PEDF regulates vascular permeability by a γ-secretase-mediated pathway

Abstract

Increased vascular permeability is an inciting event in many vascular complications including diabetic retinopathy. We have previously reported that pigment epithelium-derived factor (PEDF) is able to inhibit vascular endothelial growth factor (VEGF)-induced angiogenesis through a novel γ-secretase-dependent pathway. In this study, we asked whether inhibition of VEGF-induced permeability by PEDF is also γ-secretase-mediated and to dissect the potential mechanisms involved. Vascular permeability was assessed in vitro by measuring transendothelial resistance and paracellular permeability to dextran and in vivo by following leakage of intravenous FITC-labelled albumin into the retina in the presence or absence of VEGF and PEDF in varying combinations. Experiments were undertaken in the presence or absence of a γ-secretase inhibitor. To assess junctional integrity immunohistochemistry for the adherens junction (AJ) proteins, VE-cadherin and β-catenin, and the tight junction (TJ) protein, claudin-5 was undertaken using cultured cells and flat mount retinas. Protein expression and the association between AJ proteins, VEGF receptors (VEGFRs) and γ-secretase constituents were determined by immunoprecipitation and Western Blot analysis. In selected experiments the effect of hypoxia on junctional integrity was also assessed. PEDF inhibition of VEGF-induced permeability, both in cultured microvascular endothelial cell monolayers and in vivo in the mouse retinal vasculature, is mediated by γ-secretase. PEDF acted by a) preventing dissociation of AJ and TJ proteins and b) regulating both the association of VEGF receptors with AJ proteins and the subsequent phosphorylation of the AJ proteins, VE-cadherin and β-catenin. Association of γ-secretase with AJ proteins appears to be critical in the regulation of vascular permeability. Although hypoxia increased VEGFR expression there was a significant dissociation of VEGFR from AJ proteins. In conclusion, PEDF regulates VEGF-induced vascular permeability via a novel γ-secretase dependent pathway and targeting downstream effectors of PEDF action may represent a promising therapeutic strategy for preventing or ameliorating increased vascular permeability.

Figure 1. PEDF blocks microvascular endothelial permeability via a γ-secretase dependent mechanism.

(a) Temporal changes in transendothelial resistance across a microvascular endothelial monolayer grown on a Transwell insert treated with vehicle (unstimulated), VEGFA alone, PEDF alone, simultaneous VEGFA+PEDF, PEDF 6 hours post VEGF (n = 4 independent experiments). VEGFA and PEDF were used at 100 ng/ml. (b) Paracellular macromolecular permeability to 40 kDa Dextran-FITC using the conditions described in (A) (n = 4). In the case of PEDF 6 hours post VEGF, Transwell inserts were transferred to new wells containing basal medium without fluorescent dextran. (c) and (d) The effect of γ-secretase inhibitor (γ-SI) on PEDF reduction of VEGF-induced endothelial permeability; (c) TER, (d) paracellular flux. γ-secretase inhibitors were used at 1 nM and data is shown for L685485. Data are represented as means (n = 4) ± SEM.

Figure 2. The temporal response of AJs and TJs to VEGF or VEGF plus PEDF.

Representative pictures of confluent cultures of microvascular endothelial cells treated with VEGFA alone or VEGFA followed by PEDF 6 hours later triple stained for VE-cadherin (red), claudin-5 (green) and nuclei (DAPI) and assessed at different times over 24 hours using confocal microscopy. VEGFA and PEDF were used at 100 ng/ml. Scale bar = 100 µm.

Figure 3. PEDF prevents VEGF-induced dissociation of endothelial AJs and TJs.

Representative immunofluorescent images of confluent cultures of microvascular endothelial cells treated with vehicle (unstimulated), VEGFA alone, PEDF alone, simultaneous VEGFA+PEDF, PEDF 6 hours post VEGF and PEDF 6 hours post VEGF+γ-secretase inhibitor (γ-SI) (n = 4 independent experiments). VEGFA and PEDF were used at 100 ng/ml and γ-secretase inhibitor at 1 nM. (a) Cultures were triple stained for VE-cadherin (red), claudin-5 (green) and nuclei (DAPI, blue) and assessed at different times over 24 hours using confocal microscopy. Merged images are shown with colocalization of VE-cadherin and claudin-5 appearing as yellow. (b) The effect of PEDF on β-catenin using the conditions described in (A) (n = 4). Scale bar = 100 µm.

Figure 4. PEDF blocks VEGF-induced retinal vascular permeability in mice.

C57BL/6 mice received intravitreal injections of vehicle (unstimulated), VEGFA alone, PEDF alone, simultaneous VEGFA+PEDF, PEDF 6 hours post VEGF and PEDF 6 hours post VEGF+γ-secretase inhibitor (γSI) (n = 18 animals per treatment). Test compounds (1 µl per eye were given at the following concentrations, VEGF 80 ng/µl; PEDF (80 ng/µl); L685485 24 ng/µl; DAPT 16 ng/µl. 46 hours post the first injection mice received tail vein injections of FITC-labeled albumin. Uninjected animals acted as the baseline control. (a) 46 hours post the first injection mice received tail vein injections of FITC-labeled albumin and retinas were taken for analysis 2 hours later (n = 10–20 per group). Leakage of systemic FITC-labeled albumin into the retina was assessed by measuring total fluorescence in homogenized retina using a fluorescence plate reader. Control is the baseline fluorescence in untreated animals. Data are represented as means ± SEM. (b) Representative confocal microscopy showing dilated vessels and leakage of FITC-labeled albumin in the retinas of mice treated with vehicle (unstimulated), VEGF, PEDF, PEDF 6 hours post VEGF and PEDF 6 hours post VEGF+γ-secretase inhibitor (γ-SI). Scale bar = 50 µM.

Figure 5. PEDF prevents VEGF-induced dissociation of endothelial AJs and TJs in the retinal vasculature of mice.

(a) Representative confocal images of retinal vessels in flat mount preparations from unstimulated and animals, treated as described in Figure 4, immunostained with VE-cadherin or claudin-5 (red) and FITC-conjugated agglutinin (green) to visualize retinal vessels. Merged images are shown with colocalization of VE-cadherin and claudin-5 appearing as yellow. Scale bar = 50 µM. The lower panel (b) shows representative merged higher power images of retinal vessels stained for VE-cadherin (i–iii) or claudin-5 (iv–vi) (red) and FITC-conjugated agglutinin (green). i, iv = unstimulated; ii, v = VEGF treatment; iii, vi = PEDF 6 hr after VEGF. Scale bar = 10 µM.

Figure 6. PEDF maintains the association between AJ proteins and VEGF receptors in VEGF treated cells.

Confluent retinal endothelial cells were either unstimulated (1) treated with VEGF alone (2), PEDF alone (3), VEGF followed by PEDF 6 hours later without (4) and with (5) γ-secretase inhibitor for 24 hours. VEGFA and PEDF were used at 100 ng/ml and γ-secretase inhibitor at 1 nM. Total cell lysates were split into four equal portions and immunoprecipitated with anti-VE-cadherin, anti-β-catenin, anti-VEGFR1-C-terminus and anti-VEGFR2-C-terminus, and then subsequent Western blot analyzed using these four antibodies, respectively. (a) A panel of representative Western blots. (b) Band intensities of replicate experiments (n = 4 independent experiments) were quantified as described in the Methods and regression analysis undertaken to assess the association of these proteins.

Figure 7. PEDF inhibits VEGF-induced phosphorylation of membrane-bound AJ proteins and promotes association of presenilin and nicastrin with VE-cadherin.

Confluent retinal endothelial cells were either unstimulated (1) treated with VEGF alone (2), PEDF alone (3), VEGF followed by PEDF 6 hours later without (4) and with (5) γ-secretase inhibitor over 24 hours. VEGFA and PEDF were used at 100 ng/ml and γ-secretase inhibitor at 1 nM. Cells were separated into membrane and cytosolic fractions followed by immunoprecipitation. (a) Total VE-cadherin levels were determined by immunoprecipitation of the membrane fraction with anti-PY20 and Western blots for VE-cadherin. A panel of representative Western blots is shown on left and band intensities of replicate experiments (n = 4) on the right. (b) Western blot analysis to assess the effect of the different treatments on tyrosine phosphorylation of VE-cadherin at sites Y658 and Y731 (n = 4 independent experiments). (c) Total phosphorylation of β-catenin was assessed by immunoprecipitation of the membrane fraction with anti-PY20 and Western blots for β-catenin. (n = 4) (d) The association of presenilin-1 (PS1) and nicastrin (NTC) were confirmed by immunoprecipitation with anti-VE-cadherin, followed by Western blot analysis using anti-presenilin-1 and nicastrin. (e) To further confirm a role for γ-secretase in regulating vascular permeability presenilin-1 or nicastrin were knocked down with by transfection with AAV 2 expressing PS1 and NTC siRNAs. Scrambled siRNA acted as a control. Paracellular macromolecular permeability to 40 kDa Dextran-FITC was determined in cells treated with VEGF, PEDF, VEGF+PEDF and compared to unstimulated control. Densitometry analyses in (a), (c) and (d) are shown as the relative ratio of phosphorylated pVE-cadherin, pβ-catenin, presenilin-1 and nicastrin to heavy IgG chains. Densitometry in b is shown as the relative ratio to α-tubulin. The data are mean ± SEM. *p<0.05, **p<0.01 versus unstimulated group.

Figure 8. Hypoxia increases vascular permeability.

Temporal changes in transendothelial resistance across a microvascular endothelial monolayer grown on a Transwell insert treated with vehicle (unstimulated), VEGFA alone, PEDF alone, simultaneous VEGFA+PEDF, PEDF 6 hours post VEGF and PEDF 6 hours post VEGF+γ-secretase inhibitor (γ-SI) under either (a) hypoxia or (b) retinal normoxia. (n = 4 independent experiments). VEGFA and PEDF were used at 100 ng/ml and γ-secretase inhibitor at 1 nM. In the case of PEDF 6 hours post VEGF, Transwell inserts were transferred to new wells containing basal medium without fluorescent dextran. Data are respresented as means± SEM.

Figure 9. Hypoxia increases the expression of VEGF receptors and γ-secretase along with VEGFR dissociation from AJ proteins.

(a) Confluent retinal microvascular endothelial cells were exposed to hypoxia for up to 24 hours. Cell lysated were examined by Western blot for VEGFR1, VEGFR2, presenilin-1 (PS-1), nicastrin (NCT), VE-cadherin and β-catenin. α-tubulin acted as the house keeping control. A representative Western blot is shown together with densitometric analysis from three separate experiments. (b) Confluent retinal microvascular endothelial cells were maintained under standard incubator conditions, retinal normoxia and hypoxia for 24 hours. Total cell lysates were split into four equal portions and immunoprecipitated with anti-VE-cadherin, anti-β-catenin, anti-VEGFR1-C-terminus and anti-VEGFR2-C-terminus, and then subsequent Western blot analyzed using these four antibodies, respectively. A panel of representative Western blots is shown. Band intensities of replicate experiments (n = 4) were quantified as described in the Methods and regression analysis undertaken to assess the association of these proteins.