O-GlcNAcylation Inhibition Upregulates Connexin43 Expression in the Endothelium to Protect the Tight Junction Barrier in Diabetic Retinopathy

Abstract

Purpose: This study aimed to investigate the effects of O-linked N-acetylglucosamine modification (O-GlcNAcylation) on connexin43 (Cx43) expression and its subsequent effects on tight junction properties in diabetic retinopathy (DR).

Methods: O-GlcNAcylation levels in primary human retinal vascular endothelial cells (HRVECs) and retinas from rats with diabetes were regulated by treatment with Thiamet G or alloxan. Immunoprecipitation was used to examine the relationship between O-GlcNAcylation and Cx43 expression. Stable overexpression and knockdown of Cx43 in HRVECs were achieved using lentivirus constructs; further, their effects on occludin and zonula occluden-1 (ZO-1) expression and tight junction barrier function were determined.

Results: O-GlcNAcylation level increased significantly, whereas Cx43 expression decreased in retinas from rats with diabetes and HRVECs cultured under high-glucose conditions. Immunoprecipitation revealed that Cx43 was modified by O-GlcNAcylation and phosphorylation simultaneously. O-GlcNAcylation inhibition negatively regulated both total Cx43 and phosphorylated Cx43 expression, subsequently disrupting tight junction properties. Conversely, Cx43 overexpression reversed the disruption of tight junction properties and downregulated vascular endothelial growth factor expression. Consistently, Cx43 overexpression increased transendothelial electrical resistance values in HRVEC layers.

Conclusions: O-GlcNAcylation negatively regulated Cx43 expression, contributing to the disruption of the blood retinal barrier. However, O-GlcNAcylation inhibition and Cx43 overexpression could reverse the tight junction disruption. Therefore, O-GlcNAcylation inhibition is a potential target for avoiding tight junction disruption through the Cx43 pathway in DR.

Currently, diabetic retinopathy (DR) is the leading cause of blindness in the working age population worldwide. Clinically, DR is commonly caused by diabetic macular edema (DME), which is characterized by exudative fluid accumulation within retinal layers due to the disruption of the blood retinal barrier (BRB).^1,2 In recent years, structural development and functional maintenance of BRB have been well studied.^3–6 A previous study revealed that vascular endothelial growth factor (VEGF) is the most important factor associated with the loss of tight junction properties; therefore, anti-VEGF drugs are the mainstay of treatment for DME.⁷ However, different patients with DME respond poorly to anti-VEGF drugs.⁸ In addition, long-term anti-VEGF treatment might cause unintended microvascular occlusion and neural structural changes.^9,10 Therefore, there is an urgent need to develop new molecular mediators targeting barrier disruption and alternative pathways involved in upstream mechanisms to prevent BRB disruption. 

We previously focused on the pathological changes associated with O-linked N-acetylglucosamine (O-GlcNAc) modification in DR;^11,12 this modification is a form of GlcNAcylation targeting serine and threonine residues that is regulated by a pair of enzymes: O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA).¹³ OGT catalyzes the attachment of O-GlcNAc to proteins, whereas O-GlcNAcase facilitates the removal of sugar.¹⁴ Studies have revealed over 15,000 types of proteins modified by O-GlcNAcylation, which participate in various biological functions and are associated with various pathologies, such as diabetes, diabetic complications, cancer, neurodegenerative abnormalities, and other vascular diseases.^15,16 Enhanced O-GlcNAcylation under high-glucose conditions was found to be one of the causes of vascular damage,^11,17 which was associated with VEGF secretion, loss of tight junction properties, and apoptosis of retinal pericytes and endothelial cells.^18,19 

We recently revealed that Cx43 is closely related to BRB function and that Cx43 expression is substantially upregulated in astrocytes but significantly downregulated in retinal capillary endothelium.²⁰ As O-GlcNAcylation inhibition can downregulate Cx43 expression in astrocytes,²⁰ this study aimed to investigate the effect of O-GlcNAcylation on Cx43 expression in the endothelium. 

Cx43 is abundantly expressed in glial, vascular endothelial, and retinal pigment epithelial cells,^21,22 where it regulates the exchange of intercellular ions, metabolites, and neuroactive substances. Numerous studies have shown that Cx43 expression is involved in the development of diabetic complications, such as tight junction disruption and glial activation.^23,24 Therefore, the current study aimed to investigate the effects of O-GlcNAcylation on Cx43 expression in the endothelium and its subsequent effects on tight junction properties under high-glucose conditions. 

Methods

Reagents

CTD110.6 (#9875) was purchased from Cell Signaling Technology (Danvers, MA, USA); primary Cx43 (ab11370/ab235585), phospho-Cx43 (ab30559), occludin (ab167161), zonula occluden-1 (ZO-1; ab221547), and GAPDH (ab8245) antibodies were purchased from Abcam (Cambridge, MA, USA); and Evans Blue (E104208-25g) was purchased from Aladdin R&D Center (Los Angeles, CA, USA). The following reagents were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, USA): MEM Medium (32571036), Opti-MEMMedium (31985088), IP Lysis Buffer (87787), and Classic Magnetic IP/Co-IP Kit (88804; Thermo Pierce). Further, Thiamet G (SML0244) and alloxan (A7413) were obtained from Sigma-Aldrich (St. Louis, MO, USA). The drug solutions were diluted as previously described.²⁰ Thiamet G, an inhibitor of OGA, enhances O-GlcNAcylation, whereas alloxan, an inhibitor of OGT, reduces O-GlcNAcylation. Therefore, in this study, O-GlcNAcylation was enhanced via Thiamet G (2.5 µM) treatment and decreased via alloxan (2.5 mM) treatment. 

Cell Culture

Human retinal vascular endothelial cells (HRVECs) were acquired from Cell System Corporation (CSC, Kirkland, WA, USA) and confirmed using Von Willebrand Factor (vWF) (Supplementary Material). HRVECs were cultured in endothelial cell medium (ECM; Gibco, Thermo Fisher Scientific Inc., Waltham, MA, USA) supplemented with 1% endothelial cell growth supplement, 5% fetal bovine serum, 100 U/mL penicillin, and 0.1 mg/mL streptomycin at 37°C in a 5% CO₂ incubator. HRVECs were cultured under normal-glucose (5 mM) or high-glucose (25 mM) conditions. Lentivirus infection was carried out to overexpress or decrease Cx43 expression. Then, it was treated with Thiamet G (2.5 µM) or alloxan (2.5 mM) for another 48 hours as the experimental design. 

Experimental Animals

Sixty adult male Brown Norway rats (age = 6–8 weeks, mean weight = 305 g) were purchased from Charles River Laboratories (Shanghai, China). The rats were divided into four groups equally: control group, diabetic rat group, diabetic rat treated with Thiamet G (O-GlcNAcylation augmentation) group, and diabetic rat treated with alloxan (O-GlcNAcylation inhibition) group. A rat model of diabetes was established via the intraperitoneal injection of streptozotocin (STZ; 45 mg/kg) dissolved in 10 mM sodium citrate.²⁰ The night before injection, all animals were fasted to maximize the efficacy of STZ. Daily blood glucose level was measured 48 hours after STZ injection using OneTouch Glucose Monitoring System. A consistent blood glucose level of >16.7 mmol/L indicated the successful establishment of the rat model of diabetes. Drinking water must be provided. 

Intravitreal Injections

One month after model establishment, the rats were subjected to general anesthesia via the intramuscular injection of sodium pentobarbital (2–5 mg/kg), whereas 0.5% proparacaine was used for topical anesthesia. A 5-µL Thiamet G or alloxan solution was injected into one eye of the rats with diabetes (selected randomly) through the pars plana, as reported previously.²⁰ Upon Thiamet G or alloxan treatment, the blood glucose level was assessed once weekly, and the mean blood glucose level was compared between the groups (27.44 ± 3.395, 26.533 ± 3.152, and 27.08 ± 3.489 mmol, P = 0.75). One month after treatment, retinal tissues were dissected from enucleated eyes and stored in a freezer at −80°C or in 4% paraformaldehyde solution. 

Cx43 Overexpression and Knockdown Using Lentivirus Constructs

Cx43 overexpression and knockdown in HRVECs were induced using lentivirus constructs (Structure of the lentivirus). To achieve Cx43 overexpression, PCDH-CX43-copGFP-Puro was constructed by inserting Cx43 sequences into pCDH-CMV-MCS-EF1-copGFP-T2A-Puro lentiviral vector. Further, to achieve Cx43 knockdown, HBLV-CX43 shRNA1-ZsGreen-Puro was constructed by inserting Cx43 shRNA sequences (top strand, GATCCGCAGTCTGCCTTTCGTTGTCTCGAGACAACGAAAGGCAGACTGCTTTTTTG; bottom strand, AATTCAAAAAAGCAGTCTGCCTTTCGTTGTCTCGAGACAACGAAAGGCAGACTGCG) into pHBLV-U6-MCS-CMV-ZsGreen-PGK-Puro lentiviral vector. 

The integrative lentiviruses were obtained by cotransfecting lentivirus constructs and lentivirus packaging plasmids into human embryonic kidney 293T cells using LipoFiter version 3.0 transfection reagent (HB-TRLF-50; HanBio); subsequently, viral supernatants were collected. HRVECs were infected with viral supernatants; the integrative lentiviruses were randomly integrated in each cell, and selected using 2.5 µg/mL puromycin (P8833; Sigma-Aldrich), followed by culture in DMEM containing 1 µg/mL puromycin. Cx43 overexpression and knockdown in HRVECs were verified via Western blotting using anti-Cx43 antibody (1:1000). 

Western Blotting

Cx43, occludin, ZO-1, and O-GlcNAc levels were determined via Western blotting. Total proteins in HRVECs and isolated retinas were lysed using RIPA buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 1% Triton X-100, 0.5% SDS, 1% sodium deoxycholate, and protease inhibitor cocktail). Protein concentrations were measured using enhanced BCA protein assay kit (Thermo Scientific Inc., Grand Island, NY, USA). Equal amounts of proteins (30 µg) from cell extracts or retinal samples were subjected to 10% SDS–PAGE and then transferred onto a nitrocellulose (NC) membrane. After blocking with 5% nonfat milk for 1 hour, the membrane was incubated at 4°C overnight with rabbit anti-Cx43 (1:1000), anti-CTD110.6 (1:1000), anti-occludin (1:1000), anti-ZO-1 (1:1000), or anti-GAPDH (1:5000) antibody. After washing 3 times with 0.1% Tween-20 in TBS, the blots were incubated for 1 hour with HRP-conjugated secondary antibody (1:5000; Cell Signaling). The blots were exposed to a chemiluminescent protein detection system using enhanced chemiluminescence detection kit (Thermo Scientific Inc.). Densitometry measurements were performed using Image J. The experiments were repeated at least three times. 

Immunoprecipitation

Cx43 immunoprecipitation was performed using anti-Cx43 antibody (15 µL) and rabbit IgG antibody (3 µL, control) conjugated to protein A/G magnetic beads. The antibody–magnetic bead complex (50 µL) was incubated with protein lysate (equivalent to 300 µg of total protein) for 1 hour at room temperature. After washing 3 times with IP lysis buffer, 20 µL of 3 × SDS–PAGE loading buffer was added, and the mixture was boiled and centrifuged at 2000 × g for 1 minute to collect conjugated proteins. The proteins were run on 10% Tris–glycine gradient gels and then analyzed via immunoblotting. 

Immunofluorescence Staining

To obtain eye tissues, the enucleated eyes were incubated overnight in 4% paraformaldehyde, followed by embedding in optical coherence tomography (OCT; Electron Microscopy Science) and slicing into 6 to 10-µm-thick sections using a cryostat microtome (CM1860; Leica Biosystems, Danvers, MA, USA). HRVECs were seeded into 24-well culture plates containing ECM and subjected to treatment for 48 hours with normal glucose (5 mM), high glucose (25 mM), Thiamet G (2.5 µM), and alloxan (2.5 mM). The cells were washed with cold phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde for 20 minutes. The sections and HRVECs were rinsed 3 times with PBS for 5 minutes and treated with Triton X-100 (0.3%) for 10 minutes. They were then blocked with serum (5% goat serum and 5% bovine serum albumin) at room temperature for 1 hour, followed by overnight treatment at 4°C with primary antibodies against Cx43 (1:400), occludin (1:400), or ZO-1 (1:400). After washing 3 times with PBS, the sections and cells were incubated for 1 hour at room temperature with secondary antibodies conjugated with AlexaFluor 488 and 555 (1:200). They were then observed under a confocal microscope (DMi8, Confocal system TCSSPE; Leica) by a double-blind technician. 

Evans Blue Quantitation

Rats with diabetes and age-matched controls were anesthetized via the intramuscular injection of sodium pentobarbital (2–5 mg/kg). The left femoral veins were exposed under an operating microscope and cannulated with polyethylene tubing (PE10 or 50; BD Intramedic, Franklin Lakes, NJ, USA) filled with heparinized normal saline (200 U/mL NS). Evans blue (45 mg/kg) was injected into the body through the femoral vein and circulated for 1 hour. Retinas were dissected from enucleated eyes and analyzed by a double-blind technician under a fluorescence microscope at a magnification of 20 times. Images of five random fields were obtained and analyzed using ImageJ. The vascular morphology and leakage areas were compared among different groups. Areas with density values between 230 and 255 were considered to reflect the leakage of Evans blue dye. Leakage was calculated as the total fluorescence of Evans blue dye minus the vessel area (mm²) per retina. 

Transendothelial Electrical Resistance Measurements

HRVECs were cultured on Transwell permeable membranes (VWR International Inc., West Chester, PA, USA) until confluency. The transendothelial electrical resistance (TEER) values were determined using a voltmeter resistance measurement chamber (World Precision Instruments, Sarasota, FL, USA). At 24 hours after treatment with drugs, TEER measurements were repeated. Resistance across the HRVEC layer was calculated as follows: total resistance minus the background resistance/per unit area (Ω/cm²). Resistance was compared among the groups. 

Study Approval

This study was approved by the Ethics Committee of Shanghai Tenth People's Hospital. All experimental methods were conducted in accordance with the guidelines of the Declaration of Helsinki as well as ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Statistical Analysis

All experiments were repeated at least three times and measured at the same time. Statistical analysis was performed using Statistical Package for the Social Sciences for Windows (version 27.0; IBM Inc., Armonk, NY, USA). For all parameters, the mean and standard error of the mean were calculated. The effects of O-GlcNAcylation on the expression of Cx43 and other proteins were analyzed via two-tailed one-way analysis of variance followed by post hoc comparison. The P values of < 0.05 were considered to indicate statistical significance. 

Results

Effects of O-GlcNAcylation on Tight Junction Properties in Diabetic Retinas

We investigated the effects of O-GlcNAcylation on tight junction properties in vitro. We confirmed its effect on BRB in vivo by determining occludin, ZO-1, and VEGF expression using Western blotting and assessing vascular leakage using the Evans blue leakage assay. 

As expected, compared with the control group, retinas from rats with diabetes showed decreased expression of occludin and ZO-1 and increased expression of VEGF (Fig. 1A, P < 0.001). The O-GlcNAcylation augmented (treated with Thiamet G) group, which was subjected to intravitreal injection with Thiamet G, showed a significant decrease in occludin and ZO-1 expression and increase in VEGF expression (see Fig. 1A, P < 0.001). Conversely, the O-GlcNAcylation inhibition (treated with alloxan) group showed significantly increased expression of occludin and ZO-1 and decreased expression of VEGF (see Fig. 1A, P < 0.01). These changes were consistent with our previous findings in HRVECs.¹¹ 

Figure 1.

 

Effects of O-GlcNAcylation on Cx43 expression and BRB properties in diabetic retinas. O-GlcNAc, Cx43 (ab235585), ZO-1, occludin, and VEGF-A levels were determined via Western blotting (N = 3 independent retinas per group). (A) In groups with diabetes, Thiamet G treatment led to O-GlcNAcylation augmentation, resulting in decreased expression of occludin and ZO-1 and increased expression of VEGF. In contrast, alloxan treatment led to O-GlcNAcylation inhibition, resulting in increased expression of occludin and ZO-1 and decreased expression of VEGF. (B) Vascular leakage was confirmed using the Evans blue leakage assay (N = 6 independent retinas per group); the arrows in yellow indicate the sites of vascular leakage. Compared with the diabetic group, vascular leakage increased in the O-GlcNAcylation augmented (treated with Thiamet G) group, whereas it improved in the O-GlcNAcylation inhibition (treated with alloxan) group (magnification, 20 ×). *: versus control, P < 0.001; **: versus DM, P < 0.01; #: versus DM, P < 0.001.

Consistent with Western blotting results, the Evans blue leakage assay revealed vascular leakage in diabetic retinas (Fig. 1B, diabetes mellitus [DM], P < 0.01). The vascular leakage increased in the O-GlcNAcylation augmented group (see Fig. 1B, DM + Thiamet G, P < 0.001), whereas it improved in the O-GlcNAcylation inhibition group (alloxan treatment; see Fig. 1B, DM + alloxan, P < 0.01). Our results revealed that O-GlcNAcylation enhancement under hyperglycemia was the main cause of BRB disruption in diabetic retinas. 

Cx43 Expression in HRVECs Under High-Glucose Conditions

To explore the mechanisms underlying the effects of O-GlcNAcylation on the endothelial barrier, Cx43 expression in HRVECs was investigated. The results showed that compared with the normal-glucose group, a significant decrease in Cx43 expression was observed in the high-glucose group (Fig. 2A, P < 0.001). Within the high-glucose groups, Cx43 expression did not differ significantly (see Fig. 2A, P > 0.05). 

Figure 2.

 

Cx43 expression in HRVECs under high-glucose conditions. (A) The first lane shows the immunoblot of Cx43 (ab11370) in the normal-glucose group (5 mM); the other 3 lanes indicate Cx43 blots under high-glucose conditions (25 mM; 24, 48, and 72 hours; N = 3 independent experiments per group). Compared with the normal-glucose group, Cx43 expression was significantly decreased in high-glucose groups. (B) Cx43 (red) and DAPI staining (blue) in HRVECs under normal-glucose (5 mM) (a, b, c) and high-glucose conditions (25 mM) for 24 hours (magnification, 63 ×).

Immunofluorescence staining detected Cx43 particles as densely distributed puncta primarily localized on the cellular membranes; they were sparsely distributed on the plasma membrane (Figs. 2Ba–c). Compared with the normal-glucose group, Cx43 staining was markedly decreased in the high-glucose group (Figs. 2Bd–f), which was consistent with the western blotting results. Our previous study also reported similar results in diabetic retinas.²⁰ 

Effect of Cx43 on Occludin and ZO-1 Expression

To investigate the effect of abnormal Cx43 on the expression of occludin and ZO-1, HRVECs with stable overexpression or knockdown of Cx43 were obtained via lentivirus infection. Western blotting revealed that Cx43 expression was significantly upregulated in the Cx43 overexpression group and downregulated in the Cx43 knockdown group (Fig. 3A, P < 0.01). Figure 3B shows Cx43 staining in HRVECs subjected to high-glucose conditions and lentivirus infection. Cx43 staining was markedly increased in the Cx43 overexpression group, whereas it was evidently decreased in the Cx43 knockdown group. 

Figure 3.

 

Cx43 expression was determined via Western blotting and immunofluorescence staining. (A) Western blotting analysis of Cx43 (ab11370) in control, Cx43-Adv (overexpression), and Cx43-shRNA-Adv (knockdown) groups (N = 3 independent experiments per group). (B) Representative Cx43 staining in HRVECs. Compared with the control group, Cx43 expression was significantly upregulated in the overexpression group and downregulated in the knockdown group (P < 0.01; magnification, 63 ×). HG + NC-Adv: High glucose combined with lentivirus control; HG + Cx43-Adv: High glucose combined with Cx43 overexpression; and HG + Cx43-shRNA-Adv: High glucose combined with Cx43 knockdown. *: versus NC-Adv, P < 0.01.

As shown in Figure 4, the expression of occludin and ZO-1 decreased in the high-glucose group, which was further decreased via Cx43 knockdown (see Fig. 4, P < 0.01). However, Cx43 overexpression significantly increased occludin and ZO-1 expression (see Fig. 4, P < 0.01). The results showed that Cx43 positively regulated occludin and ZO-1 expression, indicating that Cx43 is a potential target for avoiding tight junction barrier disruption. 

Figure 4.

 

Effect of Cx43 on the expression of occludin and ZO-1. (A) Immunoblotting of Cx43 (ab235585), occludin, ZO-1, and GAPDH in the normal-glucose, high-glucose (lentivirus control), high-glucose with Cx43 overexpression, and high-glucose with Cx43 knockdown groups. (B, C, D) Relative expression of Cx43, occludin, and ZO-1 was quantified using ImageJ. Compared with the lentivirus control group, Cx43 overexpression significantly increased the expression of occludin and ZO-1 (P < 0.01), whereas Cx43 knockdown significantly decreased them. *: versus NG, P < 0.05; **: versus NG, P < 0.01; #: versus HG + NC-Adv, P < 0.01.

Crosstalk of O-GlcNAcylation With Cx43

We previously revealed that GlcNAcylation contributed to enhanced Cx43 expression in astrocytes.²⁰ It remains unclear whether enhanced GlcNAcylation causes decreased Cx43 expression in HRVECs, further disrupting tight junction properties. Under high-glucose conditions, O-GlcNAcylation was enhanced with reduced Cx43 expression (Fig. 5B, P < 0.05). Our results revealed a significant decrease in Cx43 expression in the GlcNAcylation augmented group (treated with Thiamet G) under both normal- and high-glucose conditions (see Fig. 5A, P < 0.001, Fig. 5B, P < 0.01). In contrast, the GlcNAcylation inhibition (treated with alloxan) group showed a significant increase in Cx43 expression (see Fig. 5A, P < 0.05, Fig. 5B, P < 0.01). 

Figure 5.

 

Effects of O-GlcNAcylation on Cx43 expression in HRVECs. (A) Immunoblotting of CTD110.6 and Cx43 (ab11370) in HRVECs under normal-glucose conditions (N = 3 independent experiments per group). (B) Immunoblotting of CTD110.6 and Cx43 in HRVECs under high-glucose conditions. Compared with the normal-glucose group, the GlcNAcylation augmented (treated with Thiamet G) group significantly decreased Cx43 expression, which was elevated in the GlcNAcylation inhibition (treated with alloxan) group. #: versus NG, P < 0.001; *: versus NG, P < 0.05; **: versus HG, P < 0.01.

The results showed that enhanced O-GlcNAcylation is essential for decreased Cx43 expression; thus, O-GlcNAcylation inhibition could reverse the reduction in Cx43 expression under high-glucose conditions. 

To explore the mechanism underlying the effect of O-GlcNAcylation on Cx43 expression, we used immunoprecipitation to determine whether Cx43 was modified by O-GlcNAcylation. Anti-CTD110.6 antibody was used to assess O-GlcNAcylation in precipitated Cx43 protein. The results showed that O-GlcNAcylation negatively regulated the expression of both total and phosphorylated Cx43 (Fig. 6). We concluded that Cx43 was competitively modified by O-GlcNAcylation and phosphorylation simultaneously. 

Figure 6.

 

Immunoprecipitation of Cx43 in HRVECs. (A) Immunoblotting of Cx43 (ab235585), p-Cx43, and O-GlcNAc (CTD110.6) in different groups. (B) O-GlcNAcylation was detected in precipitated Cx43 protein, with reduced expression of Cx43 and pCx43 in the O-GlcNAcylation augmented (treated with Thiamet G) group and increased expression in the O-GlcNAcylation inhibition (treated with alloxan) group.

In addition to the negative effects of O-GlcNAcylation on Cx43 expression, our results revealed that compared with the high-glucose group, O-GlcNAcylation was significantly inhibited in the Cx43 overexpression group (Fig. 7, P < 0.01). Cx43 overexpression combined with alloxan treatment downregulated O-GlcNAcylation to normal levels. 

Figure 7.

 

Effect of Cx43 on O-GlcNAcylation. (A) Immunoblotting of CTD110.6 and Cx43 in HRVECs (N = 3 independent experiments per group). (B) Compared with the lentivirus control group, O-GlcNAcylation was significantly decreased in the Cx43 overexpression group. (C) Cx43 overexpression combined with Thiamet G and alloxan treatments. Cx43 overexpression combined with alloxan treatment downregulated O-GlcNAcylation to normal levels. *: versus NG, P < 0.01; #: versus HG + NC-Adv, P < 0.05; **: versus HG + Cx43-Adv + Thiamet G, P < 0.001.

O-GlcNAcylation Inhibition Protects Against the Disruption of the Tight Junction Barrier Through the Cx43 Pathway

The effect of O-GlcNAcylation on tight junction protein expression was investigated in diabetic retinas (see Fig. 1). The results revealed increased ZO-1 and occludin expression and reduced VEGF expression in the Cx43 overexpression group (Figs. 8A–E, P < 0.01). Further, compared with high-glucose treatment, Cx43 overexpression combined with O-GlcNAcylation inhibition (treated with alloxan) significantly reversed the reduction in occludin and ZO-1 expression and downregulated VEGF expression (see Figs. 8A–E, P < 0.01). In contrast, enhanced O-GlcNAcylation due to Thiamet G treatment decreased the upregulated expression of occludin and ZO-1 induced by Cx43 overexpression (see Fig. 8C, Fig. 7D, P < 0.05); this is because enhanced O-GlcNAcylation can reverse the Cx43 overexpression. 

Figure 8.

 

Effects of Cx43 on the expression of ZO-1, occludin, and VEGF-A in HRVECs under high-glucose conditions. (A) Immunoblotting of Cx43 (ab235585), ZO-1, occludin, and VEGF-A. (B, C, D, E) Expression of ZO-1, occludin, and VEGF-A was compared in the Cx43 overexpression combined with Thiamet G or alloxan treatment group. **: versus NG, P < 0.001; #: versus HG + NC-Adv, P < 0.01; *: versus HG + Cx43-Adv, P < 0.05; *#: versus HG + Cx43-Adv, P < 0.01; ##: versus HG + Cx43-Adv, P < 0.01.

Next, TEER measurements were performed to determine the paracellular permeability within an HRVEC monolayer. Baseline TEER measurements revealed no significant differences among the groups. High-glucose treatment decreased the TEER values by 14.69% (Fig. 9A, P < 0.01; 658.6 ± 4.67 Ω/cm² vs. 561.9 ± 6.51 Ω/cm²), which was further decreased by 20.52% with Cx43 knockdown (see Fig. 9A, P < 0.05; 658.6 ± 4.67 Ω/cm² vs. 523.5 ± 9.33 Ω/cm²). Further, Cx43 overexpression significantly improved the reduction in TEER values (see Fig. 9A, P < 0.01; 561.9 ± 6.51 Ω/cm² vs. 620.2 ± 13.84 Ω/cm²). 

Figure 9.

 

TEER measurements. (A) Baseline TEER measurements were comparable among the groups. The effects of Cx43 on TEER measurements were significant. (B) Although enhanced O-GlcNAcylation contributed to decreased resistance across the HRVEC monolayer, Cx43 overexpression combined with O-GlcNAcylation inhibition could reverse the TEER reduction (N = 3 independent experiments per group). *: versus NG, P < 0.01; **: versus HG + NC-Adv, P < 0.01; #: versus HG + Cx43-Adv, P < 0.05; ##: versus HG + Cx43-Adv, P < 0.001.

Consistent with the changes in occludin and ZO-1 expression in diabetic retina (see Fig. 1) and high-glucose-treated HRVECs,¹¹ TEER values were decreased by 15.78% in the high-glucose group (see Fig. 9B, P < 0.01; 640.1 ± 7.31 Ω/cm² vs. 539.5 ± 11.21 Ω/cm²). In addition, enhanced O-GlcNAcylation (treated with Thiamet G) could reverse the increase in TEER values induced by Cx43 overexpression (see Fig. 9B, P < 0.01; 608.9 ± 6.84 Ω/cm² vs. 515.52 ± 7.67 Ω/cm²). However, Cx43 overexpression combined with O-GlcNAcylation inhibition (treated with alloxan) could completely restore the reduction in TEER values (see Fig. 9B, P < 0.001; 539.5 ± 11.21 Ω/cm² vs. 640.64 ± 7.04 Ω/cm²). These results demonstrated that enhanced O-GlcNAcylation under high-glucose conditions was an important cause of BRB disruption, and GlcNAcylation inhibition could protect against endothelium barrier disruption through the Cx43 pathway. 

Discussion

Our results revealed that O-GlcNAcylation negatively regulates Cx43 expression under both normal- and high-glucose conditions, indicating that enhanced O-GlcNAcylation in DR causes reduction in Cx43 expression within the endothelium, which subsequently leads to endothelial barrier disruption. These results also demonstrated that Cx43 overexpression could prominently reverse the endothelial barrier disruption. To explore the mechanisms potentially underlying this process, we investigated the interactive effects of O-GlcNAcylation and Cx43 expression. 

In addition to the negative effects of O-GlcNAcylation on Cx43 expression, O-GlcNAcylated proteins were detected in precipitated Cx43, suggesting that Cx43 is a target for O-GlcNAcylation. Thus, O-GlcNAcylation could regulate Cx43 expression directly or indirectly (such as via the state of Cx43 phosphorylation). 

Several phosphorylated sites of Cx43, such as Ser325/328/330/365 and Tyr247/265, have been reported.^25,26 Moreover, the phosphorylated state of Cx43 is reportedly involved in the assembly, degeneration, and activity of the gap junction, resulting in changes in protein expression and molecular functions.²⁷ As both phosphorylation and O-GlcNAcylation target serine and tyrosine residues, the specific sites that can be simultaneously modified and effects of O-GlcNAcylation on phosphorylated Cx43 should be investigated in the future. 

In general, O-GlcNAcylation is considered a process complementary to phosphorylation, but the relationship between them is more complicated than simple competitive inhibition or commodification. In this study, Cx43 phosphorylation decreased in the enhanced O-GlcNAcylation group and increased in the O-GlcNAcylation inhibition group. This finding suggests that Cx43 expressed in the endothelium is competitively modified by both phosphorylation and O-GlcNAcylation. 

In addition, O-GlcNAcylation and phosphorylation could dynamically modify the enzymes that regulate their modification levels, that is, O-GlcNAcylation of phosphate cycling enzymes and phosphorylation of O-GlcNAc cycling enzymes. A previous study demonstrated that OGT is phosphorylated at both tyrosine and serine residues, whereas OGA is only phosphorylated at serine residues.²⁸ CAMKI is an important kinase involved in regulating the phosphorylation of transcription factors, which may be O-GlcNAcylated at multiple residues.²⁹ 

Cx43 can be phosphorylated by several kinases, such as protein kinase A, protein kinase C, protein kinase CK1, and mitogen-activated protein kinase,²⁶ and these kinases could be modified by O-GlcNAcylation.³⁰ Thus, enhanced O-GlcNAcylation can dramatically reduce phosphorylation by altering the levels of upstream regulatory kinases or Cx43 phosphorylation itself. 

In addition to the negative effects of O-GlcNAcylation on Cx43 phosphorylation, we detected decreased O-GlcNAcylation in the Cx43 overexpression group. As increased Cx43 phosphorylation was detected in the Cx43 overexpression group, it is possible that enhanced Cx43 phosphorylation might explain the decrease in O-GlcNAcylation. However, the specific sites that undergo commodification or competitive inhibition of phosphorylation and O-GlcNAcylation remain unclear. 

Previous studies have investigated the effects of Cx43 on tight junctions,^23,31 revealing that Cx43 plays a protective role against the disruption of endothelial tight junction. In contrast, other studies have reported opposite results that Cx43 upregulation might contribute to retinal vessel leakage.^32,33 To clarify the impact of Cx43 on tight junctions, lentivirus infection was performed to achieve stable overexpression and knockdown of Cx43. Our results demonstrated that Cx43 overexpression significantly upregulated occludin and ZO-1 expression, whereas Cx43 knockdown promoted the loss of tight junction properties. 

Structurally, Cx43 is colocalized with occludin and ZO-1 on the membranes of vascular endothelial cells.^34,35 In addition to physical associations, molecular interactions were detected, such as propagation of ions and signaling molecules by Cx43 to tight junction complexes.³⁶ Therefore, enhanced Cx43 expression and its phosphorylated state may promote tight junction integrity.³⁷ 

Consistent with the effect of Cx43 on tight junction expression, Cx43 overexpression significantly reversed the reduction in paracellular permeability within the HRVEC monolayer and decreased the VEGF expression induced by high-glucose treatment. Moreover, notably, Cx43 overexpression combined with O-GlcNAcylation inhibition could completely reverse the reduction in paracellular permeability to normal levels. These results indicate that O-GlcNAcylation inhibition can be used to upregulate the expression of ZO-1 and occludin through the Cx43 pathway, significantly reducing the vessel leakage induced by hyperglycemia. 

Taken together, our results demonstrate that enhanced O-GlcNAcylation plays an important role in reducing Cx43 expression in DR, which subsequently causes the loss of ZO-1 and occludin. Conversely, Cx43 overexpression positively regulated tight junction properties and downregulated O-GlcNAcylation. This is the first study to provide in vivo evidence regarding the effect of O-GlcNAcylation on BRB disruption through the Cx43 pathway. Based on the protective effects of O-GlcNAcylation inhibition on Cx43 expression and tight junctions, an upstream pathway can be considered a new target for managing DR. 

Acknowledgments

Supported by National Natural Science Foundation of China (No. 81700840), Scientific Research Project of Shanghai Municipal Health Commission (No. 202040247) and Bethune Langmu Young Ophthalmology Research Fund of China (Grant No. BCF-KH-YK-20221123-05). The authors appreciate the assistant of language editing from Duane Decker. 

Disclosure: G. Liu, None; L. Feng, None; X. Liu, None; P. Gao, None; F. Wang, None