VEGF Activation of Protein Kinase C Stimulates Occludin Phosphorylation and Contributes to Endothelial Permeability

Abstract

purpose. VEGF is a potent permeabilizing factor that contributes to the pathogenesis of diabetic retinopathy and brain tumors. VEGF-induced vascular permeability in vivo and in cell culture requires PKC activity, but the mechanism by which PKC regulates barrier properties remains unknown. This study was conducted to examine how VEGF and diabetes alter occludin phosphorylation and endothelial cell permeability.

methods. Chemical PKC inhibitors and activators were used to treat primary retinal endothelial cells in culture. In vitro kinase assays and Western blot analysis of two-dimensional (2D) and one-dimensional (1D) gel retardation assays were used to analyze occludin phosphorylation. Endothelial cell permeability was determined by measuring the flux of 70-kDa dextran through a cell monolayer in culture. Exogenous expression of a dominant negative PKCβII mutant (S217A) was used to assess the PKC dependence of VEGF-induced occludin phosphorylation and endothelial permeability. Occludin phosphorylation was also determined in retinas of streptozotocin-induced diabetic rats.

results. VEGF stimulated the phosphorylation of occludin in primary retinal endothelial cells. Chemical inhibitors of PKC activity blocked the VEGF-induced increase in occludin phosphorylation, as assessed by 2D gel and gel retardation in Western blot analysis, and blocked part of the VEGF-induced monolayer permeability to 70-kDa dextran. Expression of a dominant negative PKCβII mutant blocked VEGF-induced occludin phosphorylation and endothelial permeability. Finally, elevated occludin phosphorylation was observed in the retina of diabetic animals.

conclusions. These results strongly suggest that VEGF-induced endothelial permeability requires PKC-dependent phosphorylation of occludin. Regulation of PKC activity and tight junction protein modifications may have therapeutic implications for treatment of diabetic retinopathy and brain tumors.

Vascular endothelial growth factor (VEGF) contributes to the pathogenesis of a variety of diseases in which the integrity of blood vessels is compromised, including brain tumors and diabetic retinopathy. VEGF is elevated in the retina and vitreous of diabetic patients and has been implicated in increased vascular permeability and breakdown of the blood–retinal barrier, ^{1 2 3 4 5} but the molecular mechanism by which VEGF increases vascular permeability is not well understood. Members of the protein kinase C (PKC) family of serine-threonine kinases are activated by many growth factors, including VEGF. Dysregulation of PKC has been linked to the onset of diabetic retinopathy, as evidenced by its increased activation in the retinas of diabetic rats. ⁶ Modulation of PKC activity in vivo by general PKC inhibitors or by inhibitors specific for the PKCβ isoform significantly reduces fluorescein leakage induced by VEGF. ¹ Moreover, activation of PKC by phorbol esters or hormonal signaling is associated with increased permeability in epithelial and endothelial cell culture model systems. ^{7 8 9 10 11 12 13 14 15 16 17} These studies have demonstrated an important role for PKC isoforms in the pathways by which VEGF stimulates vascular permeability. However, the specific signal transduction mechanism of VEGF-induced endothelial permeability, particularly in retinal endothelium, is uncertain, and it is essential to understand these pathways to develop therapeutic strategies. 

Vascular permeability in the retina is controlled by the blood–retinal barrier, which is formed by well-developed tight junctions between endothelial cells of the inner retina and pigmented epithelial cells in the outer retina. ^{18 19} Tight junctions form a barrier to paracellular vascular permeability and maintain cell polarity by preventing lipids and proteins from diffusing between the apical and basolateral plasma membranes. ^{20 21} Several transmembrane tight junction proteins, including occludin and members of the claudin family, contribute to formation of the paracellular barrier. The expression of occludin correlates directly with the effectiveness of barriers in various tissues. For example, occludin expression is higher in the endothelium of neuronal tissues in the brain and retina than in nonneuronal tissues that have lower barrier properties ^{22 23} and, to a greater degree, in retinal arterioles that have greater barrier properties than in venules that are more permeable. ²⁴ In addition, induction of occludin protein and mRNA content by glucocorticoids is associated with increased barrier properties in bovine retinal endothelial cells (BRECs) in culture. ²⁵ Conversely, occludin antisense oligonucleotides or peptides directed to the extracellular loops of occludin decrease occludin content and increase solute flux. ^{23 26 27} Further, suppression of occludin expression in tight junctions by small interfering (si)RNA increases barrier permeability to mono- and divalent inorganic cations and to monovalent organic cations. ²⁸ Finally, 1 month of diabetes disrupts retinal vascular occludin integrity at endothelial cell borders associated with increased permeability to fluorescence-labeled concanavalin A, ^{24 29} and by 3 months occludin content is reduced in the diabetic rat retina by an apparent increase in permeability to fluorescence-labeled albumin. ³⁰ Together, these studies strongly suggest that the expression levels of occludin and its localization at tight junctions contribute to the regulation of the blood–retinal barrier. 

We and others have shown evidence that the phosphorylation state of occludin contributes to regulation of permeability across the endothelial tight junction in response to chemical and physical stimulation. We have shown that increased occludin phosphorylation precedes increased tissue or monolayer permeability in VEGF-treated rat eyes or VEGF-stimulated BRECs. ³¹ Shear stress, which increases permeability to water (hydraulic conductivity) in bovine aortic endothelial cells, also stimulates occludin phosphorylation. ³² Lysophosphatidic acid and histamine increase the permeability of ECV304 cell monolayers and increase occludin phosphorylation. ³³ Recently, Stamatovic et al. ³⁴ observed that macrophage chemotactic protein (MCP)-1 induces occludin phosphorylation and redistribution while increasing permeability in brain endothelial cells, and these effects could be reversed with dominant negative PKC isoforms. ³⁴ Together these studies suggest a causal link between occludin phosphorylation and the regulation of endothelial barrier properties. 

The role of PKC in the regulation of retinal endothelial cell tight junctions remains incompletely elucidated. In the present study, we examined the role of conventional and novel PKC isoform activation in diffusive permeability and occludin phosphorylation in BRECs. VEGF induced occludin phosphorylation and occludin was shown to possess five to seven phosphorylation sites, as observed on 2D gels. Activation of PKC is necessary and sufficient for VEGF-induced phosphorylation of occludin. The classic PKC isoforms, specifically the β isoforms of PKC, appear to be involved, since a PKCβ-specific inhibitor was effective in blocking VEGF-induced occludin phosphorylation. In addition, exogenous expression of wild-type PKCβII enhanced and a dominant negative PKCβII mutant (S217A) abolished VEGF-induced occludin phosphorylation and endothelial permeability. Finally, increased occludin phosphorylation was observed in the retinas of diabetic rats after 1 and 3 months, further suggesting a causal role in the breakdown of the blood–retinal barrier. Together, these results demonstrate the requirement for classic PKC isoforms in VEGF-induced occludin phosphorylation and in regulated retinal endothelial permeability, but also implicate non-PKC–mediated pathways in vascular permeability. This combination of in vivo and in vitro data provides novel insights into the mechanism of VEGF- and diabetes-induced retinal permeability that have clinical importance for treatment of diabetic retinopathy and brain tumors. 

Materials and Methods

Recombinant human VEGF₁₆₅ was purchased from R&D Systems (Minneapolis, MN). Rhodamine-isothiocyanate labeled 70-kDa dextran was purchased from Sigma-Aldrich (St. Louis, MO). Hydrocortisone sodium phosphate solution was obtained from Merck (Rahway, NJ). LY379196 was obtained from Lilly Research Laboratories (Indianapolis, IN). Phorbol 12-myristate 13-acetate (PMA) and 4α-phorbol 12-myristate 13-acetate (4α-PMA) were purchased from Promega (Madison, WI). Bisindolylmaleimide I HCl (BIM I) was purchased from Calbiochem (San Diego, CA). The ZO-1 rat monoclonal antibody, R40-76, was kindly provided by Bruce Stevenson (Salk Institute, San Diego, CA). Polyclonal rabbit anti-occludin was from Zymed (South San Francisco, CA). Polyclonal rabbit anti-phospho-p42/44 MAPK (Thr202/Tyr204), rabbit anti-p42/44 MAPK antibodies, rabbit anti-phospho-Akt (Ser 473) and rabbit anti-Akt antibodies were obtained from Cell Signaling Technology (Beverly, MA). Anti-rabbit IgG-alkaline phosphatase, anti-mouse IgG-alkaline phosphatase, anti-mouse IgG-horseradish peroxidase, and anti-rabbit IgG-horseradish peroxidase were obtained from GE Healthcare (Piscataway, NJ). Cy2-conjugated donkey anti-rat IgG was obtained from Jackson ImmunoResearch Laboratories (West Grove, PA). 

Cell Culture

Primary BRECs were isolated as described previously. ³⁵ BRECs were cultured on 1 μg/cm² fibronectin (Sigma-Aldrich) in media (MCDB-131; Sigma-Aldrich) supplemented with 10% FCS (Hyclone, Logan, UT), 10 ng/mL epidermal growth factor (Sigma-Aldrich), 0.2 mg/mL endothelial cell growth medium additive (EndoGro; Vec Technologies, Rensselaer, NY), 0.09 mg/mL heparin (Fisher Scientific, Pittsburgh, PA), and 0.01 mL/mL antibiotic-antimycotic (Invitrogen-Life Technologies, Rockville, MD). Cells were used at passages 6 to 8 for experimentation. When BRECs reached confluence, media were changed to complete medium (MCDB-131; Sigma Aldrich) supplemented with 0.2 mg/mL endothelial cell growth medium additive (EndoGro; Vec Technologies), 0.09 mg/mL heparin, 0.01 mL/mL antibiotic, and 138 nM hydrocortisone for 3 days, unless otherwise stated. In experiments with the PKC inhibitors BIM I or LY379196, treatment was applied 30 minutes before the addition of VEGF. 

Measurement of BREC Permeability

BRECs were grown to confluence on fibronectin-coated filters with 0.4-μm pores (Transwell; Corning Costar, Acton, MA). VEGF (1.2 nM) was applied to both the apical and basolateral sides of the membrane for 1 hour before the addition of 10 μM RITC-dextran to the apical chamber of inserts. Aliquots were removed from the basolateral chamber at 1, 2, 3, and 4 hours after the application of dextran to the apical chamber and placed in a 96-well plates (black with clear bottoms, polystyrene; Corning Costar). A sample was taken from the apical chamber at the last time point and placed in the 96-well plate. The amount of fluorescence in the apical chamber remained unaltered over the course of the experiment (data not shown). The fluorescence of the aliquots was quantified with a fluorescence imager (FluorImager 595; Molecular Dynamics, Sunnyvale, CA), and the rate of diffusive flux (Po) was calculated by the following formula ³⁶ :  

\[P_{o}{=}{[}(F_{\mathrm{A}}/{\Delta}t)V_{\mathrm{A}}{]}/(F_{\mathrm{L}}A)\]

where, Po was in centimeters per second; F _A is basolateral fluorescence; F_L is apical fluorescence; Δt is change in time; A is the surface area of the filter (in square centimeters); and V _A is the volume of the basolateral chamber (in cubic centimeters). 

Animals

Male Sprague-Dawley rats weighing 150 to 175 g were purchased from Charles River Laboratories (Wilmington, MA) and were housed in the Penn State College of Medicine animal facility in accordance with the Institutional Animal Care and Use Committee guidelines as well as the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. Rats were maintained in a 12-hour alternating light–dark cycle and received food and water ad libitum. Diabetes was induced by a tail vein injection of streptozotocin (STZ; 65 mg/kg dissolved in 1 mM sodium citrate buffer, pH 4.5) and was confirmed 3 days later by a blood glucose level reading higher than 250 mg/dL (Lifescan; Johnson & Johnson, Milpitas, CA). Rats were anesthetized with ketamine/xylazine (50/0.5 mg/kg) and were killed by decapitation 1 to 3 months after induction of diabetes. Retinas were removed and placed directly in ice-cold homogenization buffer for subsequent experimentation. 

Immunoblot Analysis

Confluent BRECs on 60-mm polystyrene dishes were harvested by washing two times with ice-cold PBS containing phenylmethylsulfonyl fluoride (PMSF; 200 μM) and then scraped in lysis buffer using a cell lifter. Lysis buffer was a Triton-deoxycholate-SDS buffer (100 mM NaCl, 1% Trition X100, 0.5% sodium deoxycholate, 0.2% SDS, 2 mM EDTA, 10 mM HEPES [pH 7.5], 1 mM benzamidine) along with a protease inhibitor cocktail tablet (EDTA free; Complete; Roche, Indianapolis, IN). In addition, buffers were brought to 1 mM NaVO₄, 10 mM NaF, and 10 mM sodium pyrophosphate. Retinas were sonicated in extraction buffer. For both retinas and cells, samples were rocked for 15 minutes at 4°C, and insoluble material was pelleted in a microfuge at 14,000g for 10 minutes. Protein concentrations were determined (D_C Protein Assay kit; Bio-Rad Laboratories, Hercules, CA), and proteins diluted in Laemmli sample buffer were separated on 10% SDS-polyacrylamide gels. All measures of occludin gel shift were performed on 16-cm gels. Proteins were transferred to nitrocellulose (MSI, Westborough, MA), blocked with 5% milk in TBS-T, and immunoblotted with rabbit anti-occludin (1:1500), rabbit anti-phospho-p42/44 MAPK (1:1000), rabbit anti-p42/44 MAPK (1:1000), rabbit anti-phospho-Akt (ser473; 1:1000), and rabbit anti-Akt (1:1000). Primary antibodies were detected by alkaline phosphatase-conjugated anti-rabbit or anti-mouse IgG and enhanced chemifluorescence (ECF; GE Healthcare) or with horseradish peroxidase–conjugated anti-rabbit or anti-mouse IgG and chemiluminescence (LumiGlo; Cell Signaling Technology, Beverly, MA). Bands were then quantified (ImageQuant 1.2 software; Molecular Dynamics, Sunnyvale, CA; or GeneSnap software; SynGene, Cambridge, UK). 

2D Electrophoresis

Confluent BRECs maintained in serum conditions were scraped directly into lysis buffer (7 M Urea, 2 M thiourea, 4% 3-[3-cholamidopropyl]dimethylammonio-2-hydroxy-1-propanesulfonate [CHAPS], 40 mM Tris, and 2% ampholyte carriers [Pharmalytes 3-10]; Sigma-Aldrich), and 2 mM tributylphosphine). Cells were rocked for 1 hour at room temperature and microfuged for 20 minutes. Cell supernatants were subjected to a modified Bradford protein assay, with ovalbumin as the standard (Bio-Rad Laboratories). Protein (750 μg) was applied to 18-cm linear pH-4 to -7 dry strips (Immobiline; GE Healthcare), which were allowed to rehydrate overnight. Proteins were focused for 81,900 V/h at 20°C on an electrophoresis system (Multiphor II; GE Healthcare). IPG (immobilized pH gradient) strips were equilibrated for 30 minutes at room temperature with agitation in equilibration buffer (50 mM Tris-Cl [pH 8.8], 6 M urea, 30% glycerol, 2% SDS, bromphenol blue, and 5 mM tributylphosphine). Proteins were separated in the second dimension on 10% SDS-polyacrylamide gels, transferred to nitrocellulose, and immunoblotted for occludin as just described, except the development agent was different (SuperSignal; Pierce). 

Transfections

BRECs were transiently transfected using the nucleofection technique (Amaxa Biosystems, Gaithersburg, MD). BRECs were grown in culture to 70% confluence. Cells were harvested by trypsinization, and 5 × 10⁵ cells per transfection were resuspended in 2 mL HCAEC (human coronary artery endothelial cell) nucleofection solution (Amaxa Biosystems). After resuspension, 100 μL cell suspension was mixed with 3 μg of the appropriate plasmid DNA, transferred to an electroporation cuvette, and shocked in the electroporation device (setting S-5; Amaxa Biosystems). Cells were diluted with 500 μL MCDB-131 (Sigma Aldrich) complete medium (see cell culture conditions), plated onto 20 fibronectin-coated filters (1.1 cm²; Transwell; Corning Costar) or 20 wells in 6-well plates coated with fibronectin and incubated at 37°C. After 16 hours, the medium was replaced with 500 μL fresh complete medium (MCDB-131; Sigma-Aldrich). Transfection efficiency was determined after 24 hours under a fluorescent microscope by assessing the number of green fluorescent cells in a pEGFP transfected control. Transfection efficiency typically ranged from 50% to 70%. After cells reached confluence, they were switched to medium without serum and treated with 50 ng/mL VEGF for 30 minutes. A transport assay was performed with 70 kDa RITC dextran as described (see Transport Assay Method) or cells were harvested from the 6-well plates for immunoblot analysis. 

In Vitro Kinase Assay

Confluent BRECs were harvested in lysis buffer (100 mM NaCl, 2 mM EDTA, 2 mM EGTA, 50 mM HEPES [pH 7.5], 50 mM NaF, 5 mM β-blycerophosphoate, 1 mM Na₃VO₄, 1 mM benzamidine, 10 mM sodium pyrophosphate, and 1% NP-40] along with a protease inhibitor cocktail tablet (EDTA-free Complete; Roche). Samples were rocked for 15 minutes at 4°C, and insoluble material was pelleted in a microfuge at 14,000g for 10 minutes. Protein concentrations were then determined (D_C Protein Assay kit; Bio-Rad Laboratories), and 20 μg of protein was used for each kinase reaction. Bovine occludin fragments (amino acids 416-523) were cloned into a pET151 vector (Invitrogen) containing a 6Xhis tag. Transformed bacteria were induced with isopropyl-β-d-thiogalactopyranoside (IPTG), pelleted, and lysed with guanidinium lysis buffer followed by sonication. Resin (Probond; Invitrogen) was equilibrated with binding–wash buffer (pH 7.8; 8 M urea, 20 mM NaPO₄, and 500 mM NaCl) and the bacterial lysates were incubated with the resin for 30 minutes at room temperature. The bound occludin fragments were pelleted and washed with binding–wash buffer (pH 6.0), followed by binding–wash buffer (pH 5.3) and finally PBS. 

The kinase assay was performed by adding 25 μL of 2× kinase assay buffer (50 mM HEPES [pH 7.5], 50 mM β-blycerophosphoate, 12.5 mM NaOH, 10 mM MgCl₂, and 1 μM microcystin) to 50 μL of ‘on beads’ occludin fragments (∼0.5 μg). One mM Mg/ATP containing 2 μCi of adenosine 5′-[γ-³²P]triphosphate was added, and the samples were heated to 37°C for either 0 or 5 minutes. Samples were pelleted, washed with PBS, eluted with Laemmli sample buffer, and subjected to SDS-PAGE. The gels were dried, and phosphorylated occludin was assessed with autoradiography. 

Statistical Methods

Student’s t-test, one-way analysis of variance (ANOVA), Kruskal-Wallis nonparametric ANOVA, and posttest analysis were performed (InStat 2.0 software or InStat Prism software; GraphPad, San Diego, CA), with statistical significance set at P < 0.05. 

Results

Activation of Multiple Signaling Pathways in Hydrocortisone-Treated BRECs

Previous studies of primary BRECs in our laboratory have demonstrated that VEGF treatment acutely stimulates occludin phosphorylation and mediates a chronic reduction of occludin content. ^{30 31} In these studies, BRECs were grown in serum-containing medium and developed a barrier to solute flux. Other studies by this laboratory have demonstrated that hydrocortisone treatment of BRECs for 2 or 3 days improved barrier properties and decreased water and 70-kDa dextran flux under 10-cm water pressure and decreased diffusive dextran flux. ²⁵ In addition, hydrocortisone increased occludin protein and mRNA content, reduced occludin phosphorylation, and enhanced occludin and ZO-1 immunostaining at cell borders. ²⁵ Based on the results of these studies, hydrocortisone-treated BRECs were used as a physiologically relevant endothelial cell model. Briefly, BRECs were grown to confluence in serum-containing medium, and then the medium was changed to serum-free medium containing hydrocortisone (138 nM) for 3 days. Diffusive flux (Po) of 70-kDa dextran was assessed in this model system and yielded consistent rates of ∼1 × 10⁻⁶ cm/s. 

Before initiation of studies examining the effect of VEGF on tight junctions, it was important to confirm that the ability of VEGF to stimulate intracellular signaling pathways in cells treated with hydrocortisone remained intact. To this end, the VEGF-mediated activation of Akt and p42/44 MAPK, protein kinases that are phosphorylated in response to VEGF ^{37 38} was examined. Phosphorylation of Akt at Ser₄₇₃ and phosphorylation of p42 and p44 MAPK at positions Thr₂₀₂ and Tyr₂₀₄ are essential for activation of these kinases. ^{39 40} Confluent BRECs treated with hydrocortisone (138 nM) for 3 days were stimulated with VEGF (1.2 nM) for 0, 5, 15, 30, and 60 minutes, and immunoblot analysis was performed on lysate supernatants for Ser₄₇₃ phospho-Akt and total Akt. VEGF mediated a 20-fold increase in Akt phosphorylation at 5 minutes (P < 0.001) and decreased to control levels by 60 minutes (Fig. 1A) . The effect of VEGF on p42/44 MAPK was determined next. Phosphorylation of p44 MAPK was significantly increased: 1.7-fold at 5 minutes (P < 0.01), 2.5-fold at 15 minutes (P < 0.001), 2.2-fold at 30 minutes (P < 0.05), and 2.1-fold at 60 minutes (P < 0.05; Fig. 1B ). Phosphorylated p42 MAPK exhibited a similar trend with VEGF treatment, albeit the increase was not statistically significant. Together, these results demonstrate that VEGF signaling pathways remain intact after 3 days of hydrocortisone treatment. 

VEGF-Induced Occludin Phosphorylation

The effect of PKC on occludin phosphorylation was examined by two-dimensional (2D) gel electrophoresis using VEGF and the PKC inhibitor LY379196. LY379196 is a PKCβ-specific inhibitor ⁴¹ that is an analogue of LY333531, another PKC inhibitor that effectively reduced VEGF-stimulated retinal vascular permeability in rats and is currently in stage-3 clinical trials for diabetic retinopathy. ¹ At low concentrations (30 nM), LY379196 is a specific inhibitor of PKCβ, whereas at higher concentrations (600 nM) it broadly attenuates activation of conventional and some novel isoforms of PKC. Confluent BRECs were maintained in serum-containing medium, pretreated with LY379196 (30 nM) for 30 minutes, and stimulated with VEGF (1.2 nM) for 15 minutes or maintained in control conditions. Cells were harvested as described in the Materials and Methods section, and proteins were focused on IPG strips (pH 4-7) for 81,900 V/h. Proteins were separated in the second dimension on SDS-polyacrylamide gels and immunoblotted for occludin. Figure 2Ais a representative 2D blot from repeated experiments. A minimum of four forms of occludin were resolved under control conditions and VEGF induced the modification of occludin to more acidic forms, consistent with an increase in phosphorylation. LY379196 treatment alone exhibited no effect on the occludin migration pattern. However, LY379196 (30 nM) completely blocked the VEGF-induced shift of occludin to more acidic forms. Together, these results demonstrate that occludin exists in five to seven phosphorylated forms and that activation of PKC is necessary for VEGF-induced occludin phosphorylation. 

To determine whether endothelial cells directly phosphorylate occudin in vitro and to demonstrate that the shift observed on the 2D gel results from a phosphorylation event, we incubated BREC lysates with a recombinant occludin C-terminal fragment and ³²P ATP in an in vitro kinase assay. Figure 2Bshows a dramatic increase in the incorporated isotope in the occludin protein fragment, demonstrating that BREC lysate can phosphorylate occludin on its C terminus in vitro. These results confirm our previous observations of VEGF-induced occludin phosphorylation that was abolished by alkaline phosphatase. ³¹  

To test further the involvement of PKC in VEGF-stimulated occludin phosphorylation, BREC lysates were also analyzed for the occludin phosphorylation state by one-dimensional (1D) Western blot analysis. As a positive control for PKC activation, BRECs were grown to confluence and treated with 90 nM PMA for 1 hour, to stimulate PKC activity. PMA treatment of BRECs under serum conditions shifted nearly all the occludin to the more slowly migrating, phosphorylated β band relative to untreated control (Fig. 3A) . To test the effect of VEGF on the observed occludin gel-shift, BRECs, 2 days after confluence, were stimulated with VEGF (1.2 nM) for 15 minutes after a 30-minute pretreatment with 30 or 600 nM LY379196. Lysate supernatants were subjected to SDS-PAGE and immunoblot analysis for occludin (Fig. 3B) . Two occludin bands were detected—α and β—and at 15 minutes, VEGF increased occludin phosphorylation 28.5% (P < 0.05); this effect was completely blocked with 30 or 600 nM LY379196 (Fig. 3C ; P < 0.05 and P < 0.001, respectively). These results further support a model in which VEGF-induced occludin phosphorylation is mediated by PKCβ. 

PKCβ in VEGF-Induced Permeability in BRECs

To test the hypothesis that PKCβ activity is necessary for VEGF-induced permeability, confluent BRECs were pretreated with LY379196 (30 or 600 nM) for 30 minutes and were then stimulated with VEGF (1.2 nM) for 60 minutes. RITC-dextran (70 kDa) was added to the apical chamber, accumulation of dextran in the basolateral chamber was measured over the next 4 hours, and diffusive flux (Po) was determined as described in the Materials and Methods section. This tracer was chosen because it is similar in size to albumin, which accumulates in retinas with diabetes. VEGF increased dextran permeability by nearly twofold (P < 0.001), and pretreatment with a high concentration of LY379196 (600 nM) attenuated the VEGF-induced permeability by approximately 50% (Fig. 4) . The low concentration of LY379196 (30 nM) demonstrated a trend toward a decrease in the VEGF-induced permeability but was not statistically significant. This result suggests that PKCβ may contribute to VEGF-induced permeability but that other isoforms may also contribute to VEGF-induced flux of 70-kDa dextran in BRECs. 

Effect of the PKC Inhibitor BIM I on VEGF-Stimulated Occludin Phosphorylation

To determine whether the VEGF-induced increase in occludin phosphorylation could be completely blocked by a more broad-acting PKC inhibitor, we tested the ability of BIM I to block phosphorylation. Bisindolylmaleimide I (BIM I, also known as Gö-6850 and GF109203X) is a competitive inhibitor of the ATP binding site of PKC and is selective for the conventional (α, βI, βII, and γ) and novel (δ and ε) isoforms of PKC. Confluent BRECs were pretreated with BIM I (5 μM) and stimulated with VEGF (1.2 nM), the PKC activator PMA (90 nM), or the negative control 4α-PMA (90 nM) for 15 minutes. Lysate supernatants were subjected to SDS-PAGE and immunoblotted for occludin (Fig. 5A) . Three occludin bands were detected: α, β, and γ, representing basal (α) and increasingly phosphorylated forms of occludin (β and γ), as shown previously. ³¹ VEGF increased occludin phosphorylation, represented by the ratio of the sum of the β and γ forms to the α form, by 38% at 15 minutes (P < 0.001 compared with control), and this effect was completely blocked with BIM I (P < 0.001 compared with VEGF and P > 0.05 compared with control). PMA also increased occludin phosphorylation relative to control (P < 0.001) and to 4α-PMA-treated cells (P < 0.001). Occludin content was unaltered with VEGF or BIM I treatment but was significantly reduced with 15 minutes of PMA and 4α-PMA treatment (data not shown). As a positive control for the effectiveness of the inhibitor BIM I or the activator PMA, phosphorylation of p42/44 MAPKinase was determined. BIM I prevented VEGF-induced phosphorylation of p42/44 MAPK, and PMA induced phosphorylation of p42/44 MAPK. These results indicate that VEGF-induced occludin phosphorylation is mediated by PKC isoforms. 

Blockage of VEGF-Induced Endothelial Cell Permeability with BIM I

The ability of the inhibitor BIM I to block VEGF-induced occludin phosphorylation prompted us to examine the effect of this inhibitor on endothelial cell permeability in vitro. Confluent BRECs were pretreated with BIM I (5 μM) for 30 minutes and were then stimulated with VEGF (1.2 nM) for 60 minutes. RITC-dextran (70 kDa) was added to the apical chamber, and accumulation of dextran in the basolateral chamber was measured over the next 4 hours (Fig. 6) . Diffusive flux (Po) was calculated for each treatment condition with control Po of 8.49 × 10⁻⁷ cm/s. VEGF increased dextran permeability by nearly 60% (P < 0.001 compared with control), whereas pretreatment with BIM I attenuated the VEGF-induced permeability by more than half (P < 0.001 compared with VEGF). However, VEGF’s effect was not completely blocked (P < 0.01 compared with control). This result indicates that activation of classic and/or novel PKCs is necessary for part of the VEGF-induced permeability to 70-kDa dextran in BRECs. 

Effect of Dominant Negative PKCβ on VEGF-Induced Occludin Phosphorylation and Endothelial Permeability

To verify further the role of PKC in the regulation of vascular permeability, BRECs were transfected with plasmids encoding either wild-type or dominant negative (S217A) PKCβII. Control cells were transfected with plasmids expressing green fluorescent protein (pEGFP) to allow determination of BREC transfection efficiency (∼75%, data not shown). BRECs were transfected with pEGFP vector (CN) or wild-type PKCβII or PKCβII (S217A), grown to confluence, and the effect of VEGF on permeability to 70-kDa RITC dextran was measured as just described. VEGF induced an approximately twofold increase in dextran permeability in the pEGFP-transfected control cells (P < 0.001), whereas transfection of wild-type PKCβII increased the permeability approximately threefold—significantly greater than pEGFP transfected cells treated with VEGF (P < 0.001; Fig. 7A ). Transfection of dnPKCβII (S217A) significantly reduced permeability to 70-kDa RITC dextran in response to VEGF compared with pEGFP (P < 0.01) and with wild-type PKC transfected cells treated with VEGF (P < 0.001). The response to VEGF in the PKCβII (S217A) transfected cells was not significantly greater than untreated cells, although there was a trend toward an increase that was comparable to that observed with the chemical inhibitors of PKC, suggesting some VEGF induction of permeability remained. These data suggest that PKCβII regulates most, but not all, of the VEGF-induced permeability to 70-kDa dextran, consistent with results reported herein. 

As a control for PKCβII activity in transfected cells, we measured the phosphorylation of the downstream effector p42/44 MAPK after VEGF stimulation. Transfection of BRECs with wild-type PKCβII significantly increased phosphorylation of p42/44 MAPK in response to VEGF stimulation compared with control cells transfected with pEGFP vector (P < 0.05). Conversely, expression of PKCβII (S217A) abolished VEGF induction of p42/44 MAPK phosphorylation (Fig. 7B) . These data demonstrate that transfection of wild-type PKCβII effectively enhanced PKC activity and that the S217A mutant acted as a dominant negative isoform in this cell system. The induction of MAPK phosphorylation was completely abolished, as determined with a phospho-specific antibody in Western blot analysis. The lack of any remaining activity when transfection rate was ∼75% could be due to the sensitivity of the assay or a greater expression level of the PKC mutant than the pEGFP plasmid, among other possibilities. However, this control experiment demonstrates the effectiveness of the S217A mutant to act in a dominant negative manner in the BRECs. 

Occludin phosphorylation in control cells and in cells transfected with wild-type PKCβII or PKCβII (S217A) was determined by gel shift on SDS-polyacrylamide gels. (Fig. 7C) . The ratio of phosphorylated-to-basal occludin was increased in control cells treated with VEGF (P < 0.05), and VEGF-induced phosphorylation of occludin was enhanced in cells transfected with wild-type PKCβII (P < 0.05 compared with control transfection treated with VEGF). In contrast, VEGF-induced occludin phosphorylation as determined by gel shift, was completely inhibited in BRECs transfected with PKCβII (S217A). These results indicate that occludin phosphorylation is mediated by PKC, and are consistent with a causal role of PKCβ in VEGF-induced occludin phosphorylation. 

Effect of Diabetes on Retinal Occludin Phosphorylation

Experimental diabetes reduces retinal occludin content and enhances retinal vascular permeability to albumin. ³⁰ To determine whether occludin phosphorylation is altered in diabetic animals, retinas from control and diabetic Sprague-Dawley rats were collected and immunoblotted for occludin. Previous immunohistochemical analysis of retina reveals that occludin expression is restricted to the vasculature. ^{29 30} In our retinal excision, the pigmented epithelium, which also expresses occludin, was not collected. Thus, the occludin changes observed by Western blot represent retinal vascular occludin. In repeat experiments, occludin phosphorylation increased significantly after both 1 and 3 months of diabetes, as determined by gel shift assay (Table 1) . In addition, the ratio of phosphorylated to basal occludin in control and diabetic states was similar to that seen in control and VEGF-treated BRECs, respectively. These data demonstrate that occludin phosphorylation changes are equivalent in our cell culture model system and in the diabetic rat retina. 

Discussion

The goal of these studies was to determine the mechanism of increased vascular permeability in response to diabetes and VEGF, and in particular, the role of PKC. Two chemical inhibitors of PKC and dominant negative PKCβII inhibited approximately half of the VEGF-stimulated permeability to 70-kDa RITC dextran, whereas the VEGF induction of occludin phosphorylation was completely prevented. Furthermore, transfection of PKCβII augmented VEGF-induced occludin phosphorylation and endothelial permeability. Finally, in vivo data from diabetic rat retina further support a role for occludin phosphorylation in vascular permeability. These studies strongly suggest that at least part of the VEGF-induced permeability to 70-kDa dextran requires classic or novel PKC activity and suggest occludin phosphorylation as a critical downstream target. Although PKC activity is apparently necessary for occludin phosphorylation, it unclear whether occludin is a direct target of PKC activity. Indeed, 2D gels suggest that the basal state of occludin exists in at least four phosphorylated forms and VEGF stimulates occludin to a minimum of five phosphorylated forms, with a clear shift toward multiple phosphorylation states. Thus, VEGF may induce a cascade of occludin phosphorylation that requires more than PKC activation alone. 

VEGF functions as a potent endothelial permeabilizing factor, as evidenced by the fact that addition of VEGF stimulates fluorescein flux in rats ¹ and increases hydraulic conductivity and permeability to albumin in BREC monolayers. ^{36 42} Inhibitors of PKC prevent changes in tracer flux across retinal vasculature in rats after intravitreal injection of VEGF ¹ or in diabetes. ⁴³ VEGF-induced vascular permeability was also prevented by PKC inhibition in isolated, perfused coronary venules ^{44 45} and in bovine pulmonary artery endothelial cells. ⁴⁶ Together, these studies establish the significance of PKC activation in the induction of vascular permeability. The results presented herein further strengthen this link and suggest that VEGF activation of PKC leads to phosphorylation of the tight junction protein occludin, regulating endothelial permeability. However, an important observation from this work is that the classic and novel PKC isoforms only account for part of the VEGF induction of permeability in retinal endothelium. 

Our results are consistent with a model in which activation of PKC is required for VEGF-induced permeability by induction of occludin phosphorylation. Occludin phosphorylation induced by VEGF or shear stress occurs at 15 minutes, the same time an increase in water flux across BREC monolayers is observed, and treatments that block occludin phosphorylation also prevent water flux. ^{31 32 36 42} Conversely, hydrocortisone treatment reduces permeability across endothelial monolayers associated with decreased occludin phosphorylation and increased occludin content and assembly in tight junctions. ²⁵ Although the increased diffusive permeability across endothelial monolayers to 70-kDa dextran was modest (∼2-fold), it is likely that blood flow and hydrostatic pressure would augment the VEGF-stimulated flux in vivo. Furthermore, the observed change in occludin phosphorylation is transient, peaking by 15 minutes and returning to baseline by 60 minutes. ³¹ However, increased endothelial permeability persists >4 hours after VEGF stimulation. There are several possible explanations for this effect. For example, occludin phosphorylation may serve as an initial trigger to change tight junction architecture and localization. Indeed, PDGF induces redistribution of occludin from the plasma membrane to the cytoplasm in epithelial cells, ⁴⁷ similar to the redistribution of occludin in diabetic rat retinas that is associated with open tight junctions. ²⁹ The mechanism by which this redistribution occurs is unclear but may include transient phosphorylation of occludin. 

Other research has demonstrated occludin phosphorylation and suggests that PKC plays a role in posttranslational modification. Histamine increases phosphorylation of occludin as observed by 1D gel electrophoresis and by an acidic shift on 2D gel electrophoresis with incorporation of ³²P, ³³ similar to the 2D gels in the present study. In cultures of brain endothelium, MCP induces permeability as measured by electrical resistance and inulin flux, which is associated with occludin phosphorylation, as well as phosphorylation of other tight junction proteins. ³⁴ In these studies, immunoprecipitation of occludin, claudin 5, and ZO-1 followed by phosphoserine blot demonstrated phosphorylation of tight junction proteins. We have not been able to detect occludin phosphorylation in BRECs by immunoprecipitation followed by phosphoserine blot (data not shown). This discrepancy could be due to differences in specific sites of phosphorylation in the different systems or differences in blotting technique. However, the ability of alkaline phosphatase to collapse the gel-shifted bands to one band, ³¹ incorporation of ³²P from radiolabeled ATP into occludin fragments treated with BREC lysate (Fig 2B) , and 2D gel electrophoresis demonstrating increased acidic shift in a prototypical phosphorylation pattern (Fig 2A) , strongly support that the gel shift assay for occludin represents phosphorylation. Although it is unclear whether PKC directly phosphorylates occludin or it activates signaling pathways leading to increased phosphorylation of occludin, a C-terminal region of mouse occludin was phosphorylated in vitro by a purified mixture of the PKC isoforms α, βI, βII, and γ. ⁴⁸  

The results herein demonstrate that classic and/or novel PKC isoforms contribute to part of the VEGF induced permeability in retinal endothelial cells. Furthermore, the data strongly suggest that PKC-dependent phosphorylation of occludin contributes to increased endothelial permeability in cell culture and in retinal blood vessels during diabetes. These findings help in understanding the beneficial effects of a PKC inhibitor on vision in patients with diabetic retinopathy. ⁴⁹ Studies to define sites of occludin phosphorylation and additional pathways for diabetes and VEGF mediated permeability are under way. 

²  Present affiliation: Department of Microbiology and Immunology, University of Miami Miller School of Medicine, Miami, Florida.

 Supported by Grant EY012021 from the National Eye Institute (DAA) and by funding by the Juvenile Diabetes Research Foundation (DAA, TWG) and the Pennsylvania Lions Sight Conservation and Eye Research Foundation (DAA). TWG is the Jack and Nancy Turner Professor of Ophthalmology.

 Submitted for publication March 24, 2006; revised June 26, 2006; accepted September 21, 2006.

 Disclosure: N.S. Harhaj, None; E.A. Felinski, None; E.B. Wolpert, None; J.M. Sundstrom, None; T.W. Gardner, None; D.A. Antonetti, Apogee (C)

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

VEGF activates Akt and p42/44 MAPK in BRECs with 3 days of hydrocortisone. (A) BRECs were stimulated with VEGF (1.2 nM) for 0 to 60 minutes. Lysate supernatants were subjected to SDS-PAGE and immunoblot analysis for Ser473 phospho-Akt and total Akt. Data depict phosphorylated Akt normalized to total Akt from three experiments (n = 9 for each condition). VEGF caused a 20-fold increase in Akt phosphorylation at 5 minutes that decreased to control levels by 60 minutes. (B) Lysate supernatants were subjected to SDS-PAGE and immunoblot analysis for phosphorylated (Thr₂₀₂ and Tyr₂₀₄) and total p42/44 MAPK. Data are phosphorylated p42 or p44 MAPK normalized to total p42 or p44 MAPK from three experiments (n = 9 for each condition). VEGF significantly increased p44 MAPK phosphorylation from 5 to 60 minutes, whereas p42 MAPK exhibited the same trend toward increased phosphorylation. Error bars, SEM; statistical comparison by nonparametric ANOVA with the Dunn multiple comparison test (*P < 0.05, **P < 0.01, ***P < 0.001).

Figure 2.

 

Occludin was phosphorylated in vitro by BREC lysate and LY379196 attenuated VEGF-induced occludin phosphorylation in BRECs, as determined by 2D gel electrophoresis. (A) Confluent BRECs were maintained in serum-containing medium, pretreated with LY379196 (30 nM) for 30 minutes, and stimulated with VEGF (1.2 nM) for 15 minutes. Cells were harvested, and 750 μg protein was focused on pH-4 to -7 IPG strips for 81,900 V/h. IPG strips were equilibrated in an SDS-containing buffer and proteins were separated in the second dimension on 10% SDS-polyacrylamide gels. Proteins were transferred to nitrocellulose and immunoblotted for occludin. At least four forms of occludin could be resolved in control conditions. VEGF increased the percentage of occludin in more acidic forms, consistent with an increase in occludin phosphorylation. LY379196 (30 nM)-treated cells yielded a similar occludin migration pattern to that in untreated cells. LY379196 blocked the VEGF-induced shift of occludin to acidic forms. (B) A His-tagged C-terminal fragment of occludin (amino acid 416-523) was expressed and purified from Escherichia coli. Lysates were made from 2-day confluent BRECs and an in vitro kinase assay was performed using ³²P-ATP for 0 or 5 minutes. The resultant labeled phosphoproteins were separated by SDS-PAGE and visualized by autoradiography.

Figure 3.

 

VEGF-induced occludin phosphorylation in BRECs depended on PKC. (A) Confluent BRECs were maintained in serum-containing medium and were treated with PMA (90 nM) for 1 hour. Whole-cell lysates were separated by SDS-PAGE and immunoblot for occludin. (B) Confluent BRECs were maintained in serum-containing medium, pretreated with LY379196 (30 nM or 600 nM) for 30 minutes, and stimulated with VEGF (1.2 nM) for 15 minutes. Lysate supernatants were subjected to SDS-PAGE and immunoblot for occludin. (C) Two occludin bands were detected: α and β, and the ratio of the β to the α bands was calculated. n =12 for control and VEGF; n = 8 for all other conditions. Statistical analysis was by ANOVA followed by the Tukey post hoc test; *P < 0.05 and *** P < 0.001.

Figure 4.

 

LY379196 attenuated VEGF-induced permeability to 70-kDa dextran in BRECs. Confluent BRECs on fibronectin-coated filters with hydrocortisone for 1 day were pretreated with LY379196 (30 or 600 nM) for 30 minutes, and stimulated with VEGF (1.2 nM) for 1 hour. To measure the effect on permeability, RITC-dextran (70 kDa) was added to the apical chamber, accumulation of dextran in the basolateral chamber was measured over the next 4 hours, and diffusive flux (Po) was calculated. VEGF increased dextran permeability by nearly twofold, whereas pretreatment with 600 nM LY379196 attenuated the VEGF-induced permeability by approximately half. Results from four experiments (n = 13 to 24 for each condition). The average Po for the control samples was 2.78 × 10⁻⁶ cm/sec. Error bars, SEM; statistical comparison by Kruskal-Wallis nonparametric ANOVA with the Dunn multiple comparison test. Conditions with a are statistically the same as one another, whereas conditions with b are statistically the same as each other. Condition ab is not statistically different from either a or b.

Figure 5.

 

PKC mediated VEGF-induced occludin phosphorylation in BRECs. (A) Confluent BRECs under hydrocortisone conditions were pretreated with BIM I (5 μM) and stimulated with VEGF (1.2 nM), PMA (90 nM), or 4α-PMA (90 nM) for 15 minutes. Lysate supernatants were subjected to SDS-PAGE and immunoblot for occludin. Three occludin bands were detected—α, β, and γ, which have been shown to be different phosphorylation states of occluding—and the ratio of the γ and β to the α bands was calculated. VEGF increased occludin phosphorylation by 38% at 15 minutes, whereas pretreatment with BIM I completely blocked this increase. PMA also increased occludin phosphorylation relative to control and 4α-PMA-treated cells. Occludin content was unchanged with VEGF or BIM I but was significantly reduced with 15 minutes of PMA and 4α-PMA treatment. Experiments with BIM were performed three times and experiments with PMA were performed two times. Data are from one representative experiment (n = 5 or 6 at each time point). Error bars, SEM; statistical comparison by one-way ANOVA with Student-Neuman-Keuls multiple comparison test, *P < 0.05, ***P < 0.001. (B) Lysate supernatants were subjected to SDS-PAGE and immunoblot for phosphorylated p42/44 MAPK. The graphs depict phosphorylated p42 or p44 MAPK relative to the control with n= 6 for each condition. Error bars, SEM. Total p42/44 MAPK content was unchanged (not shown). VEGF and PMA significantly increased phosphorylation of both p42 and p44 MAPK, and BIM I significantly reduced VEGF-induced p42 and p44 MAPK phosphorylation to control levels. Experiments with BIM I were performed three times, and experiments with PMA were performed two times, with similar results. Data are from one representative experiment; ***P < 0.001 (one-way ANOVA with Student-Neuman-Keuls multiple comparison test).

Figure 6.

 

BIM I attenuated VEGF-induced permeability to 70-kDa dextran in BRECs. Confluent BRECs on fibronectin-coated filters treated with hydrocortisone for 3 days were pretreated with BIM I (5 μM) for 30 minutes, and were stimulated with VEGF (1.2 nM) for 1 hour. RITC-dextran (70 kDa) was added to the apical chamber, and accumulation of dextran in the basolateral chamber was measured over the next 4 hours. The average Po for the control samples was 8.49 × 10⁻⁷ cm/s. VEGF increased dextran permeability by nearly 60%, whereas pretreatment with BIM I reduced the VEGF-induced permeability by more than half. Data are results from two experiments (n = 12 for each condition). Error bars, SEM; statistical comparison was made by one-way ANOVA with Student-Neuman-Keuls multiple comparison test (**P < 0.01, ***P < 0.001).

Figure 7.

 

Transfection of PKC isoforms attenuated VEGF-induced 70-kDa dextran permeability and occludin phosphorylation. (A) BRECs were grown to 70% confluence and transfected with pEGFP (control), wild-type PKCβ expression plasmid (wt), or PKCβ S217A (dominant negative), by the standard endothelial cell nucleofection technique. Transfected cells were plated onto fibronectin-coated filters with a 0.4-μm pore size. Cells were allowed to grow to confluence in media containing 10% serum and then switched to serum-free medium for 24 hours. VEGF samples were treated with 1.2 nM VEGF for 30 minutes before the start of the transport assay with 70 kDa RITC-dextran, which was performed with samples taken every 30 minutes for 4 hours. Data were analyzed by one-way ANOVA followed by the Bonferroni post hoc test (n = 8 for each group; ** P < 0.01 and *** P < 0.001). (B) Cells were transfected with GFP alone (CN or control) or with wild-type PKCβII (WT) or dominant negative PKCβII (S217A). VEGF treatment (1.2 nM) was for 15 minutes after which, cells were harvested and immunoblotted for MAP kinase or occludin (C). Statistical analysis was by one-way ANOVA followed by the Tukey post hoc test (n = 6; *P < 0.05).

Table 1.

 

Occludin Phosphorylation in Rat Retina

	Control β + γ/α	Diabetes β + γ/α
1 Month	0.82 ± 0.03 (n = 6)	1.13 ± 0.07 (n = 6), P < 0.005
3 Month	0.73 ± 0.04 (n = 6)	1.01 ± 0.06 (n = 4), P < 0.005

 Rats were made diabetic by streptozotocin, and retinas were harvested at the times indicated. Retinal extracts were separated by polyacrylamide gel electrophoresis and subjected to immunoblot analysis for occludin. The occludin phosphorylation state was determined by gel-shift assay. Results were analyzed by Student’s t-test; each time point was repeated with similar results.