Type 2 Diabetes Mellitus Makes Corneal Endothelial Cells Vulnerable to Ultraviolet A-Induced Oxidative Damage Via Decreased DJ-1/Nrf2/NQO1 Pathway

Abstract

Purpose: The purpose of this study was to investigate whether type 2 diabetes mellitus (T2DM) makes corneal endothelial cells (CECs) suffer from more severe ultraviolet A (UVA)-induced oxidative damage and explore its mechanisms via measuring the oxidant level and the antioxidant level in vitro.

Methods: Corneas of spontaneous T2DM db/db mice and non-diabetes littermate control mice were irradiated with UVA, leading to oxidative damage of CECs. Anterior segment–optical coherence tomography, corneal image, and CECs immunohistochemistry staining were taken thereafter to measure central corneal thickness, corneal edema degree, and damage extent of CECs. In vitro, human corneal endothelial cells line B4G12 (HCECs) treated with high glucose (HG) and low glucose (LG) were exposed to UVA light separately. Subsequently, cellular proliferation, apoptosis, pro-oxidant factors, such as reactive oxygen species (ROS), antioxidant factors including Parkinson's disease protein 7 (DJ-1), nuclear factor-erythroid 2 related factor 2 (Nrf2), phosphorylated-Nrf2, and NAD(P)H: quinone oxidoreductase 1 (NQO1) were measured.

Results: T2DM mice presented greater oxidant damage of CECs and more distinct corneal edema compared with control mice when they were irradiated with the 150 J/cm² UVA light. In vitro, HCECs in HG condition showed a significant decrease of proliferation, higher apoptosis extent, more ROS generation, lower expressions of DJ-1/Nrf2/NQO1, and distinct reduction of Nrf2 nuclear translocation compared to those in LG condition after exposing to 5 J/cm² UVA light.

Conclusions: Increase of ROS, downregulation of DJ-1/Nrf2/NQO1 expressions, and decrease of Nrf2 nuclear translocation could result in that T2DM makes CECs more vulnerable to oxidative damage.

Corneal endothelium is composed of a layer of corneal endothelial cells (CECs) serving as a barrier to keep corneal stroma apart from the aqueous humor.¹ Without corneal endothelium, corneal stroma will be swelled, as hydrophilic proteoglycans inside it can make contact to the fluid from aqueous humor, leading to impaired visual function.² For a healthy cornea, although the water in aqueous humor can be leaked into corneal stroma owing to the imbibition force of proteoglycans, extra solutes in the stroma can be transported out via the pump function of CECs, consequently arousing the passive withdrawal of water and further maintaining corneal transparency.³ As is well known, human CECs have limited ability to proliferate in vivo whereas the loss of CECs can only be compensated by the expansion and migration of remaining cells.⁴ When CECs’ density declined significantly, it will bring about corneal endothelial decompensation and even corneal blindness.⁵ 

On account of its location, the cornea, especially the central cornea, is directly exposed to the ultraviolet radiation.⁶ Ninety-five percent of ultraviolet light consists of ultraviolet A (UVA) light (320 nm to 400 nm), which is able to penetrate corneal full-thickness.⁷ UVA can induce the generation of intracellular reactive oxygen species (ROS), whereas excessive ROS will lead CECs to an oxidative stress and cause oxidative damage later on.⁸ Oxidative stress is generally regarded as an imbalanced state between pro-oxidative factors and anti-oxidative factors, which also takes part in the pathogenesis of corneal endothelial decompensation.⁹ ROS, as a classic pro-oxidative factor, could destroy intracellular proteins, lipids, and nucleic acids.¹⁰ Whereas, anti-oxidative factors generally refer to antioxidant enzymes mediated by antioxidant response elements (AREs).¹¹ AREs include NAD(P)H: quinone oxidoreductase 1 (NQO1), hemeoxygenase-1 (HO-1), glutamate cysteine ligase (GCS), et cetera, and are regulated by nuclear factor-erythroid 2 related factor 2 (Nrf2).¹² Liu et al.¹³ find that downregulating the expression of DJ-1 (Parkinson's disease protein 7) can impair the nuclear translocation of Nrf2, thus resulting in decreased AREs expression and increased oxidative damage. Therefore, in this study, we measured the ROS production and the expression of DJ-1/Nrf-2/NQO1 to elucidate the level of oxidant and antioxidant, respectively. 

Diabetes mellitus (DM), a common chronic systematic disease that has already been in the state of oxidative stress,¹⁴ is considered as a possible negative factor for corneal endothelial function as well. The reported effects of DM on CECs include decreased cell density, increased pleomorphism, and polymegathism compared with healthy subjects.^15,16 But some studies do not find CECs changed in patients with DM.^17,18 Moreover, both Zhang et al.¹⁹ and Chen et al.²⁰ find that patients with DM undergoing cataract surgery have more losses of CECs than non-diabetic patients. The free radical generated from phacoemulsification is an extra oxidative stimulation that could damage the CECs.²¹ Nevertheless, the reason why the CECs of patients with DM are more vulnerable to the external oxidative stimulation is still unclear. Currently, up to 11% of the population is diagnosed with DM and more than 95% of them are type 2 diabetes mellitus (T2DM).²² Hence, it is necessary to study whether T2DM makes CECs more vulnerable to oxidative stress and explore its underlying mechanism, seeking for possible ways to ease oxidative damage of CECs brought by T2DM. 

Materials and Methods

Experimental Animals

Spontaneous type 2 diabetes db/db mice (T2DM mice; BKS/DB ko/ko) and non-diabetes littermate mice (control mice; BKS/DB wt/wt) were purchased from the GemPharmatech, Nanjing, China (male mice, 8 to 12 weeks old). Mice were housed in a constant temperature and humidity environment with a 12 hour light/dark cycle and maintained on a normal diet. For anesthetizing, mice were intraperitoneally injected with a combined dose of 100 mg/kg ketamine and 20 mg/kg xylazine. Experimental protocols were approved by the Ethics Committee of Fudan University Eye Ear Norse and Throat Hospital. 

Human Corneal Endothelial Cells Culture

The human corneal endothelial cells line B4G12 (HCECs) were purchased from Creative Bioarray (Shirley, NY, USA). To improve cytocompatibility, cell culture plates were precoated with 10 mg/mL chondroitin sulfate (Sigma-Aldrich) and 10 µg/mL of laminin (Sigma-Aldrich) at 37°C for 1 hour. Then, HCECs were seeded into the plates and grown in human endothelial serum-free medium (Gibco Invitrogen), which were supplemented with 10 ng/mL recombinant human basic fibroblast growth factor (PeproTech, London, UK) and 3% fetal bovine serum (Gibco Invitrogen). HCECs were cultured in 5% CO₂ atmosphere at 37°C. For the high glucose (HG) group, HCECs were treated with 25 mmol/L glucose, after cell attachment. For the hyperosmotic (HP) group, HCECs were treated with 5 mmol/L glucose plus 24.5 mmol/L mannitol, after cell attachment. The control group was treated with 5 mmol/L glucose and marked as the low glucose (LG) group. Both groups were cultured for 3 days in the different media. 

UVA Irradiation In Vivo

Only one cornea of the mouse was irradiated with UVA. A 4-mm-diameter light spot was focused with a convex lens (LB4972-uv; Thorlabs) from a 365 nm UVA LED light source (M365LP1-C1; Thorlabs). Modulated by the LED driver (LEDD1B-T Cube; Thorlabs), the irradiance of UVA light spot was 120 mW/cm² and the UVA exposure time was 6 minutes and 57 seconds for 50 J/cm², 20 minutes and 50 seconds for 150 J/cm², and 34 minutes and 44 seconds for 250 J/cm². The UVA irradiance was measured by an ultraviolet illuminometer. 

UVA Irradiation In Vitro

HCECs were irradiated by a 365 nm UVA lamp (UVL-26; UVP, Upland, CA, USA) with an irradiance of 5 mW/cm². The UVA exposure time was adjusted to deliver the appropriate fluence (8 minutes and 20 seconds for 2.5 J/cm², 16 minutes and 40 seconds for 5 J/cm², 25 minutes for 7.5 J/cm², and 33 minutes and 20 seconds for 10 J/cm²). HG cells and LG cells were irradiated with UVA irradiation at 72 hours after culturing in different media. UVA-treated cells were incubated in 5% CO₂ atmosphere at 37°C, and then collected at 24 hours post-UVA irradiation. Controls were cultured under identical conditions except for UVA exposure. 

In Vivo Imaging

Anterior segment–optical coherence tomography (AS-OCT) was used to capture the anterior segment image of mouse, and the central corneal thickness (CCT) was measured by its inbuilt software. A slit-lamp biomicroscope with a camera was taken to photograph the mouse corneal image. Furthermore, fluorescein was dropped in the mouse conjunctival sac and observed corneal epithelial staining under the cobalt blue light to evaluate the integrity of the mouse corneal epithelium. 

Immunochemistry and Histology

Corneas of mice were dissected under a binocular microscope and fixed with 4% PFA (P0099; Beyotime) for 1 hour. Then, mouse corneas were permeabilized with 0.1% Triton X-100 (P0096; Beyotime) in PBS for 5 minutes and blocked in 3% bovine serum albumin (BSA) in PBS for 30 minutes. Subsequently, anti-zonula occludens-1 (ZO-1) antibody (40-2200; Invitrogen) and anti-sodium potassium ATPase (Na⁺-K⁺-ATPase) antibody (Ab76020; Abcam) in 1% BSA-PBS was used to incubate the corneal cups at 4°C overnight, respectively. After that, secondary anti-rabbit fluorescein isothiocyanate (FITC; ab150077; Abcam) was applied to conduct the secondary antibody incubation. Corneal cups were flattened by four radial cuts and mounted by DAPI mounting medium (0100-20; Southern Biotech). The graphs of ZO-1 and Na⁺-K⁺-ATPase staining were captured using the laser confocal scanning microscope (Leica SP8; Wetzlar, Germany). 

For periodic acid–Schiff (PAS) staining, paraffin tissue slices of mouse corneas were dewaxed and oxidized by 0.5% periodate acid solution. Then, corneal slices were stained in PAS reagent in the dark for 30 minutes. Subsequently, nuclear dying using hematoxylin, dehydration, and slice seal were performed. The images of PAS staining were obtained using an optical microscope. 

Cell Viability Assay

The proliferative capability of HCECs was detected by the Cell Counting Kit-8 (CCK-8; CK04; Dojindo). Briefly, HCECs were seeded into 96-well plates in quintuplicate (1.0 × 10⁴ cells/well), allowing them to adhere for 48 hours. After the required treatment, HCECs were added with the CCK-8 reagent (10 µl/well) and then incubated at 37°C for 2 hours. The optical density (OD) value at 450 nm was measured by an automatic microplate reader (Gen5; BioTek, Winooski, VT, USA). 

Detection of Apoptosis

HCECs with the complete medium were transferred into 1.5 mL reaction tubes after the required treatment. HCECs were centrifuged at 1000 rpm for 5 minutes at room temperature and the supernatant was discarded. Next, the HCECs were washed with 1 mL PBS solution and repeated the process of centrifugation, and discarded the supernatant. The centrifugal HCECs were resuspended with 500 µL 1 × annexin V binding solution, 5 µL annexin V FITC, and 5 µL PI solution, according to the manufacturer’s instructions (AD10; Dojindo). HCECs were incubated in the mixed reagent for 15 minutes at room temperature in a dark place. At least 10,000 single cells per sample were collected and detected using flow cytometry at Bright field (430–480 nm), annexin V FITC (505–560 nm), and PI (595–642 nm) channel. 

Quantification of ROS

After the required treatment, complete media of HCECs were removed, and reconstituted ROS indicator solution (C13293; Invitrogen) was added to the HCECs. Then, the HCECs were incubated at 37°C for 30 minutes and protected from light. After that, the ROS solution was removed and the HCECs were washed in PBS three times. At least 10,000 single cells per sample were collected and detected using flow cytometry at 488 excitation wavelength and 525 nm emission wavelength. 

Real-Time Quantitative Polymerase Chain Reaction

Total RNA was extracted (CW0581S; CWBIO) and cDNA was prepared via reverse transcription (CW2020M; CWBIO), according to the manufacturers’ instructions. The process of real-time quantitative PCR (q-PCR; CW3008M; CWBIO) was performed on Bio-Rad CFX96 (Bio-Rad Laboratories, Hercules, CA, USA). TaqMan Primers for DJ-1, Nrf2, NQO1, and β-actin were obtained from Sangon Biotech, Shanghai, China. The comparative threshold cycle (CT) method was used for data analysis. The sequences of primers are listed in Supplementary Table S1. 

Western Blot Analysis

HCECs were collected in radioimmunoprecipitation assay (RIPA) lysis buffer with protease and phosphatase inhibitor cocktail (78443; Invitrogen). After crushing using an ultrasonic pulverizer, centrifuging, and denaturation, total proteins of samples were extracted. Then, total proteins (30 to 40 µg) were separated by 4% to 20% SDS-polyacrylamide gel electrophoresis (SDS-PAGE; M00655; GenScript) and transferred to the polyvinylidene fluoride (PVDF) membranes (IPVH00010; Millipore). After blocking in 5% defatted milk for 1 hour at room temperature, PVDF membranes were incubated with the following specific primary antibodies (1:1000): DJ-1 (ab76008; Abcam), Nrf2 (ab137550; Abcam), p-Nrf2 (BS-2013R; Bioss), NQO1 (11451-1-AP; Proteintech), β-actin (4970; Cell Signaling), and histone h3 (4499T; Cell Signaling) at 4°C overnight. The secondary antibody (1:5000) applied to these membranes was horseradish peroxidase (HRP)-conjugated goat anti-rabbit antibody (ab205718; Abcam). The protein bands were detected using a Gel Doc 1000 imaging analysis system and analyzed using the Image J software (http://imagej.nih.gov/ij/). 

Statistical Analysis

Statistical analysis was carried out using Stata version 15.1. Data were expressed as the means ± standard deviation (SD) and normality of them was tested by the Shapiro-Wilk normality test. Independent sample t-test was used to compare values of two groups in normally distributed. When values were non-normally distributed, Wilcoxon rank-sum test was applied. Differences among groups were analyzed using 1-way analysis of variance (ANOVA) and Bonferroni test if they were normally distributed. If not, the Kruskal-Wallis equality-of-populations rank test and Scheffe test were used. The value of P < 0.05 was considered as statistically different. 

Results

To Determine the Energy of UVA Irradiation for Study In Vivo and In Vitro

In vivo, UVA light irradiated the left eye of BKS/DB wt/wt mouse with the energy of 50 J/cm², 150 J/cm², and 250 J/cm², respectively. The changes of the CCT values are presented in Supplementary Table S2 and Figure 1A. The corneal images and ZO-1 staining of CECs are displayed in Figures 1B and 1C. The fluence of 150 J/cm² UVA caused corneal edema, prominent endothelium damage, and fewer epithelium defect to mice. Thus, the 150 J/cm² irradiation was performed for further study. 

Figure 1.

 

The CCT, corneal images, CECs staining in vivo and cellular proliferation in vitro after different doses of UVA irradiation. (A) The changes of CCT in BKS/DB wt/wt mice treated with 50, 150, and 250 J/cm² UVA exposure (n = 3 in each group). Data are shown as mean ± SD; P < 0.05. The * and the # represent the statistical difference before and after UVA irradiation in the 150 J/cm² group and the 250 J/cm² group, respectively. (B) The images of mice cornea before and after 50, 150, and 250 J/cm² UVA exposure. (C) ZO-1 staining of mice before UVA and at day 14 post-50, −150, and −250 J/cm² UVA. (D) The reduction of cellular proliferation post-2.5, −5, −7.5, and −10 J/cm² UVA. Data are shown as mean ± SD; *P < 0.05. CCT, central corneal thickness; CECs, corneal endothelial cells; UVA, ultraviolet A; CCK8, cell counting kit-8; OD, optical density; SD, standard deviation.

In vitro, the proliferative capability of HCECs in the HG group was declined to 67%, 34%, 6%, and 3% after exposed to 2.5 J/cm², 5 J/cm², 7.5 J/cm², and 10 J/cm² UVA light (Fig. 1D). Thus, we chose 5 J/cm² to be the cellular UVA irradiation dose due to the cellular proliferative viability being decreased more than half (P < 0.05). 

UVA Irradiation Increases Central Corneal Thickness to a Greater Extent in T2DM Mice at Early Stage

The degree of corneal edema was measured by CCT (Figs. 2A, 2B; Supplementary Table S3) and also showed by anterior segment photography (Supplementary Fig. S1). Before UVA irradiating, the CCT was slightly higher in T2DM mice compared with control mice (94.8 ± 2.4 µm versus 91.3 ± 4.3 µm, P < 0.05). Under the exposure of 150 J/cm² UVA light, both mice presented acute growths in CCT. However, the CCT values of T2DM mice were 1.15, 1.30, and 1.37 times higher than those of control mice at day 0 (106.0 ± 10.7 µm versus 92.3 ± 3.4 µm, P < 0.05), day 1 (218.1 ± 29.4 µm versus 167.8 ± 37.1 µm, P < 0.05), and day 2 (213.9 ± 27.1 µm versus 156.1 ± 40.0 µm, P < 0.05), respectively. Day 0 refers to the immediate time point post UVA irradiation. The CCT of control mice gradually moved to baseline at day 14, whereas the CCT of T2DM mice was slightly lower than before at day 7 (88.6 ± 3.7 µm versus 94.8 ± 2.4 µm, P < 0.05) and day 14 (85.4 ± 6.9 µm versus 94.8 ± 2.4 µm, P < 0.05). Subsequently, PAS staining was performed to show the corneal layers of mice at day 14 post-UVA (Fig. 2C). We found that the Descemet’s membrane became thicker, the cellular layers of the epithelium decreased, and the collagen stacked in post-stroma in both mice after exposed to UVA. 

Figure 2.

 

Comparisons between T2DM mice and control mice in CCT, structure of corneal layers, and CECs oxidative injury post-150 J/cm² UVA irradiation. (A) AS-OCT images of mice corneas at different points of time. (B) The CCT changes of mice treated with 150 J/cm² UVA (n = 10 in each group). Data are shown as mean ± SD; P < 0.05. The * represents the statistical difference between T2DM mice and control mice at the same point of time. The # and + represent the statistical difference of CCT before and after UVA exposure in T2DM mice and control mice, respectively. (C) PAS staining of mice before and at day 14 post-UVA. (D) ZO-1 staining of mice before UVA, at day 3, day 7, and day 14 post-UVA (63 ×). (E) Na⁺-K⁺-ATPase staining of mice before UVA and at day 14 post-UVA (63 ×). T2DM, type 2 diabetes mellitus; CCT, central corneal thickness; CECs, corneal endothelial cells; UVA, ultraviolet A; AS-OCT, anterior segment–optical coherence tomography; SD, standard deviation; PAS, periodic acid–Schiff; ZO-1, zonula occludens-1; Na⁺-K⁺-ATPase, sodium potassium ATPase.

In line with the variations of CCT, the corneal images demonstrated that corneas of both mice became obvious edema at day 1, day 2, and day 3 post UVA irradiation, whereas they tended to be transparent gradually at day 14. Although there were some corneal epithelial defects in both mice at day 1, their corneal epithelium had recovered after 1 to 2 days (see Supplementary Fig. S1). 

UVA Irradiation Causes Progressive Alterations in Mouse Corneal Endothelial Cell Morphology and Greater Cell Loss in T2DM Mice

The ZO-1 immunofluorescence staining images of mice are displayed in Figure 2D. There was no prominent difference of CECs structure between the two mice before being exposed to UVA light. After being irradiated, the CECs in both mice became altered at day 3 and distinctly abnormal at day 7. Then the polymegathism and the pleomorphism of CECs became more obvious in T2DM mice at day 14 post-UVA lighting. In addition, the area of Na+-K+-ATPase decreased after being exposed to UVA, especially in T2DM mice (Fig. 2E). In addition, a broader view of CECs staining is shown in Supplementary Figure S3. 

High Glucose Decreased Cellular Proliferation Capability, Induced Apoptosis, and Generated More ROS Under Oxidative Stress From UVA Light In Vitro

The cell images of all four groups are displayed in Figure 3A. Under the environment of HG, the proliferation of HCECs significantly declined to 56% and it further decreased to 31% after being irradiated with UVA light, compared to the LG group. There was no statistical difference in proliferation of the LG group before or after UVA exposure (Fig. 3B). The apoptosis degree in the HG + UVA group was almost two times higher than that in the other three groups. HCECs in the LG + UVA group slightly eased the apoptosis degree compared with the LG group (Figs. 3C, 3D). No statistical difference was found between the HG group and the LG group. Similar to the outcomes of apoptosis, the level of ROS in the HG + UVA group was significantly increased compared with the other three groups. The ROS fluorescence intensity in the HG + UVA group showed a 4.4- and 3.2-fold increase than that in the LG + UVA group and the HG group, respectively. The ROS levels in the LG + UVA group and the HG group were also slightly higher than that of the LG group (Fig. 3E). 

Figure 3.

 

HG decreases proliferative capacity, induces apoptosis and increases ROS production of HCECs after exposed to 5 J/cm² UVA. (A) The images of HCECs in the HG group and the LG group before and after UVA exposure. (B) The proliferation changes of HCECs in the four groups (n = 3). (C, D) The apoptosis degree of HCECs in the four groups (n = 3). (E) The ROS production of HCECs in the four groups (n = 3). Data are shown as mean ± SD; P < 0.05. The * represents the statistical difference between the LG group and any of the other three groups. The # represents the statistical difference between the HG + UVA group and any of the other three groups. HG, high glucose; ROS, reactive oxygen species; HCECs, human corneal endothelial cells; UVA, ultraviolet A; LG, low glucose; SD, standard deviation.

On account of the hyperosmotic condition brought by HG, a standard mannitol control was set and no statistical difference was found between the LG group and the HP group, ascribing the effects that HG played on HCECs to hyperglycemia but not HP (see Supplementary Fig. S2). 

High Glucose Inhibited the Expressions of Antioxidative Genes Under Oxidative Stress From UVA Light In Vitro

To evaluate the antioxidant capacity, the expressions of the DJ-1/Nrf2/NQO1 pathway were measured in this study (Fig. 4). Under the circumstance of oxidative stress from UVA irradiation, the LG group showed a significant increase of the NQO1 gene in both mRNA and protein levels. Unlike the trend of NQO1 gene expression in the LG group, the expressions of the NQO1 gene in the HG group remained unchanged before or after UVA exposure compared to the LG group. The Nrf2 gene, as the upstream of NQO1 gene, was significantly decreased in the HG + UVA group, whereas it was significantly increased in the LG + UVA group at the protein level. Phosphorylated Nrf2 (p-Nrf2) was the function form of Nrf2 which connected with AREs in nucleus and promoted the process of antioxidants.²³ The expression of p-Nrf2 was significantly decreased in the HG + UVA group as well. Furthermore, compared with the LG group, a significant downregulation of the DJ-1 gene was found in the HG + UVA group and the HG group both in the mRNA level and protein level, whereas DJ-1 expression in the LG + UVA group remained unchanged. All the above indicated that antioxidant capacity of HCECs in HG condition was inhibited under the oxidative stress from UVA irradiation. 

Figure 4.

 

The decrease of DJ-1/Nrf2/NQO1 expression and Nrf2 nuclear translocation in HCECs of the HG group after exposed to 5 J/cm² UVA indicates its insufficiency of antioxidant capacity. (A) Representative q-PCR of DJ-1/Nrf2/NQO1 mRNA levels (n = 3). (B) Representative Western blot of DJ-1/Nrf2/p-Nrf2/NQO1 protein levels (n = 3). Data are shown as mean ± SD; P < 0.05. The * represents the statistical difference between the LG group and any of the other three groups. The # represents the statistical difference between the HG + UVA group and any of the other three groups. DJ-1, Parkinson's disease protein 7; Nrf2, nuclear factor-erythroid 2 related factor 2; p-Nrf2, phosphorylated Nrf2; NQO1, NAD(P)H quinone oxidoreductase 1; HCECs, human corneal endothelial cells; HG, high glucose; UVA, ultraviolet A; LG, low glucose; SD, standard deviation.

Discussion

Corneal endothelium utilizes the barrier function and the pump function of CECs to maintain the dehydration of cornea, which is crucial for corneal transparency.²⁴ According to reports, the obvious decrease of CECs’ density and hexagonality were observed under the irradiation of ultraviolet light.^25,26 Excessive ROS induced by ultraviolet light makes CECs suffer from oxidative stress, further leading to oxidant damage.²⁷ In our study, the way that made CECs get oxidant damage from UVA light referred to the research of Liu et al.²⁸ The irradiating dose of 150 J/cm² in vivo was determined in this study, which could cause prominent CECs damage and fewer epithelium defect. In vitro, considering the contents of total RNA and total protein, the irradiating dose of 5 J/cm² was used, which made the proliferation of HCECs in the HG environment fall by half. In addition, the determination of glucose concentration referred to the study of Zhang et al.²⁹ 

T2DM is one of the most prevalent metabolic disorders whose pathogenesis is closely related to oxidative stress.³⁰ Papadakou et al.³¹ found that the CECs’ density is significantly lower in patients with T2DM compared with healthy control subjects (2297.9 ± 311.3 vs. 2518.3 ± 243.7 cells/mm², P < 0.05). Çolak et al.³² also prove that patients with T2DM have lower CEC density (2409.4 ± 199.1 vs. 2589.1 ± 183.6 cells/mm², P < 0.001) and higher variation coefficient of CECs (41.1 ± 5.0 vs. 38.6 ± 5.4, P < 0.05) than non-DM patients. However, Cankurtaran et al.¹⁷ do not observe statistical difference between patients with T2DM and control subjects on CEC measurements. In our study, no distinct endothelium morphic variation was found between T2DM mice and control mice, but the CCT value was higher in T2DM mice (94.8 ± 2.4 vs. 91.3 ± 4.3 µm, P < 0.05). In keeping with our data, Kim et al.³³ find that streptozotocin-induced diabetic rats become corneal edemic and have higher CCT values than control rats. Nevertheless, Meyer et al.³⁴ found a significant decrease of CECs’ density and hexagonality in streptozotocin-induced diabetic rats. 

Moreover, the CECs of patients with T2DM seemed to be more vulnerable to the energy from phacoemulsification.³⁵ Khalid et al.³⁶ noticed that patients with T2DM get more CEC loss after cataract phacoemulsification than non-diabetic patients (2250.37 ± 426.68 vs. 2399.97 ± 451.08 cells/mm², P < 0.05). The study of Chen et al.²⁰ also demonstrates that the CEC density of patients with T2DM presents persistent loss compared to non-diabetic patients from day 3 (2628.07 ± 124.02 vs. 2740.92 ± 107.32 cells/mm², P < 0.05) to month 6 (2503.64 ± 104.61 vs. 2733.73 ± 108.98 cells/mm², P < 0.05) post-phacoemulsification. However, Misra et al.³⁷ do not discover that patients with DM are predisposed to greater endothelial loss after phacoemulsification. The process of phacoemulsification generates some ROS, such as hydroxyl free radicals, resulting in oxidant damage of CECs.³⁸ In our research, facing with oxidant injury from the same dose of UVA light, T2DM mice had a greater growth of CCT post-UVA from immediately to day 3, and more severe variation and defect of corneal endothelium at day 14 post-UVA compared to control mice, indicating T2DM made CECs much easier to get UVA-induced oxidant damage. In accord with the outcomes in vivo, HCECs in the HG condition showed a significant decrease in proliferative capacity and presented a further decline of proliferation after exposing them to the oxidant injury from UVA lighting, whereas HCECs treated with LG had no change of proliferation after UVA irradiating. Likewise, HCECs in the HG group appeared to have a significant increase of apoptosis degree after UVA irradiation. Interestingly, exposing with the same dose of UVA light, the apoptosis level of HCECs in the LG group was slightly decreased. This phenomenon could be explained by that lower fluences of UVA, such as 5 J/cm², are able to induce Nrf2-regulated antioxidant defense in normal condition, and then contributed to the decrease of the apoptosis degree.³⁹ 

The mechanical researches on why T2DM makes CECs more vulnerable to UVA-induced oxidant damage are still lacking now. In our study, the CECs’ line derived from human was adopted to explore the underlying mechanism. Oxidant level was evaluated by the production of ROS. UVA, as an environmental stressor, can promote the generation of ROS production.⁴⁰ As a chronic metabolic disease, T2DM also increases the level of ROS production, which disturbs the balance between oxidant and antioxidant and resulted in pro-oxidative condition.⁴¹ In line with our findings in vitro, HCECs treated with HG were in the situation of pro-oxidant and already had a higher ROS level. After receiving an extra oxidant stimulation from UVA irradiation, HCECs in the HG condition showed a further significant increase of ROS production, whereas HCECs in the LG group just had a slight increase of that. 

Confronted with the impacts of oxidant stimulation, some endogenous antioxidants would be generated to maintain the balance between them, protecting cells or tissues from oxidative stress.⁴² Endogenous antioxidants mainly refer to some antioxidant enzymes, which are induced by AREs.⁴³ NQO1, as a member of AREs, is remarkable for cell defense during cell stress.⁴⁴ We observed that the expression of NQO1 was significantly increased in HCECs of the LG environment when exposed to the oxidative stimulation from UVA irradiation, whereas no rise was found in the HG condition. Moreover, AREs including NQO1 can be induced by Nrf2, which plays an important role in regulating the physiological and pathophysiological process under oxidant exposure.⁴⁵ In the LG condition, the Nrf2 expression of HCECs increased significantly to withstand oxidative stimulation when faced with UVA light. However, the Nrf2 expression of HCECs with HG treatment showed a distinct decrease in both mRNA level and protein level. Nrf2 can be translocated into nucleus in the form of p-Nrf2 to activate the antioxidant cytoprotective pathway.⁴⁶ After exposing to UVA light, the level of p-Nrf2 was significantly decreased in HCECs treated with HG, whereas HCECs in the LG condition did not show any decline of that. Liu et al.¹³ discovered that downregulating DJ-1 could impair the nuclear translocation of Nrf2, leading to the insufficiency of antioxidative capacity and oxidative damage in HCECs. In our study, the expression of DJ-1 decreased remarkably in HCECs of HG environment before and after UVA exposure, whereas DJ-1 in the HCECs of the LG condition showed no decline. 

Oxidative stress is an imbalance between the level of oxidant and the capacity of antioxidant.⁴⁷ Thus, we speculated that HCECs in the HG condition had been in oxidative stress due to an increase of ROS production and insufficient expressions of antioxidant genes, which was consistent with the research of Wang et al.⁴⁸ on retinal pericytes. After exposure to the oxidative stimulation from UVA irradiation, the decreased expressions of antioxidant genes and the reduction of Nrf2 nuclear translocation could not act in response to over the amount of ROS, leading to HCECs in the HG condition being more easily to get damaged from the oxidative stress. 

In conclusion, T2DM made CECs more vulnerable to UVA-induced oxidative damage, which might be caused by the increase of ROS production and the decreased expressions of antioxidant genes. Improving the expressions of DJ-1/Nrf2/NQO1 gene or promoting the nuclear translocation of Nrf2 may contribute to improving the antioxidant capacity of CECs in patients with T2DM, hoping to protect them from oxidative injury. 

Acknowledgments

Supported by grants from the Science and Technology Commission of Shanghai Municipality (no. 19441901000) and the Shanghai Sailing Program (no. 21YF1405100). 

Ethics Approval: This study was approved by the Ethics Committee of Fudan University Eye Ear Nose and Throat Hospital. 

Disclosure: X. Zhang, None; J. Qiu, None; F. Huang, None; K. Shan, None; C. Zhang, None