Endothelial mechanisms for inactivation of inflammation-induced hyperpermeability

Abstract

Microvascular hyperpermeability is a hallmark of inflammation. Many negative effects of hyperpermeability are due to its persistence beyond what is required for preserving organ function. Therefore, we propose that targeted therapeutic approaches focusing on mechanisms that terminate hyperpermeability would avoid the negative effects of prolonged hyperpermeability while retaining its short-term beneficial effects. We tested the hypothesis that inflammatory agonist signaling leads to hyperpermeability and initiates a delayed cascade of cAMP-dependent pathways that causes inactivation of hyperpermeability. We applied platelet-activating factor (PAF) and vascular endothelial growth factor (VEGF) to induce hyperpermeability. We used an Epac1 agonist to selectively stimulate exchange protein activated by cAMP (Epac1) and promote inactivation of hyperpermeability. Stimulation of Epac1 inactivated agonist-induced hyperpermeability in the mouse cremaster muscle and in human microvascular endothelial cells (HMVECs). PAF induced nitric oxide (NO) production and hyperpermeability within 1 min and NO-dependent increased cAMP concentration in about 15-20 min in HMVECs. PAF triggered phosphorylation of vasodilator-stimulated phosphoprotein (VASP) in a NO-dependent manner. Epac1 stimulation promoted cytosol-to-membrane eNOS translocation in HMVECs and in myocardial microvascular endothelial (MyEnd) cells from wild-type mice, but not in MyEnd cells from VASP knockout mice. We demonstrate that PAF and VEGF cause hyperpermeability and stimulate the cAMP/Epac1 pathway to inactivate agonist-induced endothelial/microvascular hyperpermeability. Inactivation involves VASP-assisted translocation of eNOS from the cytosol to the endothelial cell membrane. We demonstrate that hyperpermeability is a self-limiting process, whose timed inactivation is an intrinsic property of the microvascular endothelium that maintains vascular homeostasis in response to inflammatory conditions.NEW & NOTEWORTHY Termination of microvascular hyperpermeability has been so far accepted to be a passive result of the removal of the applied proinflammatory agonists. We provide in vivo and in vitro evidence that 1) inactivation of hyperpermeability is an actively regulated process, 2) proinflammatory agonists (PAF and VEGF) stimulate microvascular hyperpermeability and initiate endothelial mechanisms that terminate hyperpermeability, and 3) eNOS location-translocation is critical in the activation-inactivation cascade of endothelial hyperpermeability.

Keywords: endothelium; hyperpermeability; inflammation; vascular biology.

Graphical abstract

Figure 1.

Endothelial barrier restoration is self-limiting. Human microvascular endothelial cells were grown to confluence and treated with platelet-activating factor (PAF) or vascular endothelial growth factor (VEGF) (time point 0). A, left: representative trace during a continuous 100 nmol/L PAF application initially increased the abluminal fluorescence intensity vs. time slope (an index of permeability), which subsequently declined, suggesting restoration toward control endothelial permeability. Representative data: baseline slope (−15 to −5 min) = 0.2897, PAF slope (0 to 15 min) = 2.888, and restoration slope (20 to 50 min) = 1.237. A, right: PAF slopes quantification *P < 0.05. B, left: representative trace during a continuous 1 nmol/L VEGF application initially increased the abluminal fluorescence intensity vs. time slope (an index of permeability), which subsequently declined, suggesting restoration toward control endothelial permeability. Representative data: baseline slope (−15 to −5 min) = 0.0466, VEGF slope (0 to 15 min) = 0.23, and restoration slope (20 to 50 min) = 0.1316. AU, arbitrary units. B, right: slopes quantification, *P < 0.05; N = 5.

Figure 2.

Stimulation of Epac-1 inactivates hyperpermeability in vivo. A: 100 nmol/L platelet-activating factor (PAF) induced a significant increase in permeability in mouse cremaster muscle, as indicated by an elevation in integrated optical intensity (IOI). Administration of 10 µmol/L 8cPT-cAMP (8cPT) after PAF challenge significantly reduced the peak IOI and accelerated the return of baseline permeability: N (PAF) = 5; N (PAF + 8cPT) = 5; P < 0.05 PAF vs. PAF + 8cPT at 10 to 50 min. B: application of 1 nmol/L ABT-491 (PAF receptor antagonist) after PAF administration failed to inactivate PAF-induced hyperpermeability: N (PAF) = 5; N (PAF + ABT-491) = 5. C: blocking eNOS production with 10 µmol/L l-NMMA (global NOS inhibitor) after PAF challenge failed to decrease the peak IOI: N (PAF) = 7; N (PAF + l-NMMA) = 5. D: application of 10 nmol/L LY-294002 (inhibitor of PI3K) after PAF challenge did not inhibit PAF-induced hyperpermeability: N (PAF) = 7; N (PAF + LY-294002) = 4. Data are shown as means ± SE. AU, arbitrary units.

Figure 3.

Stimulation of Epac1 inactivates agonist-induced hyperpermeability. Platelet-activating factor (PAF) treatment significantly increased human microvascular endothelial cell (HMVEC) monolayer permeability to FITC-Dx70 compared with baseline control (BL). Epac1 was stimulated with 8cPT-cAMP (8cPT). Application of 8cPT-cAMP alone did not cause significant changes in permeability. 8cPT- cAMP application after PAF strongly inactivated PAF-induced hyperpermeability. 8cPT-cAMP failed to block PAF-induced hyperpermeability in HMVECs depleted of Epac1 via siRNA. *P < 0.05 compared with BL, 8cPT, and PAF + 8cPT in control and compared with BL in Epac1 siRNA results. N = 9 in control and Epac1 siRNA.

Figure 4.

Platelet-activating factor (PAF) induces NO-dependent delayed increase in cAMP in human microvascular endothelial cells (HMVECs). A: treatment of HMVECs with 100 nmol/L PAF for 15, 20, and 25 min significantly increased cAMP concentration compared with baseline (BL). PAF administration for 3, 5, and 10 min did not change cAMP levels. B: pretreatment of HMVECs with N^G-monomethyl-l-arginine (l-NMMA; global NOS inhibitor) completely blocked PAF-induced increase in cAMP. Forskolin (FSK) was used as positive control. *P < 0.05 compared with baseline; N = 7–12.

Figure 5.

Epac-1 stimulation after platelet-activating factor (PAF) induces endothelial nitric oxide synthase (eNOS) movement from cytosol to plasma membrane. Representative immunofluorescence microscopy images of eNOS (red) and VE-cadherin (green) in confluent human microvascular endothelial cell (HMVEC) monolayers treated with PAF or PAF followed by 8cPT-cAMP (8cPT). In control cells, eNOS and VE-cadherin are present mainly in the cell membrane. PAF triggered the movement of eNOS from the plasma membrane to the cytosol, and loss of junctional integrity as indicated by “gaps” in VE-cadherin staining. Treatment with 8cPT-cAMP after PAF restored VE-cadherin integrity and translocated eNOS from the cytosol back to the cell membrane. 8cPT alone did not change eNOS distribution in endothelial cells. Scale bar = 20 µm.

Figure 6.

Epac1 and endothelial nitric oxide synthase (eNOS) are associated in endothelial cells. A: representative images obtained by proximity ligation assay (PLA) of the spatial association between Epac1 and eNOS (red dots) in human microvascular endothelial cells (HMVECs) at baseline, after 100 nmol/L platelet-activating factor (PAF) for 3 min, and pretreated with 8 cPT-cAMP (8cPT) for 10 min. Green fluorescent signal corresponds to VE-cadherin (membrane) and the blue signal to DAPI (nuclei). B: quantification of the PLA-detected association between Epac1-eNOS observed in A. Note that the association between these two proteins is significantly higher after administration of 8cPT-cAMP. We used three independent endothelial cultures and analyzed 3/4 pictures per treatment. Values are means ± SE; N = 3. *P < 0.05 vs. baseline (BL). C: representative figures obtained by PLA of the spatial association between Epac1 and eNOS (red dots) in HMVECs at baseline and with 100 nmol/L PAF for 3, 10, 20, and 60 min. Green fluorescent signal corresponds to VE-cadherin (membrane) and the blue signal to DAPI (Nuclei). Arrows indicate Epac1-eNOS proximity. D: proximity analysis of the PLA-detected association between Epac1-eNOS observed in C. Note that the association between these two proteins is significantly higher after 20 min of PAF stimulation. We used three independent endothelial cultures and analyzed 3/4 pictures per condition. Values are means ± SE. ∗P < 0.05 vs. BL. Scale bar = 20 μm.

Figure 7.

Platelet-activating factor (PAF) elicits vasodilator-stimulated phosphoprotein (VASP) phosphorylation at Ser157 and Ser239. MyEnd cells monolayers were treated with PAF for 3, 10, 20, and 60 min. A: VASP phosphorylation at Ser157 and Ser239 was assessed using Western Blot analysis. B: quantification of the phosphorylation of Ser157 from multiple Western blots. Phosphorylation of Ser157 was significantly higher than baseline (BL) for 10, 20, and 60 min after PAF administration. *P < 0.05 vs. BL control; N = 3.

Figure 8.

Epac1-mediated inactivation of hyperpermeability requires vasodilator-stimulated phosphoprotein (VASP). A: stimulation of Epac1 in VASP-knockout (KO) mouse myocardial endothelial (MyEnd) cells fails to inactivate platelet-activating factor (PAF)-induced hyperpermeability. MyEnd wild-type (WT) and VASP-KO cells monolayers were treated with PAF or PAF followed by 8cPT-cAMP (PAF + 8cPT). PAF significantly increased the permeability of MyEnd WT cells, which was readily inactivated by Epac1 stimulation with 8cPT. MyEnd VASP-KO cells showed hyperpermeability to PAF but failed to show inactivation of hyperpermeability on treatment with 8cPT. *P < 0.05 vs. baseline (BL, WT, and VASP-KO) and PAF + 8cPT (WT); N = 6. B: VASP depletion inhibits Epac1-induced inactivation of hyperpermeability in human endothelial cells. HMVECs transfected with nontargeting (NT) siRNA or VASP/Mena siRNA were treated with PAF or PAF followed by 8cPT (PAF + 8cPT). PAF caused significant hyperpermeability in both NT and VASP/Mena siRNA cells. However, stimulation of Epac1 with 8cPT after PAF failed to inactivate hyperpermeability in VASP/Mena-depleted cells. *P < 0.05 vs. baseline (BL in NT siRNA and VASP + Mena siRNA) and PAF + 8cPT in NT siRNA; N = 10.

Figure 9.

Vasodilator-stimulated phosphoprotein (VASP) is necessary for endothelial nitric oxide synthase (eNOS) translocation from cytosol to the cell membrane. Representative images of confluent wild-type (WT) or VASP-knockout (KO) myocardial endothelial (MyEnd) cells. A: eNOS (red) is localized at the cell membrane (identified by VE-cadherin, green) in control WT cells. Platelet-activating factor (PAF)-treated cells show less eNOS at the cell membrane and increased eNOS in the cytosol. Treatment with 8cPT-cAMP after PAF induced the reappearance of eNOS at the cell membrane. B: eNOS localizes mainly to the cytosol in control MyEnd VASP-KO cells. PAF did not alter eNOS distribution. eNOS remained in the cytosol following application of 8cPT-cAMP after PAF. We used three independent endothelial cultures and analyzed 3/4 pictures per treatment. Scale bar = 20 µm.

Figure 10.

Proposed hypothesis: inactivation of hyperpermeability is an intrinsic self-limiting property of the microvascular endothelium that maintains vascular homeostasis during inflammatory conditions. Proinflammatory agonists mediate the translocation of endothelial nitric oxide (NO) synthase (eNOS) from the plasma membrane to the cytosol (box with dashed lines). The eNOS-derived NO increase in cytosol [NO] leads to hyperpermeability and delayed increase in cAMP concentration ([cAMP]). The increase in [cAMP] causes the phosphorylation of vasodilator-stimulated phosphoprotein (VASP) on Ser157 and Ser239, which assists in the retro-translocation of eNOS back to the cell membrane (boxes with solid lines). eNOS retro-translocation leads to inactivation of hyperpermeability.