Molecular Vision 2014; 20:1109-1121
Received 12 December 2013 | Accepted 29 July 2014 | Published 31 July 2014
The first two authors contributed equally to this work.
1Department of Ophthalmology, the Sixth People’s Hospital, Shanghai Jiaotong University, Shanghai, China; 2Shanghai Key Laboratory of Diabetes Mellitus, Shanghai, China
Correspondence to: Qiang Wu, Department of Ophthalmology, The Sixth People’s Hospital, Shanghai Jiaotong University, 600 Yishan Road, Shanghai 200233, China. Phone: +86-18930177422; FAX: +86-021-64701361; email: Qiang.firstname.lastname@example.org
Purpose: Retinal vascular dysfunction caused by vascular endothelial growth factor (VEGF) is the major pathological change that occurs in diabetic retinopathy (DR). It has recently been demonstrated that G protein-coupled receptor 91 (GPR91) plays a major role in both vasculature development and retinal angiogenesis. In this study, we examined the signaling pathways involved in GPR91-dependent VEGF release during the early stages of retinal vascular change in streptozotocin-induced diabetes.
Methods: Diabetic rats were assigned randomly to receive intravitreal injections of shRNA lentiviral particles targeting GPR91 (LV.shGPR91) or control particles (LV.shScrambled). Accumulation of succinate was assessed by gas chromatography-mass spectrometry (GC-MS). At 14 weeks, the ultrastructure and function of the retinal vessels of diabetic retinas with or without shRNA treatment were assessed using hematoxylin and eosin (HE) staining, transmission electron microscopy (TEM), and Evans blue dye permeability. The expression of GPR91, extracellular signal-regulated kinases 1 and 2 (ERK1/2) and cyclooxygenase-2 (COX-2) were measured using immunofluorescence and western blotting. COX-2 and VEGF mRNA were determined by quantitative RT–PCR. Prostaglandin E2 (PGE2) and VEGF secretion were detected using an enzyme-linked immunosorbent assay.
Results: Succinate exhibited abundant accumulation in diabetic rat retinas. The retinal telangiectatic vessels, basement membrane thickness, and Evans blue dye permeability were attenuated by treatment with GPR91 shRNA. In diabetic rats, knockdown of GPR91 inhibited the activities of ERK1/2 and COX-2 as well as the expression of PGE2 and VEGF. Meanwhile, COX-2, PGE2, and VEGF expression was inhibited by ERK1/2 inhibitor U0126 and COX-2 inhibitor NS-398.
Conclusions: Our data suggest that hyperglycemia causes succinate accumulation and GPR91 activity in retinal ganglion cells, which mediate VEGF-induced retinal vascular change via the ERK1/2/COX-2/PGE2 pathway. This study highlights the signaling pathway as a potential target for intervention in DR.
Diabetes mellitus is characterized by hyperglycemia and a consequent functional failure of various target organs, including the eyes. Diabetic retinopathy (DR) is one of the fastest growing causes of blindness and visual impairment in the working-age population. The pathogenesis and development of DR are highly complex due to the involvement of multiple interlinked mechanisms. Various metabolic pathways triggered by hyperglycemia are involved, such as the polyol pathway, hexosamine pathway, and diacylglycerol (DAG)-protein kinase C (PKC) pathway . In parallel, classically, oxidative stress , hemodynamic changes , and the production of free radicals, cytokines , or advanced glycosylation end-products  have also been considered to be crucial for the development of DR. However, the pathogenesis of DR has not been elucidated completely, despite much investigation.
Vascular endothelial growth factor (VEGF) has been recognized as the prominent mediator in the process of DR, and overexpression of VEGF is believed to correlate with the vascular hyperpermeability and neovascularization in diabetic subjects . Because vascular supply is tightly coupled to tissue metabolic rate, it is conceivable that energy source metabolic intermediates also affect the progression of DR . Succinate, a Krebs cycle intermediate normally found in mitochondria, is released into the extracellular medium if the local tissue energy demand and supply are imbalanced . Local accumulation of succinate is recognized as an indicator of diabetic organ damage [9,10]. Most recently, it has been suggested that high levels of succinate were detected in patients with proliferative diabetic retinopathy (PDR) . G protein-coupled receptor 91 (GPR91), a known specific receptor for succinate, is expressed in the kidney, spleen, placenta, liver, and retina . It has been demonstrated that GPR91 plays a critical role in the pathogenesis of diabetic neuropathy, hypertension, heart stress, and liver damage [9,12-14]. Sapieha et al. has found that GPR91 plays a major role in the settings of both normal retinal development and proliferative ischemic retinopathy . We previously assessed the role of GPR91 in high-glucose-induced VEGF release in vitro . Nevertheless, the influence of GPR91 on retinal vascular dysfunction in DR and the underlying molecular mechanisms remain unknown. Unveiling these precise mechanisms may contribute to clarifying the pathogenesis of DR.
DR is a multifactorial disease in which a variety of signaling pathways and active substances are involved. Cyclooxygenase-2 (COX-2) and COX-2-induced prostaglandin E2 (PGE2) have been confirmed to participate in this process and regulate the expression of VEGF . In addition, extracellular signal-regulated kinases 1 and 2 (ERK1/2), a major subfamily of mitogen-activated protein kinase (MAPK) signaling, is recognized as an important pathway in the transduction of extracellular signals to cellular responses and is involved in various physiologic effects and pathological processes [17,18]. ERK1/2 has been verified to mediate VEGF release in oncoma and hematologic diseases [19,20]. The overwhelming majority of research verifies that the ERK1/2 signaling pathway plays a key role in the occurrence and development of DR [21,22]. Recently, one study of diabetic nephropathy showed that accumulating succinate under hyperglycemia conditions induced ERK1/2 activation, COX-2, and PGE2 upregulation by binding with activated GPR91 [10,23].
In this study, we constructed a lentiviral expression vector containing a GPR91 shRNA and used it to investigate the role of GPR91 in VEGF release and to dissect the potential molecular mechanisms involved in DR. We examined the hypothesis that the ERK1/2/COX-2/PGE2 signaling pathway mediates GPR91-dependent VEGF release during the early stages of retinal vascular dysfunction in a streptozotocin (STZ)-induced diabetic model.
Male Sprague-Dawley (SD) rats (2 months old, 200–250 g) were purchased from the SIPPR/BK Lab Animal Ltd (Shanghai, China). The rats were housed in a barrier facility with free access to normal food and tap water. They were maintained under conditions of standard lighting (a 12 h:12 h light-dark cycle), temperature (23–25 °C), and humidity (50%–60%). Diabetes was induced using STZ based on a previously published protocol . The rats were injected with a single intraperitoneal dose of 60 mg/kg STZ in 100 mM citrate buffer (pH 4.5). Weight- and age-matched non-diabetic control rats received injections of an equal volume of citrate buffer. Following STZ injection (48 h post-injection), a blood sample was taken from the tail vein of each rat, and the blood glucose level was measured using an automatic analyzer (Optium Xceed, Abbott Diabetes Care, Bedford, MA). The maintenance of a diabetic state was confirmed by weekly tail vein-blood glucose measurements. Animals with plasma glucose concentrations >16.7 mmol/l were deemed diabetic and were included in the study. Treatment of the animals conformed to the Guide for Care and Use of Laboratory Animals published by the National Institutes of Health (Guide for the Care and Use of Laboratory Animals, 1996), and the protocols were approved by the Animal Ethics Committee of the Sixth People’s Hospital, Shanghai Jiaotong University.
The recombinant GPR91 shRNA (AACCCTAAATACAGTCTCATT) and the scrambled shRNA were designed and packaged by Genechem Co., Ltd (Shanghai, China), as described previously . DNA oligos containing the target sequence were chemically synthesized, annealed, and inserted into the lentivirus expression vector pGCSIL-GFP by double digestion with AgeI and EcoRI and ligation with T4 DNA ligase. The ligate was transformed into competent Escherichia coli DH5α cells. Restriction enzyme analysis and DNA sequencing were used to identify the desired transformants. The lentivirus carrying GPR91 shRNA was produced by plasmid cotransfection of 293T cells. The viral supernatant was collected after transfection for 48 h, passed through 0.45-mm filters, concentrated and titered to 109 TU/ml (transfection unit). The rats were randomized into six groups and treated as follows: (1) control group, non-diabetic rats; (2) STZ group, diabetic rats not treated with the lentiviral particle; (3) LV.shScrambled group, diabetic rats that received an intravitreal injection of the scrambled shRNA lentiviral particles (1 μl, 1×108 TU/ml) 2 weeks after the induction of diabetes; (4) LV.shGPR91 group, diabetic rats that received an intravitreal injection of the GPR91 shRNA lentiviral particles (1 μl, 1×108 TU/ml) 2 weeks after the induction of diabetes; (5) U0126 group, diabetic rats that received an intravitreal injection of 0.1 mM U0126 (ERK1/2 inhibitor, Calbiochem, Gibbstown, NJ) before the diabetic model was induced; and (6) NS-398 group, diabetic rats that received an intravitreal injection of 0.5 mM NS-398 (COX-2 inhibitor, Cayman, Ann Arbor, MI) before the diabetic model was induced.
Retinal samples were freshly harvested and homogenized ultrasonically for 4 min. Next, 3 μl 2,2,3,3-2H4-succinic acid (CDN Isotopes, Pointe-Claire, Canada) was added to each tissue sample as an internal standard in water, and an additional 200 μl water was added according to the reported protocol . Briefly, 150 μl methanol and 50 μl chloroform were added and the mixture was centrifuged at 3600 ×g for 10 min. The clear supernatants were collected, and nitrogen and 80 μl methoxamine (15 mg/ml) were added to the dried samples. The samples were incubated overnight and Bis-trimethylsilyl-trifluoroacetamide (BSTFA) was added to the reaction samples. They were concentrated for 1.5 h and prepared for GC/MS analysis (Comprehensive Two-dimensional Gas Chromatography/Time-of-flight Mass Spectrometer, Pegasus 4D GC×GC-TOFMS, Leco, St. Joseph, MI).
The fixed eyes were processed routinely and embedded in paraffin wax. Paraffin-embedded retinal sections (5 μm) were dewaxed and dehydrated. The sections were taken at approximately 100 µm intervals and spanned the entire retina. For each eye, three sections were randomly chosen and evaluated. An average of 15 to 18 sections for each group was assessed. The sections stained using hematoxylin and eosin (HE) were analyzed and photographed by the same examiner, who was blinded to the source of the tissue, using a light microscope equipped with a camera (Leica, Wetzlar, Germany). The retinal areas in each image that were used to calculate the number of blood vessel profiles (BVPs) in the inner retina were chosen according to a previous study . Briefly, we divided 180×110 mm area into 16-U areas for each image. An automatic stage produced an unbiased counting frame to advance across the entire retina.
Retinal sections were blocked in 5% BSA for 1 h at room temperature, followed by incubation with retinal ganglion cell (RGC) specific marker γ-synuclein (1:50, Santa Cruz, Dallas, TX), endothelial-cell specific marker isolectin B4(1:50, Santa Cruz, CA), GPR91 (1:50, Novus Biologicals, Littleton, CO), p-ERK1/2 (1:50, Cell Signaling Technology, Boston, MA), and COX-2 (1:50, Santa Cruz, CA) primary antibodies overnight at 4 °C. The retinal sections then were incubated in CY3-conjugated anti-rabbit and FITC-conjugated anti-mouse secondary antibodies (1:200, Invitrogen, Carlsbad, CA) for 1 h at room temperature. Fluorescence image capture and analysis were performed with a fluorescence microscope (Leica, Wetzlar, Germany).
Rat retinas were fixed in 2.5% glutaraldehyde and post-fixed in 1% osmium tetroxide after dissection out of four random areas from the central and peripheral retina. The specimens were then embedded in Spurr resin. The tissue blocks were orientated using 1 μm sections stained with toluidine blue, and the ultrathin sections were contrasted using uranyl acetate and lead citrate before examination using transmission electron microscopy (TEM) (CM-120, Philips, Eindhoven, Netherlands). Computer-assisted morphometric measurements were used by AxioVision 4.8 software.
Retinal vascular leakage was quantitated using Evans blue dye according to the standard protocol . Briefly, the right iliac vein and iliac artery were cannulated using polyethylene tubing, and Evans blue was injected through the iliac vein at a dosage of 45 mg/kg. Blood (100 μl) was withdrawn from the iliac artery in 15-min intervals. After the dye had circulated for 120 min, the animal was perfused through the left ventricle using citrate buffer (0.05 M, pH 3.5, 37 °C) for 2 min at 66 ml/min. Immediately after perfusion, both eyes were enucleated and the leakage of Evans blue dye was extracted from the retinas and was determined using the formula: (retinal Evans blue in micrograms/retina dry weight in grams) / (time-averaged plasma Evans blue in micrograms/plasma volume in microliters × circulation time in hours) .
After the rats were injected with a single intraperitoneal in dose of 50 mg/kg 2% pentobarbital sodium solution to euthanasia, each retina was dissected and frozen immediately in liquid nitrogen. The total RNA was extracted from neural retinal samples using Trizol reagent (Invitrogen). The primer sequences used were as follows: rat VEGF (forward: 5′-AAA GCC AGC ACA TAG GAG AG-3′; reverse: 5′-AGG ATT TAA ACC GGG ATT TC-3′), COX-2 (forward: 5′-TAC AAC AAC TCC ATC CTC CTT G-3′; reverse: 5′-TTC ATC TCT CTG CTC TGG TCA A-3′) and rat β-actin (forward: 5′-CAC CCG CGA GTA CAA CCT TC-3′; reverse: 5′-CCC ATA CCC ACC ATC ACA CC-3′). Quantitative real-time PCR was then performed using the SYBR Green qPCR Super Mixture (Takara, Tokyo, Japan) and an ABI Prism 7500 Sequence Detection System. All reactions were performed in triplicate. The data were analyzed using the 2-△△CT method.
For the western blot analysis, the neural retinas were rapidly dissected from the euthanatized rats and lysed in RIPA buffer (Beyotime, Shanghai, China) containing a protease inhibitor (Beyotime, Shanghai, China) and phosphatase inhibitor (Roche, Mannheim, Germany). Aliquots containing 30 μg of protein were separated by SDS-polyacrylamide gel electrophoresis using a 10% gel, and the separated proteins were blotted onto PVDF membranes (Millipore, Billerica, MA) in a wet transfer unit (Bio-Rad, Hercules, CA). After blocking with 5% non-fat dry milk at room temperature for 1 h, the membranes were incubated overnight at 4°C with the following primary antibodies: GPR91 (1:1000), p-ERK1/2 (1:3000) , t-ERK1/2 (1:3000) , COX-2 (1:200), anti-VEGF (1:200, Abcam, Cambridge, MA) andβ-actin (1:1000, Abcam). After being washed with TBS-Tween 20, the membranes were incubated with the appropriate HRP-conjugated secondary antibodies (1:1000, ProteinTech Group, Chicago, IL) for 1 h at room temperature. The bands were visualized using an enhanced ECL detection kit (Pierce Biothechnology, Rockford, IL). For the ELISA, the vitreous fluid samples were collected from rat eye for enzyme-linked immunosorbent assay analysis using kits from R&D Systems (Minneapolis, MN) following the instructions provided by the manufacturer.
All data are presented as the mean± standard deviation (SD). The data were analyzed using SPSS 16.0 software. The differences between multiple groups were assessed by one-way ANOVA, followed by Student–Newman–Keuls (SNK) comparisons. p<0.05 was considered statistically significant.
The results of gas chromatography-mass spectrometry analysis indicated that the levels of succinate were markedly increased in fresh retinal samples from diabetic rats compared with non-diabetic retinas (p<0.01, Figure 1A). However, the expression of GPR91 did not change significantly between the non-diabetic and diabetic rat retinas (p>0.05, Figure 1B,C). Immunofluorescence showed that GPR91 was predominantly localized to the cell bodies of the ganglion cell layer (GCL) and to a lesser extent to cells of the inner nuclear layer (INL) and retinal pigment epithelium (RPE; Figure 1D). However, the endothelial cells did not express GPR91 (Figure 1E).
We then researched the expression of GPR91 using a shRNA approach in the retina. The retinal GPR91 level was significantly reduced in rats transduced with LV.shGPR91 compared with tissues from those transduced with LV.shScrambled at 4 weeks (p<0.01, Figure 1D,F,G).
In addition, we found that the nonfasting blood glucose was markedly higher in the diabetic rats than in the non-diabetic rats (p<0.01, Figure 1H), and local GPR91 knockdown produced no effect on hyperglycemia in the diabetic rats (p>0.05, Figure 1H). The bodyweights of the diabetic rats were significantly lower than that of the non-diabetic rats (p<0.01, Figure 1I).
Pathological damage to the retinal vessel occurred in the 14 week STZ-induced diabetic rats (Figure 2). Compared with the non-diabetic rats, HE staining revealed that the retina tissue developed telangiectatic vessels in the inner layer of retinas (black arrow in Figure 2A), and the cells of the inner nuclear layer appeared to be disorder. The number of BVPs in the inner retina was increased (Figure 2B). TEM examination revealed that swelling was observed in the mitochondria of the pericytes and endothelial cells, and the mitochondrial membrane was ruptured (white square frame in Figure 2C). It was also demonstrated that the basement membrane thickness (BMT) was significantly greater in the diabetic retinal capillary (p<0.01, black arrow in Figure 2C,D). Furthermore, at the completion of the experiment, an increase in retinal vascular permeability was detected in diabetic rats (p<0.01, Figure 2E). Treatment with GPR91 shRNA attenuated the retinal vascular dysfunction and significantly decreased the BMT and Evans blue dye permeability (p<0.01, Figure 2A-E), whereas LV.shScrambled had no such effect (p>0.05, Figure 2D,E), and the damage to the inner nuclear layer was not attenuated in the GPR91 siRNA group (Figure 2A).
Next, we sought to investigate the role of GPR91 in regulating VEGF secretion and retinal vascular damage in diabetic retinas. The retinal expression of VEGF was increased in diabetic rats in the 4 week experiment compared with non-diabetic rats (mRNA = 1.2:1; protein = 7.8:1), and the change in protein level was dramatic (p<0.01, Figure 2F,G). Furthermore, GPR91 knockdown reduced VEGF mRNA by approximately 25% in the STZ-induced diabetic rats (p<0.05, Figure 2F) and significantly decreased VEGF protein expression by approximately 75% (p<0.01, Figure 2G).
To confirm the presence and importance of the ERK1/2 signaling pathway in GPR91-dependent retinal vascular change in diabetic rats, we evaluated ERK1/2 activation in the retinas of diabetic rats transduced with LV.shScrambled or LV. shGPR91 compared with non-diabetic rats. Western blotting showed upregulation of ERK1/2 phosphorylation in the retinas of diabetic rats at 1 week and 2 weeks after the induction of diabetes and displayed a time-dependent trend (Figure 3A,B). Double immunofluorescence showed that the expression of ERK1/2 phosphorylation was increased in RGCs after STZ injection for 1 week (Figure 3C). However, the increases in p-ERK1/2 expression were significantly blocked by GPR91 shRNA (p<0.01, Figure 3D,E).
We then investigated the COX-2 and PGE2 expression and the relationship among GPR91, ERK1/2, COX-2, and PGE2 in the retinas of STZ-induced diabetic rats. The levels of COX-2 protein were increased during the period of 2 weeks to 6 weeks after the induction of diabetes (Figure 4A,B). The COX-2 expression located in RGCs was also enhanced (Figure 4C). PGE2, measured because the production of PGE2 denotes activity of COX-2, was also markedly increased in the retinas of diabetic rats at 4 weeks (p<0.01, Figure 4G). Furthermore, intravitreal injection of 0.1 mM U0126 or 0.5 mM NS-398 significantly blocked the upregulation of COX-2, PGE2, and VEGF release (p<0.01, Figure 4D-I). These findings indicate that the ERK1/2 pathway is upstream of the COX-2/PGE2 pathway, and the ERK1/2/COX-2/PGE2 pathway is associated with VEGF release.
In the retina, VEGF is expressed in retinal ganglion cells, Müller cells, endothelial cells, astrocytes, and RPE . Wang et al.  investigated the role of Müller cell-derived VEGF in retinal vascular leakage during the process of DR. Recent research found that GPR91 existed primarily in RGCs and that it was involved in angiogenesis by binding accumulated succinate in the oxygen-induced retinopathy model . This result has led to speculation that retinal ganglion cells are also a major source of VEGF and a major cellular target for the treatment of the disease. In this study, and in others, we determined that GPR91 was localized to RGCs, cells of the INL, and RPE [7,29,30]. Gnana-Prakasam et al.  reported that GPR91 was expressed in RPE, but only in the apical membrane. The increased VEGF expression that was induced by succinate was abolished in RGC-ablated retinas , indicating that the succinate-GPR91 receptor may be predominantly expressed in RGCs. Previous research has indicated that intravitreal lentiviral vector administration results in higher transduction efficiency in the inner retina than in the outer retina . Therefore, we used intravitreally administered lentiviral gene transfer technology to explore the role of GPR91 in RGCs in the early stages of DR. Our results showed that GPR91 played an important role in the upregulation of VEGF and retinal vascular dysfunction during the early stages of DR. We considered that increased succinate and activated GPR91 may be the important factors inducing VEGF overexpression in the RGCs.
STZ destroys pancreatic island β cells and is used to induce experimental diabetes in rodents . Adult rats treated with a single dose of STZ exhibit hyperglycemia within 48 h, and these animals are widely used as a model of insulin-dependent diabetes. The induction of diabetes with STZ is associated with hyperglycemia and significant weight loss. In our experiments, no supplemental insulin was administered to prevent weight loss. Our results found that the intravitreal injection of GPR91 shRNA lentiviral particles had no effect on hyperglycemia or weight loss in the diabetic animals, suggesting that the effects of GPR91 shRNA on retinal VEGF expression and vascular dysfunction are most likely mediated by a local mechanism rather than by a systemic mechanism. In this study, we demonstrated retinal telangiectatic vessels, a thickened capillary basement membrane, and vascular leakage as determined by assessing ultrastructural changes and performing Evans blue dye permeability studies in STZ-induced diabetic rats. Our results were similar to the reports of Zhang et al., who investigated VEGF upregulation and retinal vascular dysfunction during the process of DR . Numerous studies have reported that VEGF is a potent factor involved in the induction of retinal permeability [6,34]. Additionally, our previous in vitro study demonstrated that VEGF was involved in the proliferation and migration of endothelial cells . Kaur et al. demonstrated that the inner retinal barrier, which is associated with the tight junctions between the neighboring retinal capillary endothelial cells, was more sensitive to hypoxia and ischemia than the outer retinal barrier . We concluded that this vascular leakage was due to an inner retinal barrier dysfunction, but the data do not exclude the possibility of an outer retinal barrier dysfunction . Meanwhile, we also found that lentiviral-delivered GPR91 shRNA attenuated these dysfunctions in the retinal vasculature significantly, but GPR91 had no effect on the damage of the inner nuclear layer during the development of DR. These results suggested that in the retinal ganglion layer, GPR91 may modulate retinal vascular dysfunction by regulating VEGF expression in the retina. In our study, the mRNA levels of VEGF did not parallel the protein levels of VEGF; therefore, we speculated that there were temporal and spatial differences in gene transcription and translation. First, the time of the VEGF mRNA peak may be earlier than that of the VEGF protein level peak. Second, different regulation mechanisms, acting on both the synthesized mRNA and the synthesized protein, can differentially affect the relative amounts of the two molecules. Finally, there are many well studied molecular processes, such as post-transcription processing and degradation of transcription products, that can affect the relative amounts of mRNA and protein.
The present study demonstrated that succinate accumulated in the retina during the early stages of diabetes, which was consistent with the results in the kidney . Succinate, as an intermediate of the Krebs cycle, is produced by the oxidation of succinyl-CoA by the enzyme succinyl-CoA hydrolase and is further oxidized to fumarate by succinate dehydrogenase . The activity of the Krebs cycle is regulated to match metabolic demands, but pathological situations such as ischemia and hyperglycemia can disrupt the flow of substrates in this cycle, resulting in increased succinate levels [7,10]. Succinate has received intense attention for its role in cellular signaling events via its specific receptor GPR91 . Thus far, reports have demonstrated that succinate-induced activation of GPR91 mutiple biological signals change including Ca2+, PKA-dependent pathway, NO , COX-2 and renin-angiotensin system [10,23] in various tissues. Earlier studies showed that the MAPK signaling pathways are activated by G protein-coupled receptors [36,37]. This study indicates that succinate, via its receptor GPR91 in RGCs, mediates the ERK1/2/COX-2/PGE2 signaling pathway activation and results in the increase of the angiogenic factor VEGF during the pathological process in STZ-induced diabetic retinopathy (Figure 5).
ERK1/2 signaling is associated with many cellular responses, such as proliferation, differentiation, and development . However, this pathway’s inappropriate and continuous activation contributed to oncogenesis , diabetic complications , and angiogenesis . In the retinopathy of prematurity (ROP) model, ERK1/2 was found to be involved in VEGF-induced retinal microvascular endothelial cell proliferation . Our research showed that VEGF release was obviously reduced by using ERK1/2 inhibitor U0126 in the STZ-induced diabetic rats. U0126 is widely thought to act as a potent ERK1/2 antagonist to induce DR pathology . This finding suggested that ERK1/2 played an important role in the process of DR. ERK1/2 kinases belong to a large family of serine/threonine kinases that are triggered by multiple extracellular signals and ERK1/2 kinases transfer the information within the cells. Activation of ERK1/2 is associated with dual phosphorylation of the protein kinase activating loop on threonine and tyrosine residues . The ERK1/2 pathways are tightly regulated by and cross-communicate with the other signaling pathways involved in a wide variety of tissues, such as cAMP, PKC, RTK, and TNF-β and PI3K . In our studies, we demonstrated that the ERK1/2-induced VEGF secretion, at least partially, was regulated by GPR91 in DR.
COX-2 has been intensely studied for many years as an inflammatory mediator, and the upregulation of COX-2 expression has been suggested to lead to inflammation during the development of diabetic retinopathy . Recent research demonstrated that inflammation contributes to local ischemia in the retina and further induces pathological angiogenesis [44,45]. Moreover, data have shown that increased levels of COX-2 are associated with the upregulation of VEGF in various tissues [46-48]. Our results, as described in this report, are similar to those of these studies. Some research showed that the MAPK signaling pathway was involved in the activation of COX-2 [49,50]. Before our studies, the exact signals that mediated the increase in COX-2 in DR were unknown. PGE2, an important COX-2 product, is also a strong inducer of VEGF in cells, including rheumatoid synovial fibroblasts , human monocytic THP-1 cells , and Müller cells . Our findings now have identified the metabolic receptor GPR91 as a strong candidate for the underlying signaling mechanism of COX-2 in the early stages of DR. The results of our investigation also established that ERK1/2 phosphorylation, COX-2, and VEGF are upregulated in parallel, and this upregulation partially relies on the activation of GPR91. Additional studies using the ERK1/2 inhibitor U0126 and COX-2 inhibitor NS-398 abolished COX-2, PGE2 production, and VEGF secretion, providing further confirmation that ERK1/2 and ERK1/2-dependent COX-2 and PGE2 are involved in GPR91-dependent succinate-induced VEGF release. On the basis of the findings from this study, a working signaling model of the succinate and GPR91 signaling pathway occurring in retinal ganglion cells is proposed (Figure 5). Despite promising results, it is not clear that the GPR91 receptor is exclusively responsible for the observed effects because shGPR91 only knocks down the expression of GPR91 instead of knocking it out or silencing it completely. Studies with GPR91 knockout mice and GPR91 overexpression studies would help to clarify this point.
In conclusion, we demonstrated that the retinal vascular dysfunction caused by accumulated succinate and the activity of GPR91 was dependent on ERK1/2 signaling, COX-2 and PGE2 expression and subsequently increased secretion of VEGF in the early stages of DR. We cannot completely rule out other mechanisms of DR because there may be several intricate signaling pathways involved. However, these observations provide a basis for future investigations concerning the potential therapeutic implications of GPR91-dependent signaling-related inhibitors in inhibiting the development of DR.
This work was supported by grants from the Research Fund for the National Natural Science Foundation of China (No. 81070738) and the Key Basic Science Foundation of Science and Technology Commission of Shanghai Manicipality (No. 11JC1407702). No potential conflicts of interest relevant to this article were reported.