Ablation of 4E-BP1/2 Prevents Hyperglycemia-Mediated Induction of VEGF Expression in the Rodent Retina and in Müller Cells in Culture

Male Sprague-Dawley rats (120–180 g) were housed in The Pennsylvania State University College of Medicine Animal Facility in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines. Type 1 diabetes was induced in rats by a single intraperitoneal injection of streptozotocin (STZ), as previously described (21). Rats exhibiting blood glucose levels above 250 mg/dl, measured 24 h after drug administration, were considered diabetic. Average blood glucose values for the animal models used herein are shown in supplementary Fig. 1 in the online appendix available at http://diabetes.diabetesjournals.org/cgi/content/full/db10-0148/DC1.

Male C57BL/6J Ins2^Akita heterozygous mice were bred in The Pennsylvania State University College of Medicine Juvenile Diabetes Research Foundation Diabetic Retinopathy Center Animal Core, in accordance with IACUC guidelines. At 4.5 weeks of age, mice were genotyped and tested for a diabetic phenotype, that is, a blood glucose level exceeding 250 mg/dl. Wild-type littermates were used as controls (22).

Wild-type and Eif4ebp1;Eif4ebp2 double knockout mice were described previously (23). At 4 weeks of age, type 1 diabetes was induced according to the AMDCC “Low-Dose Streptozotocin Induction Protocol (Mouse)” (24). Mice with blood-glucose levels exceeding 250 mg/dl each week were considered diabetic, and those that did not meet this threshold were excluded from the study.

TR-MUL [kindly provided by Dr. Ken-ichi Hosoya (Toyama Medical and Pharmaceutical University)] and Moorfields/Institute of Ophthalmology-Müller 1 [kindly provided by Dr. G. Astrid Limb (University College of London)] retinal Müller cells, and wild-type and Eif4ebp1;Eif4ebp2 double knockout MEFs (23) were maintained in Dulbecco's modified Eagle's medium containing 5.6 mmol/l glucose and 2% heat-inactivated FBS and grown to subconfluence before seeding. Cells were maintained at either 37°C (MIO-M1 and MEFs) or 33°C (TR-MUL), in a humidified, 5% CO₂ atmosphere. TR-MUL and MIO-M1 cells (500,000) were seeded into 6-well dishes, and MEF cells (20,000) were seeded into 12-well dishes and grown to ∼75% confluence. Culture medium was aspirated, cells were washed with sterile PBS, and experimental medium was added. For studies of the effects of hyperglycemic conditions, cells were either replenished with medium containing 5.6 mmol/l glucose and 25 mmol/l mannitol (hereafter referred to as control medium) or 25 mmol/l glucose and 10% heat-inactivated FBS (hereafter referred to as high-glucose medium) for 10 h (unless otherwise indicated).

At the indicated time points, the retina was isolated, flash-frozen in liquid nitrogen, and later sonicated in 500 μl of extraction buffer (25) supplemented with inhibitors [microcystin, sodium vanadate, benzamidine, and protease inhibitor cocktail (Sigma-Aldrich)]. The homogenate was centrifuged at 12,000 rpm for 10 min at 4°C, and the supernatant was collected, a fraction of which was added to 1× Laemmli buffer and boiled for 5 min. For Western blot analysis of these samples, 60 μg of protein was loaded per lane. In studies using cells in culture, medium was collected for VEGF analysis, and cells were washed in sterile PBS, then scraped in 200 μl of either 1× Laemmli buffer, immunoprecipitation homogenization buffer (20 mmol/l HEPES, 2 mmol/l EGTA, 50 mmol/l NaF, 100 mmol/l KCl, 0.2 mmol/l EDTA, 50 mmol/l β-glycerophosphate, pH 7.4, 1 mmol/l benzamidine, 0.5 mmol/l sodium vanadate, 1 μmol/l microcystin, and 15 μl/ml protease inhibitor cocktail), or Lambda phosphatase buffer (26), supplemented with inhibitors (except microcystin and sodium vanadate). For Western blot analysis of these samples, 30 μg of protein was loaded per lane. Protein was separated by SDS-PAGE on 15%, 4–15%, or 4–20% Criterion gels (Bio-Rad Laboratories) and transferred to a polyvinylidene fluoride membrane. The membrane was blocked in 5% milk in Tris-buffered saline Tween-20, washed, and then incubated overnight at 4°C with primary antibody. The source of polyclonal antibodies, and the dilutions used in Western blot analysis, are presented in supplementary Table 1, available in an online appendix. Monoclonal antibodies to the α-subunit of eIF2, eIF4E, and eIF4G were prepared in our laboratory as described previously (27,–29). The membranes were incubated with horseradish peroxidase-conjugated secondary antibody, washed, incubated for 1 h at a 1:10,000 dilution, and then developed using enhanced chemiluminescence. Quantitation was performed using a GeneGnome HR imaging system with GeneTools software (SynGene).

For analysis of VEGF production in cells in culture, media was collected from control and treated cells and subjected to VEGF ELISA (Qiagen) following the manufacturer's protocol. VEGF concentration was normalized to total cellular protein content.

The retinas and harvested cells were sonicated in 500 and 200 μl, respectively, of Lambda phosphatase buffer (26), supplemented with benzamidine and protease inhibitor cocktail, and the homogenate was combined with 4 μl (1,600 units) of Lambda phosphatase and incubated for 1 h at 37°C. The sample was centrifuged at 1,000 g for 3 min, and the supernatant was boiled for 3 min, combined with 2× Laemmli buffer, boiled again for 5 min, and subjected to Western blot analysis.

BioMag goat-anti-mouse IgG magnetic beads (BioClone Inc.) (1.5 ml/sample) were washed thrice in low salt buffer (20 mmol/l Tris-HCl, pH 7.4, 5 mmol/l EDTA, 150 mmol/l NaCl, 0.5% Triton X-100, 0.1% β-mercaptoethanol) and then resuspended in 1/5 of the original volume in low salt buffer. Beads were combined with 175 μl PBS, 12.5 μl Triton X-100, and 4 μg monoclonal anti-eIF4E antibody (per sample) and incubated on an orbital rocker overnight at 4°C. On the day of the study, the retinas were homogenized using a Dounce homogenizer in 500 μl of eIF4E homogenization buffer (20 mmol/l HEPES, 2 mmol/l EGTA, 50 mmol/l sodium fluoride, 100 mmol/l potassium chloride, 0.2 mmol/l EDTA, 50 mmol/l β-glycerophosphate, pH 7.4) plus inhibitors. Alternatively, cells in culture were harvested in immunoprecipitation buffer containing 2.5% Triton X-100 and 0.25% deoxycholate. Retinal or cell homogenate was added to 1 ml or 200 μl, respectively, of BioMag bead-bound eIF4E antibody and incubated for 1 h at 4°C. The beads were washed, resuspended in 1× Laemmli buffer, and boiled for 5 min. After the beads were removed, 20 μl of sample was loaded onto a 4–15% Criterion gel and subjected to Western blot analysis.

RNA was isolated from retinas after using TriZol (Invitrogen) as per the manufacturer's protocol. Total RNA, at least 500 ng, was reverse transcribed using an ABI High Capacity cDNA Reverse Transcription Kit according to the manufacturer's protocol. For quantitation of mRNA, qRT-PCR was performed using Taqman RT-PCR assay. The sequence of the Taqman (R) Gene Expression Assay primers used herein are presented in supplementary Table 2, available in an online appendix.

For cell culture studies, qRT-PCR was performed using QuantiTech SYBR Green PCR Kit (Qiagen) and the following QuantiTech Primer Assay kits (Qiagen) following the manufacturer's protocol: β-actin (Rn_Actb_1_SG QuantiTect Primer Assay NM_031144), VEGFA (Rn_RGD:619991_1_SG QuantiTect Primer Assay NM_001110335), and 4E-BP1 (Rn_Eif4ebp1_1_SG QuantiTect Primer Assay NM_053857).

Results from the retina of diabetic animals were compared by ANOVA or Student t test to those from the retina of their nondiabetic counterparts using Prism 4 (GraphPad Software, Inc.). Where appropriate, data were log-transformed for normality. Data from untreated control cells were compared either by ANOVA or Student t test, whichever appropriate, with that from their treated counterparts using Prism 4. The data are presented as mean ± SEM; P < 0.05 was considered statistically significant.

Alterations in VEGF expression were assessed by Western blot analysis of retinal homogenates prepared from control and STZ-induced type 1 diabetic rats at various times after drug administration. Whereas VEGF expression was increased significantly in the retina of diabetic compared with control rats at 2, 4, and 6 weeks (Fig. 1,A), the relative abundance of VEGF mRNA remained constant (Fig. 1,B). In addition to constant expression of VEGF mRNA, expression of the transcription factor, HIF-1α, did not differ among the groups (Fig. 1 C), suggesting that the mechanism for increased VEGF expression was posttranscriptional.

Because the expression of the VEGF mRNA did not change over the time course used above, mechanisms involved in the regulation of mRNA translation were explored that might explain increased expression of the protein. Such mechanisms include changes in phosphorylation of eIF4E, eIF4G, and the α-subunit of eIF2. Therefore, the phosphorylation state of each of these proteins was assessed at various times after STZ treatment. No significant difference in phosphorylation of eIF2α on Ser51 (Fig. 2,A), eIF4G on Ser1108 (Fig. 2,B), or eIF4E on Ser209 (Fig. 2 C) was observed in the retina of diabetic compared with control rats at 1, 2, 4, or 6 weeks of diabetes.

Another mechanism through which the selection of mRNAs for translation is regulated involves the sequestration of eIF4E by 4E-BP1 (30). Therefore, 4E-BP1 association with eIF4E was assessed, and, as shown in Fig. 3,A, a significant increase in 4E-BP1 associated with eIF4E was detected in the retina of diabetic compared with control rats, with no change in the amount of 4E-BP2 bound to eIF4E (Fig. 3,A). The increased association was attributed to both a reduction in the relative phosphorylation of 4E-BP1 on T37/46, a key regulatory phosphorylation site (Fig. 3,B), and to increased expression of 4E-BP1 (Fig. 3,C). The time course of changes in 4E-BP1 expression paralleled that observed for VEGF (Fig. 1,A). 4E-BP1 mRNA expression was unchanged (Fig. 3 D).

The Ins2^Akita mutation is a single amino acid substitution (C96Y) in the insulin 2 gene that disrupts disulfide bond formation between the A and B chains, resulting in misfolding of proinsulin. The accumulation of misfolded protein leads to ER stress and, ultimately, β-cell apoptosis and development of hypoinsulinemia and hyperglycemia (22). These mice have been used in recent studies to investigate diabetic complications in light of the similar disease progression and characteristics to that of the established STZ-induced model of type 1 diabetes, including increased retinal vascular permeability and elevated apoptosis (22). Thus, to further evaluate the correlation between increased VEGF and 4E-BP1 expression observed in the STZ-induced model of type 1 diabetes in the rat, retinas were harvested from Ins2^Akita mice at various time points and analyzed for VEGF and 4E-BP1 protein and mRNA expression. The time course for increased VEGF expression in the Ins2^Akita mice was delayed compared with that of the STZ-induced diabetic rat, but the elevated expression of VEGF and 4E-BP1 protein (Fig. 4 A and B, respectively) were similar in both models of diabetes.

If an increase in 4E-BP1 availability is necessary for the increased expression of VEGF, then the absence of 4E-BP1 would be expected to prevent the diabetes-induced change in VEGF expression. To test this hypothesis, retinas from control and STZ-induced diabetic, wild-type, and 4E-BP1 and 4E-BP2 double knockout (Eif4ebp1;Eif4ebp2) mice were analyzed. While retinal 4E-BP1 protein was detected in Eif4ebp1;Eif4ebp2 wild-type mice, the protein was undetectable in the retinal extracts from Eif4ebp1;Eif4ebp2 double knockout mice, confirming lack of 4E-BP1 expression in these mice (Fig. 5,A). After 5 weeks of STZ-induced diabetes, VEGF expression was significantly higher in the retina of the diabetic compared with the control Eif4ebp1;Eif4ebp2 wild-type mice. In contrast, diabetes had no effect on VEGF expression in the retina of Eif4ebp1;Eif4ebp2 double knockout mice (Fig. 5 B), despite severe hyperglycemia (supplementary Fig. 1C). This result demonstrates that an increase in 4E-BP1 availability is necessary for the diabetes-induced increase in VEGF expression.

To assess whether hyperglycemic conditions contribute to the increase in 4E-BP1 and VEGF expression in the retina of diabetic animals, TR-MUL Müller cells were incubated in either control or high-glucose medium for various times. 4E-BP1 expression was significantly increased 4 h after exposure to high glucose and continued to be elevated through 18 h of treatment (Fig. 6,A). As the maximal increase in 4E-BP1 occurred at ∼10 h of treatment, this time point was selected for further studies. An increase in 4E-BP1 expression was also observed in the human Müller cell line, MIO-M1 (Fig. 6,B), in response to incubation in high-glucose medium, demonstrating that the effect of hyperglycemic conditions was not specific for TR-MUL cells. Although 4E-BP1 protein content was significantly increased at all time points, the expression of 4E-BP1 mRNA was similar among groups (Fig. 7,A). Additionally, hyperglycemic conditions caused a decrease in the relative phosphorylation of 4E-BP1 on T37/46 (Fig. 7,B). Increased expression and decreased phosphorylation resulted in a significant increase in the amount of 4E-BP1 bound to eIF4E, as measured in eIF4E immunoprecipitates (Fig. 7 C). Thus, similar to the effect observed in the retina of diabetic rodents, 4E-BP1 expression was increased in response to hyperglycemic conditions in Müller cells in culture. The increased VEGF expression in Müller cells is consistent with the immunolocalization of VEGF in human eyes of patients with diabetes (31).

To determine whether hyperglycemic conditions resulted in the induction of VEGF, medium was collected for measurement of VEGF secretion by ELISA. After 10 h of high-glucose treatment, the VEGF concentration in the cell culture medium was significantly elevated for VEGFA content (Fig. 8,A). However, there was no change in the expression of VEGFA mRNA in the cell lysates (Fig. 8,B), suggesting a posttranscriptional mechanism for the stimulated production of VEGF. VEGF secretion was also stimulated in MIO-M1 cells incubated in the high-glucose medium compared with the control medium (Fig. 8 C).

To confirm the dependence of VEGF induction on 4E-BP1 expression, wild-type and Eif4ebp1;Eif4ebp2 double knockout MEFs were exposed to either control or high-glucose medium for 10 h. Expression of 4E-BP1 and 4E-BP2 was undetectable in the double knockout cells (data not shown). As observed in Müller cells, 4E-BP1 expression was increased in wild-type MEFs during hyperglycemic conditions (data not shown), and this was associated with a 2.5-fold stimulation of VEGF secretion into the medium (Fig. 8,D). Secretion of VEGF from double knockout MEFs, on the other hand, was not stimulated under hyperglycemic conditions, confirming the requirement for increased 4E-BP1 expression (Fig. 8 D).

In the present study, VEGF expression was increased, without a change in VEGF mRNA expression, in the retina of three animal models of type 1 diabetes: STZ-induced diabetic rats and mice and Ins2^Akita mice. An increase in VEGF expression also occurred in the absence of changes in VEGF mRNA expression in two different Müller cell lines as well as in MEFs exposed to hyperglycemic conditions. These findings suggest that VEGF expression is regulated in a manner that is independent of changes in expression of the message. Although other groups have reported an increase in VEGF mRNA expression in the retina of diabetic animals (32,33), under the conditions used in the present study, no difference was detected.

The results presented herein indicate that 4E-BP1/2 is necessary for the increased expression of VEGF in the retina during diabetes, as the absence of the binding proteins prevents the change. 4E-BP1/2 is also required for the increase in VEGF expression in MEFs maintained in culture under hyperglycemic conditions, as VEGF expression did not increase in 4E-BP1/2 double knockout MEFs exposed to hyperglycemic conditions. In both the retina and in cells in culture, the increase in 4E-BP1 expression led to enhanced binding to eIF4E. Under conditions promoting eIF4E sequestration into the eIF4E·4E-BP1 complex, there is a switch to cap-independent translation, making mRNAs that are normally inefficiently translated, such as the one encoding VEGF, more competitive for translation (33,,,–37). While further studies are required to identify the exact mechanism of 4E-BP1/2-induced regulation of VEGF mRNA translation, it is likely that increased VEGF internal ribosome entry sites utilization accounts for the enhanced expression of the protein.

Increased 4E-BP1 protein content in the absence of changes in 4E-BP1 mRNA expression suggests regulation at the posttranscriptional level in both the retina of diabetic animals and in cells in culture exposed to hyperglycemic conditions. However, unlike the VEGF mRNA, the mRNAs encoding mouse and human 4E-BP1 have relatively short 5′-UTRs that lack significant secondary structure in prediction models. Moreover, the 5′-UTR of the 4E-BP1 mRNA lacks other regulatory sequences that might be involved in translational regulation, such as open reading frames or terminal oligopyrimidine tracts. Although other translational regulatory mechanisms such as microRNAs cannot be discounted, changes in the rate of degradation may contribute to alterations in 4E-BP1 expression. In this regard, evidence for proteasome-dependent degradation of 4E-BP1 has been shown by accumulation of the protein after proteasome inhibition (34) and by ubiquitin immunoprecipitation (35). In the present study, increased binding of 4E-BP1 to eIF4E in the retina of diabetic rats is consistent with possible decreased ubiquitination of 4E-BP1, because previous studies (35) have shown inefficient binding of ubiquitinated 4E-BP1 to eIF4E. Whether or not 4E-BP1 degradation is decreased in the retina of diabetic rodents will be explored in future studies.

A role for 4E-BP1 in regulating VEGF mRNA translation has also been reported in immortalized proximal tubular epithelial (MCT) cells (36). Treatment of MCT cells with angiotensin II promotes increased phosphorylation of 4E-BP1 and recruitment of VEGF mRNA into polysomes. Moreover, incubation in medium containing 30 mmol/l instead of 5 mmol/l glucose leads to increased synthesis of angiotensinogen, a precursor of angiotensin II, as well as VEGF (37). Together these data suggest that elevations in glucose concentration may promote VEGF mRNA translation by increasing expression of angiotensin II, and subsequently 4E-BP1 phosphorylation. Angiotensin II has also been implicated in changes in VEGF expression in the retina of diabetic rodents (38,–40). However, in the animal models of diabetes used in those studies, VEGF mRNA expression was increased, a finding not observed in the present study. Further, in the present study, the 4E-BP1 content increased causing a relative decrease in the amount of phosphorylated 4E-BP1, suggesting a distinct mechanism from that observed in MCT cells. Whether or not changes in angiotensin II expression might be involved in the increase in VEGF expression observed in the models of diabetes used herein is unknown.

In the present study, VEGF expression was increased in the retina of diabetic compared with control rats as early as 2 weeks of STZ-induced type 1 diabetes. The elevated expression was sustained for at least 6 weeks but returned to basal levels by 12 weeks. Many studies have documented that the vascular pathology associated with diabetic retinopathy, such as vascular permeability (41), appearance of acellular capillaries (42), and endothelial apoptosis (43), occur after prolonged periods, for example, 3–8 months, of diabetes in rodents, while other studies have observed vascular changes by 1 week (44). One study (33) reported blood–retinal barrier breakdown in response to a near twofold increase in VEGF expression after only 8 days of diabetes. The increase in permeability was entirely prevented by treatment with a chimeric VEGF receptor (33). Therefore, increased VEGF expression has been observed in multiple animal models, but the time of increase varies with the model and method of diabetes induction as was observed within this study comparing VEGF expression in rat and mouse models of type 1 diabetes.

In the present study, three rodent models of type 1 diabetes yielded similar results in increased expression of VEGF and 4E-BP1. Together, the evidence from these studies strongly suggests that 4E-BP1 is a key modulator of translational control of VEGF expression in the retina during diabetes. Further details about the regulation of VEGF expression may lead to novel therapies to control abnormal vascular permeability and angiogenesis in diabetic retinopathy.

T.W.G. is the Jack and Nancy Turner Professor. This project was funded by a grant from the Juvenile Diabetes Research Foundation International (4-2002-455) and a grant from The Pennsylvania Department of Health using Tobacco Settlement Funds. The Department specifically disclaims responsibility for any analyses, interpretations, or conclusions. No potential conflicts of interest relevant to this article were reported.

T.L.S. researched data and wrote the first draft of the manuscript. D.A.A., N.S., and T.W.G. contributed to discussion and reviewed the manuscript. S.R.K. designed experiments, contributed to discussion, and wrote the manuscript. L.S.J. designed experiments, contributed to discussion, and reviewed/edited the manuscript.

The authors thank The Pennsylvania State College of Medicine JDRF Animal Core for generating the animal models of type 1 diabetes used in the present study. The authors also thank Dr. Patrice E. Fort for his collaborative effort, Dr. Ken-ichi Hosoya (Toyama Medical and Pharmaceutical University) for providing the TR-MUL cells, and Dr. G. Astrid Limb (University College of London) for providing the MIO-M1 cells.