3C (Sema3C) and its receptor, neuropilin-2 (NRP-2), as well as netrin-1 and its receptor, adenosine2B, accelerates corneal epithelial and nerve regeneration.^22,23 By contrast, ephrinA1 and ephrinA2 receptor signaling restrict corneal epithelial wound healing.²⁴ These studies indicate that AGMs play a complex role in corneal regeneration. 

The Slit guidance ligand (SLIT) family, including SLIT1, SLIT2, and SLIT3, is a secretory family. These proteins function by interacting with Roundabout receptors (ROBO) receptors (ROBO1, ROBO2, ROBO3, and ROBO4). The SLIT/ROBO signal was first discovered in the nervous system, capable of regulating axonal guidance and remnants. Recently, more and more studies have shown that the SLIT/ROBO signal can also regulate other physiological and pathological processes, such as angiogenesis, inflammatory cell chemotaxis, tumor cell migration and metastasis, fibrosis, etc.^25,26 Research in fruit flies found that JNK mediated Slit-Robo2 signaling promotes epithelial wound healing.²⁷ In addition, a study in diabetic mice found that SLIT2 promotes the regeneration of the corneal epithelium and nerves.²⁸ 

In the present study, we induced ER stress in human limbal epithelial cells (HLECs) using thapsigargin (TG) and performed RNA sequencing to identify potential key molecules involved in ER stress affecting corneal regeneration. We found that ER stress caused significant changes in AGMs, with SLIT3 and its receptor, ROBO4, being particularly elevated. Therefore, in the present study, we investigated the role of ER stress in regulating corneal epithelial and nerve regeneration through the SLIT3-ROBO4 pathway. 

Materials and Methods

Animals

Male C57BL/6J mice (8–10 weeks old) were obtained from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). All animal experiments were conducted according to the guidelines and statements of the American Society for Visual and Ophthalmic Research regarding the use of animals in ophthalmic and visual research and were approved by the ethics committee of the Shandong Eye Institute. The anesthetized mice were injected subconjunctivally with 5 µL of TG (50 µm/mL), a classic ER stress agonist (MedChemExpress, Monmouth Junction, NJ, USA), or SLIT3 (100 ng/mL; R&D Systems, Minneapolis, MN, USA) 24 hours before and 0 hours after corneal epithelial scraping. ROBO4-specific siRNA (20 mmol/L), SLIT3-specific siRNA (20 mmol/L; GenePharma Company, Suzhou, China) or nonspecific control (NC) siRNAs were injected 24 hours and 4 hours before and 0 hours after wounding. The SLIT3 and ROBO4-specific siRNA sequence listed in Table 1 were synthesized using GenePharma. 

Table 1.

 

Mouse SLIT3 and ROBO4 siRNA Sequences

 

Corneal Epithelial Wound Healing

The mice were administered general anesthesia through an intraperitoneal injection of 0.6% pentobarbital sodium, followed by the topical application of proparacaine hydrochloride eye drops. A 2.5 mm diameter area of the central corneal epithelium was scraped using an electric handheld epithelial scraper (Alger Co., Lago Vista, TX, USA). Subsequently, ofloxacin eye ointment was used to prevent infection. The wound healing progression was monitored using fluorescein sodium staining, and photographs were taken with a slit lamp microscope (Topcon, Tokyo, Japan). The image analysis software ImageJ was used to measure the corneal epithelial defect area in each group. 

Cell Culture and Treatment

The corneal ring left over from corneal transplantation surgery in the eye bank was used for the primary culture of HLECs. The limbal corneal epithelium was removed after being digested with 15 mg/mL dispase Ⅱ (Roche, Indianapolis, IN, USA) in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA, USA) overnight at 4°C. Next, the epithelium was digested into single cells using 0.25% trypsin / 0.02% EDTA at 37°C for 15 minutes. The cells were subsequently suspended in DMEM/F-12 medium supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY, USA), insulin-transferrin-selenium (Invitrogen), 1% nonessential amino acids (Invitrogen), 0.1 nM cholera toxin (Sigma, St. Louis, MO, USA), 2 nM 3,3′,5-triiodo-l-thyronine sodium salt (Sigma), 0.4 ng/mL dexamethasone sodium phosphate (Wako, Osaka, Japan), 2 mM l-glutamine (Invitrogen), penicillin-streptomycin (Hyclone, Logan, UT, USA), 10 ng/mL recombinant human epidermal growth factor (EGF; R&D Systems), and 10 µM Y27632 (stem cells). The primary cells were cultured on dishes coated with a 3T3 feeder layer and passaged after 80% to 90% confluence. The passaged HLECs were treated with or without 400 nm/mL TG for 24 hours. 

A mouse corneal epithelial stem/progenitor cell line (TKE2) was provided by Dr. Tetsuya Kawakita of Keio University (Tokyo, Japan). Mouse TKE2 has been applied in the research of cell biology of corneal epithelial cells.^29–32 The mouse TKE2 cells were cultured in KSFM basal medium (Gibco/Thermo Fisher Scientific Inc., Waltham, MA, USA) supplemented with human keratinocyte growth supplement (HKGS; Gibco) and recombinant human EGF (5 ng/mL; R&D Systems) at 37°C with 5% CO₂. This cell line has been characterized in our previous studies.^33,34 Finally, the TKE2 cells were treated with recombinant at a concentration of 100 ng/mL SLIT3 or 400 nm/mL TG. 

Cell Proliferation and Migration Analysis

To detect the effect of TG on the proliferation of HLECs or mouse TKE2 cells, the cells were digested into single cells using 0.25% trypsin / 0.02% EDTA after 24 hours of treatment of TG, and the cell number was counted. To detect the effect of SLIT3 on the proliferation of mouse TKE2 cells, the latter were treated with SLIT3 protein at a final concentration of 100 ng/mL for 24 hours and measured using a CCK-8 assay kit (Bioss, Beijing, China). 

For the migration assay of mouse TKE2 cells, the cells were cultured to confluence and scraped with a 200 µL pipette tip. Subsequently, the cells were washed 3 times with phosphate-buffered saline (PBS), and exogenous SLIT3 protein was added at a final concentration of 100 ng/mL. Photographs of the samples were taken after 0 hours and 24 hours, and the percentage of the closed area was calculated using ImageJ. 

Immunofluorescence Staining

For immunofluorescence staining, a frozen section (7 µm) of the treated cells was fixed in 4% paraformaldehyde for 20 minutes at room temperature, permeabilized, and blocked for 1 hour in PBS using 5% BSA and 0.1% Triton X-100, followed by primary antibody incubation overnight at 4°C. The samples were then incubated with a fluorescein-conjugated secondary antibody for 2 hours at room temperature. The proportion of Ki67-positive cells in the corneal limbus and wound margin was calculated by counting the number of Ki67 positive–stained cells and DAPI-stained cells. The relative expression of CHOP, XBP1, SLIT3, and ROBO4 was calculated by performing a quantitative analysis of the fluorescence intensity using ImageJ. The antibodies used in this assay were shown in Table 2. 

Table 2.

 

The Antibodies Used in the Experiment

 

Whole-Mount Staining of Corneal Nerves

For corneal whole-mount staining, the corneas were dissected and fixed with 4% paraformaldehyde for 1 hour at 4°C. The corneas were permeabilized and blocked with PBS containing 0.3% Triton X-100 and 5% BSA overnight at 4°C, and then incubated with TUBB3 antibodies overnight. The staining was observed and captured using a ZEISS confocal microscope (ZEISS, Jena, Germany), and the nerve density was calculated using ImageJ. 

RNA Sequencing

After 24 hours of culture with TG-treated or without TG-treated HLECs, the cells were collected and sent to a biotech company (Oebiotech Biomedical Technology Co., LTD., Shanghai, China) for RNA sequencing. The sequencing was performed using the Illumina HiSeq 4000 high-throughput sequencing platform. Finally, GSEA, volcano plot, and heat map analyses were conducted using oebiotech.com. 

Reverse Transcription Quantitative PCR

Complete RNA was extracted using a TransZol Up Plus RNA kit (Transgen Biotech, Beijing, China). The cDNA was synthesized by reverse transcription using complementary DNA and using a HiScript III RT SuperMix Kit (R323-01; Vazyme, Nanjing, China). A ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd., Nanjing, China) and a Rotor-Gene Q Real-Time PCR cycler (Qiagen, Germany) were used to perform real-time PCR. The primers listed in Tables 3 and 4 were synthesized using GenePharma. The expression of human and mouse β-actin was used for normalization to calculate the mRNA levels of each sample. The 2-^ΔΔct method was used to analyze the relative expression levels of mRNA. 

Table 3.

 

Mouse Primer Sequences Used for qPCR

 

Table 4.

 

Human Primer Sequences Used for qPCR

 

ELISA Analysis

After 24 hours of treatment with or without TG, the mouse TKE2 cells were collected. After thorough lysis, centrifugation was performed at 3000 rpm for 10 minutes, and the supernatant was collected. The concentration of SLIT3 protein was determined using the Mouse SLIT3 ELISA Kit according to the instructions (FanKeW, Shanghai, China). Measurements were taken at an absorbance of 450 nm using a microplate reader (Molecular Devices, Sunnyvale, CA, USA). 

Western Blot

The supernatant from mouse corneal epithelium after lysis and centrifugation were collected and used for Western blot detection. Then, 50 µg of protein were electrophoresed on a 7.5% SDS-PAGE gel, and then transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore, Billerica, MA, USA). The PVDF membrane was blocked with 5% skim milk at room temperature for 2 hours. TBST was used for 3 times of washing, each for 10 minutes, followed by overnight incubation with anti-ROBO4 antibody (1:2000; Affinity, Jiangsu, China) at 4°C. On the next day, the membrane was incubated for 2 hours at room temperature with an HRP-conjugated secondary antibody (1:4000; Proteintech, Shanghai, China). The blots were visualized using the Bio-Rad Molecular Imager Che-miDoc XRS system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Nonmuscle β-actin (1:4000; Proteintech, Shanghai, China) was used as a loading control. 

TUNEL Assay

To detect the cells death levels in TG treated mouse cornea, HLECs, and mouse TKE2 cells, we used a in situ cell death detection kit (Roche Applied Science, Penzberg, Upper Bavaria, Germany) and performed as per the manufacturer's instructions. 

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 8 (GraphPad Software Inc.). The data are presented as mean ± SD. Paired t-tests were utilized to compare differences between the two groups, whereas an analysis of variance (ANOVA) was used to compare differences among three or more groups. Each experiment was repeated at least three times to ensure reproducibility. 

Results

TG-Induced ER Stress–Impaired Corneal Epithelial and Nerve Regeneration

To evaluate the pathological effects of ER stress after corneal injury, we induced corneal ER stress by subconjunctivally injecting TG, a commonly used ER stress inducer. Here, we found that TG induced higher expression levels of ER stress markers Xbp1 and Chop in corneal epithelium (Supplementary Figs. S1A–S1C). We found that the healing of the corneal epithelium was significantly suppressed by TG treatment (Fig. 1A). Compared with normal corneas, the healing rates decreased by 34.76 ± 8.62% and 16.56 ± 1.53% 24 hours and 36 hours after injury, respectively (Fig. 1B). The Ki67-positive cells in the limbus and central wound margin were markedly reduced in the TG-treated corneas during wound healing (Figs. 1C, 1D). In addition, we performed corneal nerve whole-mount staining and found that TG significantly hindered the repair of the corneal nerves (Fig. 1E, Supplementary Figs. S2A, S2C). The peripheral nerve density of TG treated corneas decreased by 85.79 ± 0.54%, 45.13 ± 1.73%, and 12.64 ± 0.95% and the central nerve density of TG treated corneas decreased by 100%, 74.87 ± 2.13%, and 19.34 ± 1.64% compared with normal corneas at 24 hours, 5 days and 30 days after epithelial scraping, respectively (Fig. 1F, Supplementary Figs. S2B, S2D). These results demonstrated that TG-induced ER stress directly impaired both corneal epithelial wound healing and nerve regeneration. By the way, although significantly delayed compared to the control group, after TG treatment, the defect of corneal epithelium was closed at 48 hours after scraping and the nerves were gradually recovering. 

Figure 1.

 

TG-induced ER stress inhibited corneal epithelial and nerve regeneration. (A) After epithelial scraping, the remaining wound defect was stained with fluorescein sodium and photographed at 0 hours, 24 hours, 36 hours, and 48 hours. (B) The epithelial defects at 24 hours and 36 hours were quantified as the percentage of defect area relative to the original wound area (n = 4/group). (C) Ki67-staining of corneas of the control and TG-treated groups 24 hours after injury. (D) The percentage of Ki67-positive cells (%) at the limbus and wound margin was calculated (n = 4/group). (E) Whole-mount corneal staining of TUBB3 24 hours after epithelial scraping; the representative areas of the central and peripheral corneas are presented below. (F) Statistics of nerve density in central and peripheral corneas in the control and TG-treated groups 24 hours post-wounding using ImageJ (n = 4/group). ***P < 0.001.

TG-Induced ER Stress Inhibited the Proliferation Ability of HLECs and Mouse TKE2 Cells

The primary cultured HLECs and mouse TKE2 cells were treated with 400 nm TG for 24 hours, and the mRNA expressions of ER stress associated genes XBP1 and CHOP were found to be upregulated (see Supplementary Figs. S1D, S1E). We observed that TG caused an obvious change in cell morphology, from regular cobblestone-like shapes to irregular shapes (Fig. 2A). Cell counting revealed that TG significantly reduced the number of cells (Fig. 2B). The proliferation marker of Ki67 staining showed that the Ki67 positive cells were reduced dramatically in HLECs after TG treatment (Figs. 2C, 2D). Similar changes in cell morphology and decreased proliferation were also observed in mouse TKE2 cells after TG treatment (Figs. 2E–H). These findings suggested that TG-induced ER stress impaired the growth ability of corneal epithelial cells. Here, we used TUNEL staining analysis to evaluate the programmed cell death level of TG treated corneal epithelial cells. We found that TG induced cell death in wounded corneas, however most of the TUNEL positive cells were distributed in the corneal stroma (Supplementary Fig. S3A). After 24 hours of TG treatment, no TUNEL positive apoptotic cells were found in both HLECs and mouse TKE2 cells (Supplementary Figs. S3B, S3C). KEGG analysis of transcriptome data revealed that genes downregulated by TG in HLECs exhibit a high enrichment of cell cycle related genes (Supplementary Fig. S3D), suggesting that under the concentration and treatment time of this experiment, TG may inhibit cell growth by affecting the cell cycle. 

Figure 2.

 

Effects of TG-induced ER stress on HLECs and mouse TKE2 cells growth. (A) Morphological changes in HLECs under light microscopy after the induction of ER stress. (B) Cell count statistics of HLECs after TG treatment for 24 hours (n = 4/group). (C) Ki67-staining of HLECs of the control and TG-treated groups after 24 hours. (D) The number of Ki67-positive cells at HLECs was calculated and represented as a percentage of Ki67-positive cells (%; n = 4/group). (E) Morphological changes in mouse TKE2 cells under light microscopy after the induction of ER stress (n = 4/group). (F) Cell count statistics of mouse TKE2 cells after TG treatment for 24 hours. (G) Ki67-staining of mouse TKE2 cells of the control and TG-treated groups after 24 hours. (H) The number of Ki67-positive cells at mouse TKE2 cells was calculated and represented as a percentage of Ki67-positive cells (%; n = 4/group). ***P < 0.001.

TG-Induced ER Stress Upregulated SLIT3 and ROBO4 Signaling in HLECs

To further investigate the potential molecular mechanism, we conducted RNA sequencing. The GSEA analysis, a powerful analytical method that compares gene sets across biological states and detects statistically significant differences,³⁵ confirmed that TG induced an excessive ER stress response in HLECs (Fig. 3A). Subsequently, according to KEGG analysis, we focus on the differential genes related with development and regeneration. We found that axon guidance–related genes constituted a substantial portion, with significant upregulation of SLIT3 expression (Fig. 3B). A volcano plot illustrating SLIT–ROBO signaling showed significant upregulation of SLIT3 and ROBO4 expression (Fig. 3C). We then performed qPCR to test the expression of SLIT1–3 and ROBO1–4, and the results were consistent with the RNA sequencing data: significant upregulation of SLIT3 and ROBO4 in the TG-treated group was observed (Figs. 3D, 3E). 

Figure 3.

 

Transcriptional profiling of HLECs subjected to ER stress. (A) Gene set enrichment analysis (GAES) of ER stress response–related genes. (B) Heat map of genes related to development and regeneration according to KEGG analysis. (C) Volcano plot of RNA sequencing data comparing the SLIT–ROBO signaling pathway–related gene transcripts in the control and TG-treated HLECs. The red dots represent upregulated genes, and the blue dots represent downregulated genes; adjusted P < 0.05 and |log2FC|>1. (D, E) The qPCR detection of mRNA expression of SLIT1–3 and ROBO1–4. ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001. n = 3/group for RNA sequencing.

TG-Induced ER Stress Increased SLIT3 and ROBO4 Expression in Mouse TKE2 Cells and Wounded Mouse Corneal Epithelium

We verified whether TG can also induce upregulation of the Slit3-Robo4 pathway in mouse TKE2 cells. The qPCR and immunofluorescence staining results showed that the mRNA and protein level of Slit3 and Robo4 in TG-treated mouse TKE2 cells were upregulated markedly (Figs. 4A–C). In addition, through ELISA assay, we found that the concentration of SLIT3 in the lysis solution of TG treated mouse TKE2 cells was increased significantly (Fig. 4D). To detect the effect of ER stress on SLIT3 and ROBO4 expression during corneal wound healing in vivo, healed corneas with or without TG treatment were collected at 24 hours post-wounding and subjected to qPCR and immunofluorescence staining. The results indicated that TG treatment induced higher mRNA and protein expression of SLIT3 and ROBO4 in the epithelium of wounded corneas (Figs. 4E–G). 

Figure 4.

 

TG treatment upregulated the expression of SLIT3 and ROBO4 in mouse TKE2 cells and wounded corneas. (A) The qPCR detection of Slit3 and Robo4 mRNA expression in mouse TKE2 cells after TG treatment (n = 3/group). (B, C) After 24 hours of treatment, the cells were collected for SLIT3 and ROBO4 staining and the relative protein expression of SLIT3 and ROBO4 were analyzed (n = 4/group). (D) The protein concentration of SLIT3 in mouse TKE2 cells lysis solution was detected through ELISA (n = 3/group). (E) Relative mRNA expression of Slit3 and Robo4 in mouse wounded corneal epithelium with or without TG treatment 24 hours after epithelial scraping. (F, G) Representative images of SLIT3 and ROBO4 staining in wounded corneas of control and TG-treated mice at 24 hours post-wounding and the relative protein expression in limbus and wound margin were analyzed (n = 4/group). ***P < 0.001.

Application of SLIT3 Delayed Corneal Epithelial and Nerve Regeneration

To explore the effect of SLIT3 on corneal epithelial repair and nerve regeneration, we performed corneal epithelial curettage in mice subconjunctivally injected with recombinant SLIT3 protein. The unhealed areas in exogenous SLIT3–treated corneas were larger than those in normal corneas, as shown in Figure 5A. The epithelial defect areas in the normal and SLIT3-treated groups were 30.25 ± 0.75% and 62.26 ± 8.15% at 24 hours and 0.78 ± 0.52% and 6.58 ± 1.70% at 36 hours after wounding, respectively (Fig. 5B). On the contrary, inhibition of SLIT3 by application of SLIT3-specific siRNA partially reversed TG-induced delayed of corneal wound healing (Figs. 6A, 6B). Compared with the untreated group, the proportion of Ki67-positive cells in the limbus and wound margin of the SLIT3-treated group were decreased significantly (Figs. 5C, 5D), indicating a reduction in proliferation ability due to SLIT3. 

Figure 5.

 

Effect of recombinant SLIT3 protein on corneal epithelial and nerve repair. (A) Subconjunctival injections of SLIT3 or control solvent were performed after scraping a 2.5 mm diameter area of the corneal epithelium. The remaining defect was stained with fluorescein sodium and photographed at 0 hours, 24 hours, 36 hours, and 48 hours after injury (n = 4/group). (B) Epithelial defects at 24 hours and 36 hours were quantified as the percentage of defect area relative to the original wound area. (C) Representative images of Ki67 staining of limbus and wound margin in the control and SLIT3-treated corneas. (D) The percentage of Ki67-positive cells of limbus and wound margin in the control and SLIT3-treated corneas was calculated (n = 4/group). (E) The corneal nerves in the control and SLIT3-treated groups were detected using whole-mount TUBB3 staining 24 hours after epithelial scraping (n = 4/group). The total (upper) and locally magnified (lower) figures are presented. (F) The peripheral and central nerve density in the control and SLIT3-treated groups was calculated through Image J analysis. ***P < 0.001.

Figure 6.

 

TG inhibited corneal epithelial wound healing through SLIT3. (A) SLIT3-specific siRNA partially blocked the inhibitory effect of TG on corneal epithelial wound healing (n = 4/group). (B) Epithelial defects at 24 hours were quantified as the percentage of defect area relative to the original wound area. (C) The qPCR detection of Slit3 and Robo4 in mouse wounded corneal epithelium after TG treatment (n = 4/group), ***P < 0.001.

The proliferation and migration abilities of cells are crucial for epithelial wound healing, here, we studied the effect of SLIT3 on the proliferation and migration in mouse TKE2 cells. The CCK8 detection results showed that SLIT3 significantly decreased the proliferation ability of mouse TKE2 cells (Supplementary Fig. S4A). Consistently, the proliferation marker of Ki67 expression in SLIT3 treated group was reduced (Supplementary Figs. S4B, S4C). Next, a scratch experiment was performed to verify the effect of SLIT3 and TG on cell migration. Compared to the control group, SLIT3-treatment inhibited the cell migration significantly, and TG treatment with or without SLIT3 almost completely blocked the migration ability of cells (see Supplementary Figs. S4D, S4E). 

We further examined TUBB3 staining 24 hours and 5 days after epithelial scraping and found that SLIT3 inhibited nerve regeneration (Fig. 5E, Supplementary Fig. S5A). SLIT3 reduced the nerve density in the peripheral cornea by 58.98 ± 2.87% and 46.61 ± 1.52% 24 hours and 5 days after injury, respectively; and SLIT3 reduced the nerve density in the central cornea by 100% and 75.25 ± 2.89% 24 hours and 5 days after injury, respectively (Fig. 5F, Supplementary Fig. S5B). These findings proved that SLIT3 suppressed the regeneration of the corneal epithelium and nerves dramatically, similar with the effects of TG. 

ROBO4 Receptor Mediated the Inhibitory Effect of TG and SLIT3 on Corneal Epithelial and Nerve Regeneration

SLIT3 treatment dramatically enhanced the mRNA and protein level of ROBO4 in mouse corneal epithelium wound healing (Supplementary Figs. S6A–S6C) and in cultured mouse TKE2 cells (Supplementary Figs. S6D–F). To verify whether TG induced ER stress increases ROBO4 levels through SLIT3, the SLIT3-specific siRNA was used. The results showed the enhancement of SLIT3 and ROBO4 induced by TG was significantly inhibited by SLIT3-specific siRNA (Fig. 6C), indicating that TG promotes the increase of ROBO4 expression through SLIT3. 

To demonstrate that ER stress–induced SLIT3 exerts inhibitory effects on corneal repair through its receptor ROBO4, we used specific siRNA to knock down the expression of ROBO4 in mouse corneas. Knocking down ROBO4 completely alleviated the inhibitory role of SLIT3 in epithelial repair (Figs. 7A, 7B). The effectively knocking down of ROBO4 by siRNA was demonstrated by Western blotting assay (Figs. 7C, 7D). The Ki67-staining results showed that knocking down ROBO4 significantly restored the proliferative capacity of the epithelial cells inhibited by SLIT3 (Figs. 7E, 7F). To investigate the recovery role of ROBO4 knockdown on SLIT3-suppressed nerve regeneration, we performed TUBB3 immunostaining. The results indicated that knocking down ROBO4 effectively attenuated the inhibitory effect of SLIT3 on nerve regeneration (Figs. 7G, 7H). These findings suggest that the inhibitory effects of SLIT3 on corneal epithelial and nerve regeneration were mediated through its receptor, ROBO4. 

Figure 7.

 

Effect of ROBO4 siRNA on SLIT3-inhibited corneal epithelial and nerve regeneration. (A, B) ROBO4-specific siRNA effectively blocked the inhibitory effect of SLIT3 on corneal epithelial wound healing (n = 4/group). (C, D) Western blot assay showed that SLIT3 induced increased ROBO4 expression was inhibited by ROBO4-specific siRNA. (E, F) ROBO4-specific siRNA increased the proportion of Ki67-positive cells in SLIT3-treated healing corneal epithelium (n = 4/group). (G, H) ROBO4-specific siRNA reversed the inhibitory effect of SLIT3 on nerve regeneration (n = 4/group). ***P < 0.001.

Considering that TG-induced ER stress significantly upregulated the expression of SLIT3 and ROBO4, we hypothesized that ER stress impacts corneal epithelial wound healing and nerve regeneration through the SLIT3–ROBO4 signaling pathway. To validate this hypothesis, we used ROBO4 siRNA under TG-induced ER stress. Results from the epithelial scraped assay and subsequent staining demonstrated that knocking down ROBO4 partially reversed the inhibitory effects of TG on corneal epithelial healing (Figs. 8A, 8B). Western blotting results demonstrated the effectively knocking down of ROBO4 by siRNA (Figs. 8C, 8D). TG induced Ki67 expression was reversed by ROBO4-specific siRNA (Figs. 8E, 8F). In addition, corneal peripheral nerve regeneration delayed by TG was partially recovered by ROBO4-specific siRNA (Figs. 8G, 8H). These findings indicate that ER stress inhibits corneal epithelial healing and nerve regeneration through the upregulation of the SLIT3–ROBO4 pathway to a certain degree. 

Figure 8.

 

Effect of ROBO4 siRNA on TG-inhibited corneal epithelial and nerve regeneration. (A, B) The inhibition of corneal epithelial wound healing induced by TG was partially reverted by ROBO4-specific siRNA application (n = 4/group). (C, D) Western blot assay showed that application of ROBO4-specific siRNA inhibited TG induced increased ROBO4 expression. (E, F) The proportion of Ki67-positive cells in wounded corneal epithelium, reduced by TG, was partially restored by the application of ROBO4-specific siRNA (n = 4/group). (G, H) Peripheral corneal nerve density reduction during wound healing due to TG was partially restored by ROBO4-specific siRNA (n = 4/group). ***P < 0.001.

By the way, the activity of four specific siRNAs for SLIT3 and ROBO4 synthesized by GenePharma was validated using a TG-treated corneal epithelial injury model, and ultimately selected SLIT3-specific siRNA4 and ROBO4-specific siRNA3, which had the most significant recovery effect on TG-inhibited corneal epithelial wound healing in the subsequent experiments (Supplementary Fig. S7). 

Discussion

ER stress plays a significant role in corneal pathology, affecting the regeneration of the epithelium and nerves, but the specific molecular mechanism remains unclear.^14,36 This study aims to investigate the potential molecular mechanisms underlying the ER stress–induced delayed repair of the corneal epithelium and nerves. TG, a cell-permeable sesquiterpene lactone derived from the plant thapsiagarganic, can induce ER stress by specifically inhibiting ER Ca²⁺-ATPase, making it a suitable inducer for studying the mechanisms of ER stress.^37,38 Here, we induced ER stress in the cornea in vivo by subconjunctival injection of TG and found that ER stress leads to delayed wound healing in the corneal epithelium and nerve regeneration. These findings are consistent with previous research on the effects of ER stress on corneal epithelial and nerve repair.^14,39 We further induced ER stress using TG in cultured HLECs and mouse TKE2 cells. There is a correlation between the morphology and function of cells. A study showed that cell size of human corneal epithelial cells are associated with stem cell property and proliferative capacity.⁴⁰ TG treated HLECs and mouse TKE2 cells displayed irregular elongated shapes and enlarged size, which may be related to the decreased stem cell activity and proliferation ability in TG treated cells. 

To gain more insight into the mechanism of ER stress intervention in regeneration, we performed RNA sequencing on cells with or without TG treatment. Through KEGG analysis, we found that the genes related to development and regeneration were mainly enriched in axonal guidance factors. In particular, the expression levels of SLIT3 and ROBO4, a protein–receptor pair,⁴¹ were markedly increased. Immunofluorescence and qPCR experiments also confirmed the significant upregulation of SLIT3 and ROBO4 in TG-treated TKE2 cells in vitro and in mice corneal epithelium in vivo, suggesting that the SLIT3–ROBO4 pathway may be a downstream pathway of ER stress inhibiting corneal repair. Studies in various tissues have found a regulatory relationship between AGMs and ER stress; most of these studies reported the inhibitory effect of AGMs on ER stress, including molecules of the netrin, ephrin, and semaphorin families.^42–46 However, in the study of neurons, it was found that ER stress has a degradation effect on netrin-1.⁴⁷ To our knowledge, this study is the first report on ER stress regulating the SLIT family. 

In the SLIT family, SLIT2 has been shown to promote diabetic corneal epithelial repair according to our group's previous study²⁸ and inhibit corneal neovascularization.⁴⁸ However, the roles of SLIT1 and SLIT3 in the cornea have not been studied. SLIT3 is a protein secreted by the SLIT family of glycoproteins and is often considered a proangiogenic factor.^49,50 Several studies have confirmed that SLIT3 can regulate bone formation and regeneration by promoting angiogenesis.^51–54 Here, we found that the cornea as an avascular tissue has low expression of SLIT3; however, under ER stress, SLIT3 expression in the corneal epithelium was greatly increased. Exogenous SLIT3 supplementation significantly inhibited corneal epithelial wound healing, and, on the contrary, TG caused a delay of corneal wound healing that was reversed by inhibition of SLIT3. Given that corneal epithelial repair relies on the proliferation and migration of corneal epithelial cells, we conducted in vitro experiments to study the effect of SLIT3 on the proliferation and migration of mouse TKE2 cells. The results indicated that SLIT3 may suppress corneal re-epithelialization by inhibiting the proliferation and migration ability of corneal epithelial cells. In line with our findings, previous research showed that SLIT3 overexpression significantly reduces granulosa cell proliferation.⁵⁵ In terms of neural repair, recent studies have shown that SLIT3 plays a crucial role in controlling the formation of new neural bridges and proper axon regeneration after peripheral nerve transection injuries,⁵⁶ as well as in peripheral nerve repair after injury.⁵⁷ The cornea is the most densely innervated tissue of the body; after corneal injury, the regeneration of the local sensory nerves is regulated by numerous complex molecules.^19,20 Different AGMs have distinct effects on corneal nerve growth. For example, during early development, both Sema3A and SLIT2 are involved in preventing nerves from entering the corneal stroma. However, in the later stages of development, Sema3A continues to negatively regulate nerve entry into the corneal epithelium, whereas SLIT2 actively regulates nerve growth by increasing nerve branching within the corneal epithelium.⁵⁸ However, there are currently no reports on the effect of SLIT3 on corneal nerve innervation. The present study revealed that SLIT3 inhibits the regeneration of nerve fibers during corneal wound healing, indicating that it acts as a negative regulator for corneal nerve fiber regeneration. These findings suggest that ER stress may impair normal corneal regeneration by inducing SLIT3 overexpression. 

SLIT3 primarily exerts its effects through binding to its receptors, which include ROBO1, ROBO2, ROBO3, and ROBO4.⁴¹ Our RNA sequencing data and in vitro/vivo experimental validation demonstrate a significant upregulation of ROBO4 expression in corneal epithelial cells under ER stress. In multiple studies on angiogenesis, it has been reported that there is a SLIT3-ROBO4 pathway.^50,59–61 Studies in vascular endothelial cells have shown that ligand SLIT3 inhibits ROBO4 shedding to enhance ROBO4 signaling.⁶² In the present study, we found that the increased expression of SLIT3 and ROBO4 induced by TG were significantly inhibited by SLIT3-specific siRNA, indicating that TG induced ER stress enhanced ROBO4 expression by SLIT3. However, the interaction mechanism between SLIT3 and ROBO4 in corneal epithelial cells still needs to be elucidated in future research. ROBO4-specific siRNA effectively restored ER stress and SLIT3 delayed corneal epithelial and nerve repair, suggesting that ER stress–induced SLIT3 overexpression may suppress the regeneration of corneal epithelium and nerves mainly through ROBO4 receptors. Zhang et al. found that ROBO4 deficiency accelerates skin wound healing.⁶³ Cai et al. showed that inhibition of ROBO4 expression promotes the proliferation and migration of human brain microvascular endothelial cells (HBMECs).⁶⁴ These studies indirectly support our conclusions. However, we have not yet identified the downstream mechanisms by which SLIT3–ROBO4 signaling affects corneal epithelial repair and nerve regeneration under ER stress. 

In summary, our findings confirmed that ER stress inhibits corneal epithelial and nerve regeneration by activating the SLIT3–ROBO4 pathway. This study reveals a new mechanism by which ER stress regulates disease pathology and provides a new potential target for the treatment of corneal wound healing. 

Acknowledgments

Support from the Natural Science Foundation of Shandong Province (ZR2022MH026 and ZR2020QH146), the National Natural Science Foundation of China (82371029 and 82000851), Qingdao Shinan District Science and Technology Plan Project (2022-2-018-YY), Youth Science Fund Cultivation Support Program Project of Shandong First Medical University (202201-121), the Innovation Project of Shandong Academy of Medical Sciences, the Academic promotion program of Shandong First Medical University (2019ZL001), and the Taishan Scholar Project of Shandong Province (tsqn20221163). 

Author Contributions: R. Chen conducted experiments, collected and analyzed data, and wrote the manuscript. Y. Wang conducted experiments and collected data. Z. Zhang conducted some experiments and analyzed some data. X. Wang, Y. Li, M. Wang, H. Wang, and M. Dong analyzed the data. Q. Zhou guided the idea of the article. L. Yang designed the study, analyzed the data, and contributed to writing the manuscript. 

Disclosure: R. Chen, None; Y. Wang, None; Z. Zhang, None; X. Wang, None; Y. Li, None; M. Wang, None; H. Wang, None; M. Dong, None; Q. Zhou, None; L. Yang, None